Ai & Asd

International Journal of Innovative Research in Medical Science (IJIRMS)
Volume 09, Issue 09, September 2024,

https://doi.org/10.23958/ijirms/vol09-i09/1969
Review article
Artificial Intelligence in Autism Spectrum Disorder:

Technological Innovations to Enhance Quality of
Life: A Holistic Review of Current and Future
Applications
1
Amália Cinthia Meneses do Rêgo , Ph.D., Irami Araújo-Filho *1,2, Ph.D.
1
Institute of Teaching, Research, and Innovation, Liga Contra o Câncer - Natal - Brazil; Full Professor
of the Postgraduate Program in Biotechnology at Potiguar University, (UnP) - Natal/RN - Brazil.
2
Full Professor, Department of Surgery, Potiguar University. Ph.D. in Health Science/ Natal-RN - Brazil.
Received 22 August 2024; Accepted 18 September 2024; Published 22 September 2024
Abstract
Integrating Artificial Intelligence (AI) into healthcare, specifically for managing Autism Spectrum Disorder (ASD), offers transformative
potential to enhance diagnostic accuracy, personalize treatment, and improve patient outcomes. This review explores the application of various
AI programs in ASD management, discussing their functionalities, ethical considerations, implementation challenges, and the need for
comprehensive regulatory frameworks. Critical AI applications such as AI-driven diagnostic imaging, predictive analytics, assisted therapy
robots, remote monitoring, treatment personalization, decision support systems, and therapeutic chatbots are examined. Each technology is
analyzed for its ability to improve the quality of life for individuals with ASD by offering more personalized, efficient, and effective care and
support. Ethical issues, particularly concerning data bias and privacy, are highlighted as significant challenges that need addressing to maximize
AI's benefits while minimizing risks. Practical hurdles like integration with existing healthcare systems, the need for scalable solutions across
diverse geographic and socio-economic contexts, and the high costs associated with AI development are also discussed. Furthermore, the review
underscores the necessity for robust regulatory policies that ensure patient safety, protect data privacy, and maintain high ethical standards in AI
deployment. The paper concludes that while AI presents substantial opportunities for advancing ASD management, achieving these benefits
requires a concerted effort from technologists, clinicians, ethicists, and policymakers to develop AI tools that are not only innovative but also
ethical, equitable, and universally beneficial.
Keywords: autism spectrum disorder, artificial intelligence, quality of life.
Introduction AI, defined as the capability of a machine to imitate

intelligent human behavior, extends into machine learning (ML) and
Autism Spectrum Disorder (ASD) is a complex neurodevelopmental deep learning (DL) subfields that are particularly pertinent to
condition characterized by a range of symptoms that include processing complex datasets and making predictive decisions [7].
challenges in social interaction, restricted interests, and repetitive These technologies have been progressively applied to
behaviors. The heterogeneity and complexity of ASD symptoms various aspects of ASD, including early screening and detection,
necessitate personalized and adaptable approaches for diagnosis, behavioral intervention, and even therapeutic settings to improve
intervention, and management [1-3]. social communication skills [8-10].
In recent years, Artificial Intelligence (AI) has emerged as a The promise of AI in ASD begins with early detection.
transformative tool in the medical field, offering unprecedented Timely diagnosis of ASD can significantly enhance intervention
opportunities for enhancing diagnostic accuracy and personalizing outcomes. However, diagnosing ASD is challenging due to its broad
treatment strategies. This review explores the various dimensions of spectrum and the subtlety of its signs in young children [11].
AI applications in the context of ASD, highlighting its potential to AI-driven models using machine learning algorithms on
improve the quality of life for individuals affected by this disorder genetic, imaging, and developmental data have demonstrated
[4-6]
. potential in identifying ASD markers earlier than traditional
methods [12].
www.ijirms.in 539
In therapeutic applications, AI has created adaptive learning essential synthesis of knowledge and a roadmap for future
environments for children with ASD. These include robot-assisted innovations in this vibrant area of research [23].
therapies and virtual reality (VR), which provide controlled yet
flexible interaction settings that can improve social skills and reduce Methods
anxiety. Furthermore, AI-driven data analysis helps customize
interventions based on individual behavioral patterns and needs [13]. The research strategy employed for this study was meticulously
AI contributes significantly to the ongoing research and designed to encompass an exhaustive review of literature across
understanding of ASD. By applying machine learning techniques to several distinguished databases known for their extensive collection
large datasets, such as genetic information and neuroimaging, AI of medical and scientific peer-reviewed publications. The databases
helps identify potential biomarkers of ASD. These biomarkers are selected for this comprehensive search included PubMed, Scopus,
crucial for understanding the underlying biological pathways and Scielo, Embase, and Web of Science, each renowned for their vast
can lead to more targeted therapies [14]. repository of scholarly articles. Google Scholar was also a
The integration of AI in managing ASD also extends to supplementary resource for accessing the so-called gray literature,
monitoring and maintaining behavioral therapies. AI systems can be which often contains significant studies and reports unavailable in
trained to observe and interpret the patient's progress and adapt the conventional academic journals. The focal point of this research was
interventions accordingly. This dynamic adjustment helps keep the the intersection of leveraging Artificial Intelligence to enhance the
treatment's effectiveness over time, a crucial aspect given the long- Quality of Life for patients with Autism Spectrum Disorder, guiding
term nature of managing ASD [15-17]. the formulation of search parameters. A carefully curated set of
However, the application of AI in ASD also presents keywords was deployed to optimize the search, comprising terms
challenges, primarily related to ethical considerations, data privacy, such as "autism spectrum disorder," "artificial intelligence," and
and the potential for bias in AI algorithms, which must be rigorously "quality of life." This strategic combination of keywords was
addressed to ensure equitable and safe use. Additionally, instrumental in filtering the literature to include studies directly
comprehensive datasets that represent the diverse populations pertinent to the research objectives. To ensure a broad yet relevant
affected by ASD are needed to train unbiased AI models [18-20]. data collection, the inclusion criteria were designed to be
This review aims to synthesize current research findings on comprehensive, welcoming a variety of study designs, including
the application of AI in the diagnosis, treatment, and management of systematic reviews, case-control studies, cross-sectional analyses,
ASD. It will cover the methodologies employed, discuss the case series, and scholarly reviews. Such diversity in study types
successes and limitations observed, and suggest directions for future aimed to capture a spectrum of evidence and viewpoints regarding
research. By integrating findings from various studies, this review the nexus between Artificial Intelligence and enhancing the quality
will highlight how AI not only promises to refine the precision of of life for patients with Autism Spectrum Disorder. The literature
ASD interventions but also enhances the quality of life of individuals review's evaluation and selection process were executed strictly with
with ASD by offering more personalized, efficient, and accessible methodological rigor. This involved a dual-review system, where
care solutions [21]. pairs of reviewers independently evaluated each study's title and
In conclusion, while AI presents a promising frontier in ASD abstract for relevance and conformity to the predefined criteria.
care, collaborative efforts between clinicians, researchers, Discrepancies between reviewers were resolved through
technologists, and ethicists are essential to harness its full potential consultation with a third independent reviewer to reach a consensus,
responsibly. The goal is to develop AI-driven tools and applications ensuring the selection was based on solid and unbiased judgment.
that are scientifically robust, ethically sound, and widely accessible This detailed and systematic approach to research methodology
to improve the lives of those living with ASD [22]. underpins the reliability and validity of the findings presented and
Through detailed examination and continued ensures that the conclusions drawn from this study are grounded in
interdisciplinary research, AI can significantly alter the therapeutic a comprehensive and critically evaluated body of scientific evidence
landscape for ASD, ushering in an era of enhanced precision in care related to Artificial Intelligence and Autism Spectrum Disorder.
and management. This comprehensive review will thus provide an
Results and Discussion
Table 1. AI Programs for ASD Management: Functionalities and Benefits
AI Program Functionality Influence on Quality of Life for Patients with ASD
DeepScan AI Advanced image processing for Facilitates quicker and more accurate ASD diagnoses, potentially
medical diagnostics. identifying subtle neurological markers that might be missed during
standard evaluations.
Predictive Health Utilizes big data to forecast Enables personalized early interventions, improving long-term care
Analytics developmental outcomes and risks. strategies and preventing comorbidities, enhancing patient adaptability and
health outcomes.
RoboTherapist 360 Interactive robots providing Enhances development of social and communication skills through
consistent therapeutic activities. consistent, controlled therapy sessions, making therapeutic interactions
more accessible.
VitaMon AI Continuous real-time monitoring of Provides constant vigilance without direct human supervision, alerting
vital signs and behavioral patterns. caregivers about potential health crises or behavioral issues needing
immediate attention.
TailoredRx AI Customizes treatment plans based Improves the efficacy of treatments by tailoring strategies specifically to the
on individual genetic and medical individual's needs, thereby reducing the trial-and-error process in
data. medication adjustments.
www.ijirms.in 540
CliniHelp Decision AI AI-driven decision support for Assists clinicians in making informed, evidence-based decisions quickly,
selecting optimal treatment plans. reducing treatment errors and enhancing patient trust in medical
interventions.
TalkEase Bot AI chatbots designed for therapeutic Offers continuous emotional support and anxiety management, teaches
interaction and support. coping mechanisms, and can interactively enhance life skills training.
AI in the Diagnosis and Treatment of Ensuring that AI systems learning plainable is paramount so
that both clinicians and patients can understand and trust AI-
Psychiatric and Emotional Disorders generated outputs. This transparency helps in maintaining informed
Artificial Intelligence (AI) offers transformative potential in the consent, where patients are continually educated on how their data
diagnosis and treatment of psychiatric and emotional disorders, is used and are given the choice to accept or decline AI-driven
including Autism Spectrum Disorder (ASD). By leveraging AI, interventions [45].
healthcare providers can achieve more accurate diagnoses, develop Furthermore, preserving the human elements of care is
personalized treatment plans, and enhance patient monitoring and crucial, particularly in mental health and behavioral therapy, where
interventions [24]. However, ensuring these AI systems are fair, personal interactions are often therapeutic themselves [46]. AI should
unbiased, and respect patient autonomy requires careful planning, support, not supplant, the relational dynamics between patients and
ethical consideration, and meticulous implementation [25]. providers. Offering patients the option to opt-out or choose less AI-
AI can significantly assist in diagnostic processes through its intensive care preserves their autonomy and caters to personal
ability to analyze large datasets rapidly, identifying patterns that treatment preferences [47].
might not be evident to humans. Machine learning algorithms, for In this sense, while AI presents significant opportunities, for
instance, can evaluate speech patterns, facial expressions, and social advancing psychiatric care, its integration into clinical practice must
interactions, which can help diagnose conditions like ASD, be handled with care. This includes ensuring algorithmic fairness,
schizophrenia, or depression swiftly and with greater accuracy [26- transparency in AI operations, and maintaining a patient-centered
28]
. approach that respects and preserves human connections and patient
Additionally, predictive analytics utilized in AI models can autonomy [48]. By adhering to these principles, AI can be effectively
use historical and real-time data to predictt psychiatric episodes or harnessed to improve the outcomes and quality of life for individuals
deteriorations in mental health, thereby enabling proactive suffering from psychiatric and emotional disorders [49].
management of the condition [29].
In terms of treatment, AI algorithms can analyze a wealth of
Integration with Wearable Technologies
personal health data to recommend customized therapeutic The integration of AI with wearable technologies represents a
strategies. These might include specific medications, behavioral frontier in the personalized management of Autism Spectrum
therapies, or lifestyle changes tailored to the individual’s genetic Disorder (ASD). Recent advancements have seen the development
profile, symptomatology, and previous treatment responses [30-32]. of wearable devices that can monitor physiological signals and
AI-driven Virtual Reality (VR) and Augmented Reality behaviors in real-time, providing a continuous stream of data
(AR) programs can also play a crucial role, especially for patients invaluable for dynamic treatment regimens [50].
with ASD. For example, VR can simulate social scenarios to teach These devices, ranging from smartwatches to clothing
and improve social skills in a comment, while AR can provide real- embedded with sensors, facilitate the hidden collection of health
time, contextual information to help patients navigate daily tasks and metrics such as heart rate variability, sleep patterns, and activity
reduce anxiety [33-35]. levels, which are crucial for managing ASD effectively [51].
Furthermore, AI-powered wearable technologies enable Studies highlight the potential of these technologies when
continuous monitoring of physiological indicators such as heart rate combined with AI to offer real-time insights and proactive
variability and galvanic skin response. This real-time monitoring can interventions, enhancing the quality of life and independence of
help in immediately identifying and managing acute stress or anxiety individuals with ASD [52].
conditions, potentially averting crises [36-38]. Wearable technologies are proving to be invaluable in the
Emotion recognition systems that analyze facial expressions management of Autism Spectrum Disorder (ASD), mainly through
or vocal tones can further assess a patient’s emotional state, their ability to monitor a wide array of physiological signals and
providing critical information for individuals who may struggle with behaviors that individuals with ASD might find challenging to
self-reporting their emotions [39]. communicate effectively [53].
Despite the advantages, the deployment of AI in this field These devices can capture critical data points such as heart
must address potential biases which could lead to unfair treatment or rate variability (HRV), a crucial indicator of the autonomic nervous
misdiagnosis. To combat this, AI models should be trained on system's function, and signify stress level changes. Similarly, skin
diverse, representative datasets encompassing various ethnicities, conductance measurements, which increase with physiological
genders, ages, and socioeconomic backgrounds [40]. arousal, provide insights into emotional states and stress responses
Regular audits and updates of these AI models are essential [54]
.
to ensure they perform equitably across all demographic groups and Moreover, monitoring body temperature variations offers
adapt to new health trends and findings. Involving ethicists and clues about stress or anxiety levels, while accelerometers and
community representatives in the development process can help gyroscopes record movement patterns, detecting stereotypical
oversee the ethical deployment of these technologies, ensuring they repetitive behaviors common in individuals with ASD. Sleep, often
conform to high standards of equity and justice [41-43]. disrupted in those with ASD, can also be tracked in terms of quality,
Balancing AI use with the need to preserve patient autonomy duration, and disturbances, providing essential data for effective
in clinical decision-making is also critical. AI should augment the management strategies [55,56].
capabilities of clinicians rather than replace them, providing insights
and recommendations that enhance clinical judgments [44].
www.ijirms.in 541
Advanced wearables with cameras enhance monitoring compromising individual rights or dignity [73,74]. The ethical
capabilities by analyzing facial expressions and eye movements to considerations are multifaceted, ranging from privacy and data
infer emotional states and focus areas. Some devices even assess security concerns to bias, autonomy, social implications, and the
vocal patterns and speech to evaluate emotional and stress levels, need for comprehensive regulatory oversight [20].
which are particularly beneficial for non-verbal individuals or those Privacy and data security are paramount, as AI systems
who struggle with expressive communication [57]. involved in ASD care require collecting, storing, and analyzing large
Integrating Artificial Intelligence (AI) with these wearable volumes of sensitive personal data. This data includes physiological
technologies facilitates the generation of real-time insights and the signals, behavioral patterns, and potentially identifying information,
initiation of proactive interventions that significantly improve daily necessitating robust protective measures to guard against
management and intervention strategies for individuals with ASD unauthorized access and breaches [75]. Data integrity involves using
[58]
. AI algorithms are adept at analyzing physiological data to encrypted storage solutions and secure transmission protocols and
promptly detect signs of anxiety or stress, allowing for immediate maintaining transparency about data use practices. Additionally,
intervention by caregivers or through automated systems that initiate individuals with ASD and their caregivers should be thoroughly
calming strategies [59]. informed about the data collection processes, their usage, and the
AAI's capability extends to predicting potential behavioral parties accessing them to secure informed consent that respects the
outbursts by analyzing movement data, thus allowing preemptive patients' cognitive capabilities and autonomy [14,76].
actions to mitigate challenging behaviors before they escalate [60]. However, deploying AI can inadvertently perpetuate
Similarly, sleep data analyzed by AI helps identify disturbances and existing biases if not carefully managed. Algorithmic biases may
patterns, enabling personalized recommendations and arise when AI models are trained on datasets that do not adequately
environmental adjustments to enhance sleep quality [61]. represent the diverse ASD population, potentially leading to skewed
AI systems personalize intervention plans by learning outcomes that favor specific demographics over others [77].
individual patterns from the collected data. This approach not only Such biases could manifest in misdiagnoses or unequal
helps tailor interventions and therapies to individual needs but also access to therapies, underscoring the need for AI systems to be
provides personalized feedback through interactive applications designed and trained on representative data sets. Moreover,
such as virtual reality (VR) or augmented reality (AR), which offer equitable access to AI-driven tools must be ensured to prevent
engaging formats for therapy adapted to user responses [62]. exacerbating health disparities within the community, particularly
Furthermore, AI supports developmental progress tracking affecting those from minority or lower socio-economic backgrounds
[39,78]
over the long term, adjusting strategies as needed to align with the .
individual’s evolving needs. It also enhances communication for The autonomy of individuals with ASD is another critical
those with ASD by recognizing emotions through analysis of facial concern. AI's role in supporting clinical decision-making should
and vocal expressions and supporting real-time language processing, enhance rather than undermine the independence of both patients
thus aiding in more effective communication [63-65]. and healthcare providers [10].
For individuals prone to wandering, GPS-enabled wearables Over-reliance on AI recommendations might impede
can immediately alert caregivers if the individual moves beyond a personalized care approaches that consider unique patient needs and
safe perimeter. Continuous monitoring of vital signs can prompt circumstances, making it essential to strike a balance that supports
timely medical interventions, such as during seizures, which are but does not replace human judgment [79]. Additionally, there's a risk
common in individuals with ASD [66]. that dependence on AI could reduce meaningful human interactions,
In essence, the confluence of AI and wearable technologies which are crucial for the social development of individuals with
not only enriches the toolkit for managing ASD by providing ASD. Maintaining a balance where technology supplements but
continuous, detailed physiological and behavioral data but also does not supplant human care is vital [80].
empowers individuals with ASD to achieve better communication, The broader social implications of AI in ASD care also
reduced anxiety, and a generally enhanced quality of life through include the potential for stigmatization and marginalization if these
personalized, adaptive interventions [67]. These technologies technologies highlight the differences between individuals with
represent a significant step forward in the context-sensitive, real- ASD inappropriately. There is a delicate balance to maintain where
time management of ASD, making daily care and intervention more AI tools should empower patients without marking them
effective and responsive to the individual's unique needs [68]. unnecessarily or invasively, which could reinforce societal
prejudices [81-83].
Ethical and Social Implications Finally, rigorous regulatory and ethical oversight is essential
to navigate the complexities of using AI in healthcare. AI
The deployment of AI in ASD care raises significant ethical and applications must comply with healthcare regulations that govern
social considerations that must be addressed. The privacy of patient privacy and medical devices [44].
sensitive data, the consent process for vulnerable populations, and Moreover, the development and updating of ethical
the potential for AI systems to perpetuate biases present substantial standards specific to AI in ASD care need to involve a broad
challenges [69]. spectrum of stakeholders including ethicists, affected individuals,
There is an urgent need to develop robust ethical frameworks healthcare providers, and technologists to ensure that these
that govern the use of AI in healthcare, emphasizing data protection, technologies are developed and implemented in ways that prioritize
transparency in AI decision-making processes, and equitable access patient welfare, privacy, consent, and justice [84-86].
to technology. Furthermore, these systems must be designed to avoid In sum, while AI holds significant promise for improving the
stigmatization and support the dignity and rights of individuals with management of ASD, realizing this potential requires careful
ASD, ensuring that AI tools are used responsibly and justly [70-72]. consideration of a range of ethical issues [48]. Addressing these
Integrating Artificial Intelligence (AI) in managing Autism concerns through thoughtful AI design, responsible implementation,
Spectrum Disorder (ASD) presents many ethical challenges that and strict adherence to moral and legal standards is crucial for
must be conscientiously navigated to ensure that these technologies harnessing AI's capabilities effectively and equitably [22]. This
are leveraged responsibly, enhancing patient care without
www.ijirms.in 542
approach will optimize patient outcomes and support the broader Regulation (GDPR) in Europe provide foundational data privacy and
goal of fostering a more inclusive and equitable healthcare landscape patient rights frameworks [95].
for individuals with ASD [87]. However, these regulations require updates and expansions
to address the unique challenges posed by AI, particularly
Multimodal Approaches concerning algorithmic bias and continuous learning systems. The
Food and Drug Administration (FDA) provides guidelines for AI in
Employing multimodal approaches is essential for harnessing AI's medical devices. Still, more specific regulations are needed that
full potential in ASD diagnostics and treatment. By integrating data cater to the nuances of AI applications in psychiatry and emotional
from diverse sources such as genetic profiles, neuroimaging, and health [96-98].
behavioral assessments, AI can provide a more comprehensive To strengthen these standards, AI-specific regulations
understanding of ASD [88]. should be developed that address the particularities of algorithmic
This integration allows for developing sophisticated models decision-making and data handling in healthcare [11]. Enhanced
that can predict outcomes more accurately and tailor interventions patient consent processes must be implemented, ensuring patients
more finely to individual needs. However, the complexity of are thoroughly informed about AI use, including potential risks and
multimodal data integration poses significant challenges in data the nature of data used. Privacy protections must be robust,
harmonization, requiring advanced algorithms and substantial incorporating advanced encryption methods, stringent data
computational resources [89]. minimization practices, and strict access controls [58-60].
In treating Psychiatric and Emotional Abnormalities (PEA), It is recommended that interdisciplinary oversight
including Autism Spectrum Disorder (ASD), the deployment of committees be established to monitor AI implementations. These
Artificial Intelligence (AI) introduces a series of ethical, regulatory, committees, comprising ethicists, technologists, legal experts,
and practical challenges that must be meticulously managed [90]. clinicians, and patient advocates, would ensure that AI applications
These challenges center around ensuring the representativeness of adhere to ethical and clinical guidelines and review them regularly
AI training data, balancing AI's role in clinical decision-making, and for compliance and efficacy. Furthermore, international
adhering to stringent ethical and regulatory standards to safeguard collaboration is essential to harmonize regulatory frameworks,
patient interests [52]. which can facilitate AI's global development and deployment in
To avoid perpetuating existing biases, AI models must be healthcare, ensuring that benefits are universally accessible while
trained on datasets that are diverse and representative of the minimizing risks [99-101].
population. This can be achieved by incorporating data across While AI presents significant opportunities for enhancing
different demographics such as age, ethnicity, gender, the diagnosis and treatment of PEA, it brings many ethical and
socioeconomic status, geographic locations, and various levels of practical challenges that must be carefully addressed [44]. Through
disease severity [37]. the conscientious application of diverse training data, balanced
Ensuring the diversity of training data helps develop fair and integration into clinical practices, and rigorous regulatory oversight,
effective algorithms across all patient groups. Additionally, these AI can be harnessed effectively and ethically, ultimately improving
datasets need to be continually updated with new information to outcomes for patients across the spectrum of psychiatric and
reflect changing population dynamics and disease characteristics, emotional disorders [19].
supported by collaborations across multiple healthcare institutions,
allowing for a richer aggregation of data and sharing best practices
[74,91]
Longitudinal and Large-scale Study Outcomes
.
However, merely collecting comprehensive data is not There is a notable deficiency in longitudinal studies addressing the
sufficient. Synthetic data generated through simulations may also be long-term efficacy of AI-powered interventions in ASD. Most
employed to enhance dataset diversity without compromising current research provides only a snapshot based on short-term
individual privacy. This synthetic data must be rigorously validated studies, which does not adequately capture the progressive nature of
to ensure its accuracy and reliability in reflecting real-world ASD and the long-term impacts of interventions [56-58].
conditions [92]. Future research must focus on long-term, large-scale studies
In clinical settings, AI should function to augment human to validate the preliminary findings and ensure that AI applications
decision-making, not replace it. Healthcare providers must utilize AI provide sustained benefits, adapting to the evolving needs of
to aid in diagnostics and treatment planning, providing insights and individuals with ASD [37].
recommendations to assist clinicians and patients. The transparency As Artificial Intelligence (AI) becomes increasingly
of AI systems is paramount; clinicians should be able to understand integrated into healthcare, particularly in the diagnosis and treatment
and explain how AI-derived recommendations are made [69]. This of psychiatric and emotional disorders, including Autism Spectrum
transparency reinforces trust and facilitates informed consent, where Disorder (ASD), it presents unique challenges that necessitate
patients are fully aware of how their data is used and the basis for AI significant updates and expansions to existing legal frameworks like
recommendations [93]. the Health Insurance Portability and Accountability Act (HIPAA) in
Moreover, patient autonomy must be preserved. AI the U.S. and the General Data Protection Regulation (GDPR) in
implementations should consider patient preferences and values, Europe [81-83]. Additionally, synthetic data and inherent risks
integrating these into personalized treatment plans. AI systems ought associated with AI technologies must be rigorously managed to
to be adaptable to individual patient profiles, enhancing ensure medical treatments' safety, privacy, and efficacy [90].
personalization in treatment approaches [22-24]. It is also vital to The introduction of AI in healthcare raises complex data
ensure that AI does not supplant valuable human interactions, privacy and security challenges, particularly with the increased data
especially in the context of PEA, where patient-provider rapport can sharing required between systems and providers. This integration
significantly influence treatment outcomes [94]. heightens the risk of data breaches, suggesting a need for HIPAA and
Regulatory and ethical oversight is another critical area GDPR to enhance data transmission security protocols [45-47].
current standards, such as the U.S. Health Insurance Portability and Furthermore, AI's ability to potentially re-identify
Accountability Act (HIPAA) and the General Data Protection anonymized data through complex pattern recognition exposes
www.ijirms.in 543
inadequacies in current de-identification standards under these efforts are crucial for tackling the multifaceted challenges of ASD,
regulations, which could allow for unintended disclosures of from early diagnosis to intervention tailoring and outcome
personal information [49-52]. monitoring [70].
Another significant concern is the secondary use of data, As Artificial Intelligence (AI) becomes progressively
where AI applications might utilize healthcare data for training integrated into healthcare, particularly in the diagnosis and treatment
purposes, which was not the original intent upon collection [40]. This of psychiatric and emotional disorders such as Autism Spectrum
practice highlights the need for these regulations to be updated to Disorder (ASD), substantial revisions must be made to existing legal
ensure that any secondary use of data involves explicit patient frameworks[4-6]. The Health Insurance Portability and
consent, thereby preventing any unauthorized use of patient data [18]. Accountability Act (HIPAA) in the U.S. and the General Data
Synthetic data, generated through simulations to create Protection Regulation (GDPR) in Europe must be updated to address
artificial datasets, offers a solution for training AI without AI's nuanced challenges. These challenges include ensuring robust
compromising patient privacy. However, ensuring the validity and data privacy, enhancing patient consent protocols, mitigating AI
reliability of synthetic data involves several crucial steps [20]. biases, and establishing international standards for AI applications
[90-92]
Synthetic data must undergo statistical equivalence testing against .
real-world data to confirm that it preserves essential characteristics The deployment of AI in healthcare demands enhanced data
and distributions [8-10]. protection measures to safeguard sensitive patient information
Performance benchmarking should follow, where AI models against the increased risks of breaches and unauthorized access that
trained on synthetic data are compared with those trained on actual come with AI technologies [33,34]. This necessitates including
data to evaluate performance metrics such as accuracy, precision, improved encryption methods and more stringent anonymization
recall, and F1 scores [33]. techniques in HIPAA and GDPR. Additionally, to foster
Iterative refinement of data generation algorithms is transparency, these regulations should mandate that patients are
necessary to minimize discrepancies with real-world conditions. comprehensively informed about their data usage detailing what data
Additionally, involving domain experts in the validation process will be collected, its intended uses, and who will have access to it [58-
61]
helps ensure that synthetic data are realistic and that AI model .
predictions based on this data are clinically sound [65-67]. Moreover, given AI's potential to re-identify individuals
The deployment of AI in healthcare also introduces several from anonymized datasets, stricter de-identification standards are
risks that must be communicated to patients during the consent required to prevent unintended disclosures. Protecting privacy must
process. Privacy risks are a primary concern, as AI systems can be coupled with precise, understandable, and robust consent
potentially expose sensitive health information. Misdiagnosis or processes, ensuring patients are fully aware of AI-driven decisions
inaccurate predictions by AI, mainly if based on biased or non- in their healthcare [85-88].
representative data, is another significant risk [74-76]. Regulations also need to enforce that AI systems are trained
There is also the danger of clinicians over-relying on AI on diverse datasets that accurately reflect all patient groups to
decisions without adequate scrutiny, which could lead to suboptimal prevent biases in AI-driven healthcare decisions. To ensure fairness
care. Moreover, AI systems can perpetuate or amplify existing biases and equity, regular audits should be institutionalized to monitor AI
in training data, leading to discriminatory practices in patient care. systems for biases, performed by internal stakeholders and
Unforeseen consequences in complex medical scenarios where AI independent external bodies to maintain impartiality and
might not fully comprehend the intricacies involved pose additional thoroughness [23-26].
risks [17-19]. Synthetic data offers a pathway to train AI without
To mitigate these risks, patients must be thoroughly compromising patient privacy. Validating this synthetic data
informed about how AI will be used in their care, the data security involves rigorous statistical testing against real-world data to
measures in place, the potential risks involved, and their rights confirm its accuracy in reflecting accurate patient demographics and
regarding opting out of AI-driven processes [48-50]. Ensuring robust clinical scenarios [34-37].
security measures, continuous monitoring of AI systems, regular This includes statistical and equivalence testing to ensure
updates to training datasets, and comprehensive legal protections are that synthetic data maintains fundamental statistical properties
essential steps in safeguarding patient interests [72]. similar to patient data. Performance validation follows where AI
The integration of AI into psychiatric and emotional disorder models trained on synthetic data are tested against models trained on
treatments necessitates careful consideration of privacy, security, real-world data to ensure they retain efficacy and reliability when
and ethical issues. HIPAA and GDPR require substantial revisions applied in genuine clinical situations [63-65].
to address the challenges posed by AI [81]. Moreover, iterative refinement of synthetic data generation
Validating synthetic data and transparently communicating algorithms based on expert feedback and validation outcomes helps
potential risks to patients are critical for maintaining trust and improve their precision and adaptiveness to evolving medical
ensuring the effectiveness of AI applications in healthcare. These knowledge [22].
measures will protect patient data and enhance the safety and Deploying AI in healthcare settings also introduces risks
efficacy of AI-enhanced medical treatment, paving the way for more such as privacy violations, potential biases, dependency on
informed, personalized, and effective healthcare solutions [66-69]. technology, and errors that could lead to misdiagnoses or
inappropriate treatments [5-7]. Comprehensive risk assessments must
Interdisciplinary Collaborations be performed to identify these risks and develop appropriate
mitigation strategies. Enhancing data security measures and
The complexity of ASD and the sophisticated nature of AI continuously monitoring for biases and errors are critical in
technologies necessitate interdisciplinary collaborations. These protecting patient interests [16,87].
collaborations should span across fields such as neurology, Maintaining a balance between AI-driven processes and
psychiatry, computational sciences, ethics, and machine learning to clinician oversight is vital to preserving the human element in
foster innovations that are not only technologically advanced but healthcare. AI should support but not replace the nuanced judgments
also clinically relevant and ethically sound. Such collaborative
www.ijirms.in 544
of human providers. Patients should have the autonomy to opt out of community leaders to build trust and facilitate engagement within
AI-driven care, allowing them to choose traditional treatment these groups can also help mitigate underrepresentation [16].
modalities [94-96]. Another critical measure is regular auditing of AI algorithms
The advancement of AI in treating psychiatric and emotional for biases. These audits, using statistical and computational
disorders requires careful consideration of legal, ethical, and techniques, can identify and rectify biases that might disadvantage
practical challenges. By updating HIPAA and GDPR, validating any specific group [34,58].
synthetic data rigorously, and effectively mitigating potential risks, Validating these tools across various subgroups to confirm
healthcare providers can leverage AI to improve treatment outcomes equitable performance before deployment is essential. This process
responsibly [9-11]. should include rigorous cross-validation techniques to test the AI
These measures will not only enhance the safety and efficacy systems in varied real-world settings and among diverse patient
of AI applications in healthcare but also ensure they are used to profiles [79].
protect patient privacy and uphold their rights, thereby fostering trust Regulatory and ethical frameworks must be established or
and dependability in AI as a transformative healthcare tool [14,88]. updated to require that AI tools demonstrate fairness and accuracy
across all demographic characteristics as a condition of their
Global Diversity and Inclusion approval. Transparency in the demographic characteristics of the
data used for training and testing these systems should be maintained
Research in AI applications for ASD has predominantly been to inform stakeholders about the applicability and limitations of the
centered on populations that do not adequately represent global AI tools [98-100].
diversity. There is a pressing need to extend these studies to include Furthermore, developing and implementing iterative
underrepresented groups to ensure the generalizability of AI tools feedback loops that continuously collect performance data across
across different racial, ethnic, and socioeconomic groups [51]. different populations can facilitate the dynamic adaptation of AI
Addressing this gap is crucial for developing AI-driven tools. Such mechanisms ensure that these technologies evolve in
interventions that are effective and accessible universally, response to new insights and changing demographics [86-88].
preventing disparities in healthcare outcomes. Collaborations among technologists, clinicians, sociologists,
Scalability and Implementation in Clinical Settings [24]. and community advocates are also vital. These multidisciplinary
Scaling AI innovations from research laboratories to real- teams can drive the development of culturally competent AI
world clinical settings is fraught with challenges, including technical technologies that effectively address and incorporate the needs of
integration with existing healthcare systems, data interoperability, underrepresented groups [66-69].
and user (clinician and patient) acceptance [66]. By implementing these inclusive and comprehensive
The healthcare industry must address these challenges by approaches, AI tools in ASD research can be made more effective
developing standardized protocols for implementing AI tools, and equitable. Ensuring the broad generalizability of these
training clinicians to utilize these technologies effectively, and technologies not only enhances their clinical efficacy but also
designing user-friendly AI interfaces that patients with varying upholds ethical standards, promoting wider trust and acceptance of
levels of technological literacy can quickly adopt [52-54]. AI in healthcare [26,45].
Artificial Intelligence (AI) applications in Autism Spectrum This holistic strategy is essential to harnessing AI's full
Disorder (ASD) research and treatment have shown promising potential in revolutionizing the diagnosis and treatment of ASD
advancements. However, the efficacy and equity of these across all segments of the population, thereby ensuring that no group
technologies are often compromised by the underrepresentation of is left behind as these innovative technologies continue to advance
certain groups in the datasets used to train such AI systems [77-79]. [38,96]
.
Specifically, racial, and ethnic minorities, individuals from
lower socioeconomic backgrounds, non-English speaking Comparative Analyses with Traditional
populations, and certain genders and ages are frequently
Methods
underrepresented. This skew in data can lead to biased AI tools and
potentially less effective for some groups than others [40-43]. While AI offers many advantages over traditional methods, such as
Racial and ethnic minorities such as African Americans and increased efficiency and the ability to handle large datasets,
Hispanics, as well as people from varied socioeconomic statuses, are comparative analyses are essential to evaluate the benefits and
less likely to be included in biomedical research that involves limitations of AI approaches objectively [73].
sophisticated AI technologies [6]. This exclusion is often due to Such analyses will help identify areas where AI can replace
economic barriers, limited access to technology, and lower overall or augment traditional methods and where it falls short. These
engagement with healthcare systems that conduct such research [32]. insights are vital for guiding the development of AI tools that
Additionally, non-English speakers are disadvantaged, as AI complement existing practices and optimize clinical outcomes [81].
applications are commonly developed for English-speaking users Artificial Intelligence (AI) is increasingly being integrated
and tested predominantly within these populations [28]. Furthermore, into healthcare, offering significant improvements over traditional
gender and age disparities also exist, with evidence suggesting that methods, particularly in the diagnosis and treatment of psychiatric
females with ASD are often underdiagnosed, potentially due to the and emotional disorders, including Autism Spectrum Disorder
AI tools being predominantly trained and validated on data from (ASD) [36].
males [45]. AI enhances healthcare efficacy through superior image
To enhance the generalizability of AI tools across these analysis, predictive analytics, personalized medicine, and patient
diverse groups, it is crucial to adopt comprehensive strategies that monitoring, alongside automating administrative tasks such as
ensure equitable training and validation of these technologies. scheduling and data management. However, the application of AI in
Actively recruiting participants from diverse demographics is healthcare is not without challenges [90-93].
essential [68]. This involves not only including individuals of It demands large volumes of high-quality data and can suffer
different races and ethnicities but also balancing the participants' from issues of interpretability, generalizability, and potential biases,
socioeconomic statuses, genders, and ages. Collaborating with which could exacerbate disparities in healthcare delivery's ability to
www.ijirms.in 545
process and analyze large datasets with complex algorithms, By embracing rigorous development practices, ensuring
allowing it to identify patterns and make predictive insights with ethical compliance, and fostering transparency and collaboration, AI
greater accuracy than traditional methods [74,87]. can effectively augment traditional healthcare methods, leading to
For instance, in radiology, dermatology, and pathology, AI enhanced patient outcomes and streamlined healthcare services.
algorithms have successfully analyzed images to detect early signs These efforts are essential for realizing the full potential of AI in
of diseases such as cancer more accurately and swiftly than human healthcare, ensuring it serves as a beneficial tool across all sectors of
practitioners [12-15]. Moreover, AI-driven predictive analytics can the population [47-50].
effectively integrate various patient data to forecast health outcomes, Artificial Intelligence (AI) in healthcare is poised to
making it a powerful tool for managing chronic diseases and significantly enhance diagnostic accuracy, personalize treatment
preventive healthcare [75]. plans, and improve patient outcomes. However, the effective
In surgical applications, robotic AI systems provide integration of AI systems into medical practices must carefully
enhanced precision, which can lead to better outcomes and faster address inherent challenges, such as algorithmic biases and ethical
recovery times. Their role in continuous and remote patient implications, to ensure equity and optimize clinical utility [65-68].
monitoring through wearable technologies and mobile apps Algorithmic bias in AI systems manifests when systematic
exemplifies their capability to extend healthcare beyond traditional and repeatable errors create unfair outcomes, such as privileging one
settings, providing real-time updates on patient vitals and facilitating group of users over others. In healthcare, this can occur in various
timely interventions [78]. forms, including racial and ethnic biases, where AI tools perform
Additionally, AI applications in administrative domains less effectively for minorities [12]. For instance, an algorithm used to
optimize healthcare workflows, significantly reducing the time spent allocate healthcare resources was found to favor white patients over
on routine tasks and improving overall healthcare efficiency [84]. Black patients by assigning riskier scores to the latter when
Despite these advantages, AI systems heavily depend on the conditions were comparable [79].
data they are trained on, which requires vast amounts and high Gender biases are also prevalent, particularly when AI
diversity and quality to function optimally. Inadequate or biased data diagnostic tools for diseases like heart disease underpredict risks for
can lead to inaccurate AI predictions, particularly affecting women because the models were trained primarily on male data [94].
underrepresented groups and potentially leading to healthcare Additionally, socioeconomic, and age-related biases can lead to less
disparities [92]. practical AI applications in low-income settings or among older
Furthermore, the" "black b"x" nature of many AI systems adults, respectively, due to underrepresentation in training data [50].
poses significant challenges in medical settings where transparency To integrate AI systems effectively into existing medical
and trust are paramount. The inability of some AI models to provide practices, it is crucial to involve healthcare professionals directly in
interpretable explanations for their decisions is a critical drawback developing and training AI models. This collaborative approach
in clinical environments where understanding the basis of diagnostic ensures that the AI tools are practical and tailored to the specific
and therapeutic choices is crucial [44-46]. needs of medical practitioners. Utilizing comprehensive and diverse
Moreover, AI systems can develop algorithmic biases if not datasets for training including a wide range of demographics such as
carefully monitored. These biases could manifest as disparities in age, gender, race, and socioeconomic status is fundamental to
treatment efficacy across different patient demographics, mitigating biases and enhancing the generalizability of AI systems
[61-63]
necessitating regular audits and updates to AI algorithms to ensure .
fairness and accuracy [52-54]. The formation of interdisciplinary teams comprising data
Ethical considerations also play a significant role, with scientists, ethicists, clinicians, and patient advocates can address
privacy concerns paramount due to the sensitive nature of health various aspects of AI integration, from technical development to
data. Ensuring robust data protection and adhering to stringent ethical implications and user acceptance [37-39].
ethical standards is essential to maintaining patient trust and Moreover, AI systems require regular updates and
complying with legal standards [94-96]. monitoring to adapt to new medical findings and demographic
Developing AI tools that complement existing medical trends. This continual learning approach is essential to maintain the
practices while optimizing clinical outcomes involves several best accuracy and relevance of AI tools. Implementing explainable AI
practices. Incorporating diverse and comprehensive datasets during (XAI) practices is also vital, allowing practitioners to understand and
training ensures that AI models are well-rounded and applicable validate AI-driven recommendations based on transparent evidence
across various patient groups [30-32]. and clinical knowledge [64-66].
Collaboration among experts from multiple disciplines Several challenges accompany the development of AI
clinicians, data scientists, ethicists, and patient advocates is vital to systems in healthcare. Ensuring the privacy and security of sensitive
align AI developments with practical healthcare needs and ethical medical data is paramount; employing advanced encryption methods
standards [36]. Transparency in AI methodologies, extensive real- and adhering to strict data privacy laws can help protect patient
world testing, and continuous feedback loops enhance AI systems' information [28-30].
reliability, performance, and acceptance in healthcare [53]. AI tools must also be designed to seamlessly integrate into
Regulatory compliance with healthcare standards and existing clinical workflows without disrupting the routines of
privacy laws like HIPAA and GDPR is also crucial to ensure that AI medical professionals. Addressing ethical concerns involves
tools are safe and privacy-preserving. Adhering to these regulations establishing clear guidelines and robust oversight mechanisms to
and continuously engaging in ethical AI practices promotes an AI- oversee AI's decision-making processes involving patient care [77-79].
enabled healthcare environment that is efficient and predictive but Resistance to new technologies is another significant
also fair, understandable, and respectful of patient rights and privacy challenge. Effective change management strategies, including
[77-79]
. training programs and pilot projects, are critical to facilitating
AI presents opportunities to significantly advance healthcare medical staff adoption of AI systems [72]. These programs help
practices, particularly in diagnostics and patient care management; healthcare providers become accustomed to AI tools and understand
it necessitates careful consideration of its limitations and risks [81-83].
www.ijirms.in 546
their benefits, promoting a smoother transition and greater The successful regulation of AI in healthcare requires a
acceptance [74-76]. collaborative approach involving multiple stakeholders.
While AI substantially benefits healthcare by making patient Interdisciplinary teams comprising AI technologists, healthcare
care more personalized, efficient, and accessible, addressing the providers, ethicists, and regulatory experts are essential [2]. These
challenges of algorithmic bias, data integration, and ethical usage are teams should work together to create regulations that address the
crucial [22]. technical, clinical, and ethical dimensions of AI use in healthcare [11].
The proactive mitigation of these issues and strategic Continuous engagement with all stakeholders, including patients, is
implementation plans will be essential for leveraging AI to its full crucial to ensure that regulations remain relevant and effectively
potential in healthcare settings. By ensuring equitable, transparent, address public concerns and technological advancements [35].
and ethically sound applications of AI, the medical community can Given the global nature of AI development and application,
revolutionize healthcare practices and deliver superior clinical international cooperation is another critical element. Sharing
outcomes [86-89]. insights, best practices, and safety data across borders can help
standardize regulations and ensure that AI applications are safe and
Regulatory and Policy Frameworks beneficial globally [59-62].
Regulating AI in healthcare, especially in ASD management,
The regulation of AI in healthcare, particularly in sensitive areas involves a complex interplay of technological innovation, regulatory
such as ASD management, is lagging technological advancements. oversight, and collaborative ethics. By crafting clear, comprehensive
There is a need for comprehensive regulatory frameworks that not regulatory pathways, enhancing data protection laws, and fostering
only ensure patient safety and privacy but also facilitate innovation international cooperation, policymakers can harness AI's benefits to
[92-94]
. improve healthcare outcomes while safeguarding patient safety and
Policymakers must work closely with technologists, privacy [14,24]. Such efforts require the collective input and
clinicians, and ethicists to create regulations that balance these collaboration of technologists, clinicians, and ethicists to ensure that
priorities to support AI's safe and effective use in ASD care [38]. AI tools are innovative, practical, ethical, and equitable [87].
Artificial Intelligence (AI) is increasingly pivotal in
healthcare, particularly in the management of Autism Spectrum Future Directions and Emerging Technologies
Disorder (ASD). This prompts the need for robust regulatory
frameworks to ensure patient safety and privacy while fostering Looking forward, AI in ASD is poised for significant breakthroughs
innovation [16]. The integration of AI into healthcare hinges not only with the advent of technologies such as augmented reality for
on advanced technological applications but also on comprehensive enhanced social training, advanced deep learning models for
regulations and collaborative efforts among various stakeholders, predictive analytics, and blockchain for secure and transparent data
including technologists, clinicians, and ethicists [80]. management. These technologies have the potential to revolutionize
Current regulatory landscapes, such as the Food and Drug ASD diagnosis and treatment, making interventions more effective,
Administration (FDA) in the United States and the European personalized, and accessible [67-70].
Medicines Agency (EMA), alongside the European Commission,
play critical roles in shaping the use of AI in healthcare. The FDA Patient-Centered Designs
categorizes AI-driven software as a medical device, necessitating
premarket approval that confirms safety and efficacy before The design of AI tools must prioritize the user experience of
deployment [95-97]. individuals with ASD. This involves creating intuitive interfaces that
Such regulations are part of the FDA's Digital Health accommodate the sensory and cognitive preferences of individuals
Innovation Action Plan, which is crucial for technologies involved with ASD. A patient-centered approach to developing AI tools
in ASD care. In Europe, similar oversight is provided by the MDR enhances user engagement and satisfaction and improves therapeutic
and the GDPR, the latter safeguarding personal data and setting outcomes by ensuring that interventions align more with the users'
stringent limitations on the use of AI in processing sensitive health needs [99-101].
data without explicit consent [71-74].
Policymakers are tasked with a delicate balance: ensuring AI
Conclusion
applications uphold patient safety without stifling innovation. This In conclusion, the adoption of Artificial Intelligence (AI) in the
balance can be achieved by establishing clear regulatory pathways management of Autism Spectrum Disorder (ASD) presents a
for AI applications classified as medical devices. This would remarkable potential to revolutionize diagnostic and therapeutic
streamline the approval processes and clarify clinical trial and practices. However, the realization of this potential is contingent
evaluation requirements [57-59]. upon meticulously addressing a spectrum of ethical, practical, and
Enhancing data protection laws is also vital, primarily to regulatory challenges. These challenges include mitigating biases in
address unique challenges posed by AI, such as the potential for re- AI algorithms, ensuring the protection of sensitive data, enhancing
identification in large datasets. Laws like GDPR need to emphasize the transparency and explainability of AI systems, and overcoming
more robust consent processes, uphold data minimization principles, significant hurdles in system integration and scalability.
and enhance individuals' rights to understand the use of their data [30- Ethical considerations are paramount, as AI systems must
33]
. fairly represent and effectively serve diverse populations without
Furthermore, promoting transparency and accountability is perpetuating existing disparities. This necessitates the development
essential. Regulations should compel AI developers to disclose the of comprehensive, clear regulatory policies that enforce rigorous
algorithms used, the training data, and any potential biases. testing and validation of AI tools to confirm their safety and efficacy
Additionally, integrating ethical considerations into the regulatory before widespread implementation.
frameworks to guide the development and use of AI applications, Moreover, fostering robust cross-sector partnerships is
particularly those involved in sensitive areas like ASD treatment, is essential for crafting these regulations, requiring cooperation among
crucial. Guidelines should focus on fairness, non-discrimination,
and ensuring human oversight [81-83].
www.ijirms.in 547
technologists, clinicians, ethicists, and policymakers to ensure that [6] Joudar SS, Albahri AS, Hamid RA. Triage and priority-
AI tools are both ethically sound and practically beneficial. based healthcare diagnosis using artificial intelligence for
Furthermore, international collaboration is crucial to autism spectrum disorder and gene contribution: A
standardize AI applications in healthcare globally, ensuring systematic review. Comput Biol Med. 2022
consistent, fair, and effective care enhancements across all regions. Jul;146:105553. doi:
By addressing these challenges head-on, AI can significantly 10.1016/j.compbiomed.2022.105553.
improve the quality and efficiency of ASD care, providing more [7] Feng W, Liu G, Zeng K, Zeng M, Liu Y. A review of
personalized, efficient, and effective support and treatment options. methods for classification and recognition of ASD using
This comprehensive approach will not only facilitate the fMRI data. J Neurosci Methods. 2022 Feb 15;368:109456.
integration of AI into existing healthcare frameworks but also ensure doi: 10.1016/j.jneumeth.2021.109456.
that it complements and enhances medical practices without [8] Helmy E, Elnakib A, ElNakieb Y, Khudri M, Abdelrahim
compromising ethical standards or patient safety. Through deliberate M, Yousaf J, Ghazal M, Contractor S, Barnes GN, El-Baz
and thoughtful policymaking, research, and application design, AI A. Role of Artificial Intelligence for Autism Diagnosis
can fulfill its promise as a transformative tool in the realm of ASD Using DTI and fMRI: A Survey. Biomedicines. 2023 Jun
management. 29;11(7):1858. doi: 10.3390/biomedicines11071858.
[9] Hosseini SA, Molla M. Asperger Syndrome. In: StatPearls
Data Availability [Internet]. Treasure Island (FL): StatPearls Publishing;
2024 Jan.
All data is available upon request to the corresponding author. [10] Xu M, Calhoun V, Jiang R, Yan W, Sui J. Brain imaging-
based machine learning in autism spectrum disorder:
Funding Statement methods and applications. J Neurosci Methods. 2021 Sep
None 1;361:109271. doi: 10.1016/j.jneumeth.2021.109271.
[11] Li J, Kong X, Sun L, Chen X, Ouyang G, Li X, Chen S.
Conflict of interest Identification of autism spectrum disorder based on
electroencephalography: A systematic review. Comput
The authors declare that there is no conflict of interest. Biol Med. 2024 Mar;170:108075. doi:
10.1016/j.compbiomed.2024.108075.
Acknowledgments [12] Alqaysi ME, Albahri AS, Hamid RA. Diagnosis-Based
Hybridization of Multimedical Tests and
The authors thank the Federal University of Rio Grande do Norte, Sociodemographic Characteristics of Autism Spectrum
Potiguar University, and Liga Contra o Cancer for supporting this Disorder Using Artificial Intelligence and Machine
study. Learning Techniques: A Systematic Review. Int J Telemed
Appl. 2022 Jul 1;2022:3551528. doi:
References 10.1155/2022/3551528.
[13] Wawer A, Chojnicka I. Detecting autism from picture
[1] Pandya S, Jain S, Verma J. A comprehensive analysis
book narratives using deep neural utterance embeddings.
towards exploring the promises of AI-related approaches
Int J Lang Commun Disord. 2022 Sep;57(5):948-962. doi:
in autism research. Comput Biol Med. 2024
10.1111/1460-6984.12731.
Jan;168:107801. doi:
[14] Jia Q, Wang X, Zhou R, Ma B, Fei F, Han H. Systematic
10.1016/j.compbiomed.2023.107801.
bibliometric and visualized analysis of research hotspots
[2] Marciano F, Venutolo G, Ingenito CM, Verbeni A,
and trends in artificial intelligence in autism spectrum
Terracciano C, Plunk E, Garaci F, Cavallo A, Fasano A.
disorder. Front Neuroinform. 2023 Dec 6;17:1310400.
Artificial Intelligence: the "Trait D'Union" in Different
doi: 10.3389/fninf.2023.1310400.
Analysis Approaches of Autism Spectrum Disorder
[15] Khodatars M, Shoeibi A, Sadeghi D, Ghaasemi N, Jafari
Studies. Curr Med Chem. 2021;28(32):6591-6618. doi:
M, Moridian P, Khadem A, Alizadehsani R, Zare A, Kong
10.2174/0929867328666210203205221.
Y, Khosravi A, Nahavandi S, Hussain S, Acharya UR,
[3] Alharthi AG, Alzahrani SM. Do it the transformer way: A
Berk M. Deep learning for neuroimaging-based diagnosis
comprehensive review of brain and vision transformers for
and rehabilitation of Autism Spectrum Disorder: A review.
autism spectrum disorder diagnosis and classification.
Comput Biol Med. 2021 Dec;139:104949. doi:
Comput Biol Med. 2023 Dec;167:107667. doi:
10.1016/j.compbiomed.2021.104949.
10.1016/j.compbiomed.2023.107667.
[16] Bahado-Singh RO, Vishweswaraiah S, Aydas B, Mishra
[4] Moridian P, Ghassemi N, Jafari M, Salloum-Asfar S,
NK, Yilmaz A, Guda C, Radhakrishna U. Artificial
Sadeghi D, Khodatars M, Shoeibi A, Khosravi A, Ling
intelligence analysis of newborn leucocyte epigenomic
SH, Subasi A, Alizadehsani R, Gorriz JM, Abdulla SA,
markers for the prediction of autism. Brain Res. 2019 Dec
Acharya UR. Automatic autism spectrum disorder
1;1724:146457. doi: 10.1016/j.brainres.2019.146457.
detection using artificial intelligence methods with MRI
[17] Gao K, Sun Y, Niu S, Wang L. Unified framework for
neuroimaging: A review. Front Mol Neurosci. 2022 Oct
early stage status prediction of autism based on infant
4;15:999605. doi: 10.3389/fnmol.2022.999605.
structural magnetic resonance imaging. Autism Res. 2021
[5] Ali MT, Gebreil A, ElNakieb Y, Elnakib A, Shalaby A,
Dec;14(12):2512-2523. doi: 10.1002/aur.2626.
Mahmoud A, Sleman A, Giridharan GA, Barnes G, Elbaz
[18] Barik K, Watanabe K, Bhattacharya J, Saha G. Functional
AS. A personalized classification of behavioral severity of
connectivity-based machine learning approach for autism
autism spectrum disorder using a comprehensive machine
detection in young children using MEG signals. J Neural
learning framework. Sci Rep. 2023 Oct 9;13(1):17048.
Eng. 2023 Mar 14;20(2). doi: 10.1088/1741-2552/acbe1f.
doi: 10.1038/s41598-023-43478-z.
www.ijirms.in 548
[19] Shahamiri SR, Thabtah F, Abdelhamid N. A new trainings for adults with high-functioning autism spectrum
classification system for autism based on machine disorder (ASD): study protocol for a randomized
learning of artificial intelligence. Technol Health Care. controlled trial. Trials. 2021 Apr 8;22(1):261. doi:
2022;30(3):605-622. doi: 10.3233/THC-213032. 10.1186/s13063-021-05205-9.
[20] Farooq MS, Tehseen R, Sabir M, Atal Z. Detection of [33] Erden YJ, Hummerstone H, Rainey S. Automating autism
autism spectrum disorder (ASD) in children and adults assessment: What AI can bring to the diagnostic process.
using machine learning. Sci Rep. 2023 Jun 13;13(1):9605. J Eval Clin Pract. 2021 Jun;27(3):485-490. doi:
doi: 10.1038/s41598-023-35910-1. 10.1111/jep.13527.
[21] Harmer B, Lee S, Duong TVH, Saadabadi A. Suicidal [34] Yoshinaga K, Egawa J, Watanabe Y, Kasahara H,
Ideation. In: StatPearls [Internet]. Treasure Island (FL): Sugimoto A, Someya T. Usefulness of the autism spectrum
StatPearls Publishing; 2024 Jan. quotient (AQ) in screening for autism spectrum disorder
[22] Akdeniz G. Face-like pareidolia images are more difficult and social communication disorder. BMC Psychiatry.
to detect than real faces in children with autism spectrum 2023 Nov 13;23(1):831. doi: 10.1186/s12888-023-05362-
disorder. Adv Clin Exp Med. 2024 Jan;33(1):13-19. doi: y.
10.17219/acem/162922. [35] Zhang S, Wang S, Liu R, Dong H, Zhang X, Tai X. A
[23] Thabtah F. Machine learning in autistic spectrum disorder bibliometric analysis of research trends of artificial
behavioral research: A review and ways forward. Inform intelligence in the treatment of autistic spectrum disorders.
Health Soc Care. 2019 Sep;44(3):278-297. doi: Front Psychiatry. 2022 Aug 29;13:967074. doi:
10.1080/17538157.2017.1399132. 10.3389/fpsyt.2022.967074.
[24] Supekar K, Ryali S, Yuan R, Kumar D, de Los Angeles C, [36] Bahado-Singh RO, Vishweswaraiah S, Aydas B,
Menon V. Robust, Generalizable, and Interpretable Radhakrishna U. Artificial intelligence and placental DNA
Artificial Intelligence-Derived Brain Fingerprints of methylation: newborn prediction and molecular
Autism and Social Communication Symptom Severity. mechanisms of autism in preterm children. J Matern Fetal
Biol Psychiatry. 2022 Oct 15;92(8):643-653. doi: Neonatal Med. 2022 Dec;35(25):8150-8159. doi:
10.1016/j.biopsych.2022.02.005. 10.1080/14767058.2021.1963704.
[25] Del Valle Rubido M, McCracken JT, Hollander E, Shic F, [37] Varma M, Washington P, Chrisman B, Kline A, Leblanc E,
Noeldeke J, Boak L, Khwaja O, Sadikhov S, Fontoura P, Paskov K, Stockham N, Jung JY, Sun MW, Wall DP.
Umbricht D. In Search of Biomarkers for Autism Identification of Social Engagement Indicators Associated
Spectrum Disorder. Autism Res. 2018 Nov;11(11):1567- With Autism Spectrum Disorder Using a Game-Based
1579. doi: 10.1002/aur.2026. Mobile App: Comparative Study of Gaze Fixation and
[26] Mayor Torres JM, Medina-DeVilliers S, Clarkson T, Visual Scanning Methods. J Med Internet Res. 2022 Feb
Lerner MD, Riccardi G. Evaluation of interpretability for 15;24(2):e31830. doi: 10.2196/31830.
deep learning algorithms in EEG emotion recognition: A [38] Lyu K, Li J, Chen M, Li W, Zhang W, Hu M, Zhang Y,
case study in autism. Artif Intell Med. 2023 Feng X. A bibliometric analysis of autism spectrum
Sep;143:102545. doi: 10.1016/j.artmed.2023.102545. disorder signaling pathways research in the past decade.
[27] D'Mello AM, Frosch IR, Meisler SL, Grotzinger H, Front Psychiatry. 2024 Feb 12;15:1304916. doi:
Perrachione TK, Gabrieli JDE. Diminished Repetition 10.3389/fpsyt.2024.1304916.
Suppression Reveals Selective and Systems-Level Face [39] Valizadeh A, Moassefi M, Nakhostin-Ansari A, Heidari
Processing Differences in ASD. J Neurosci. 2023 Mar Some'eh S, Hosseini-Asl H, Saghab Torbati M, Aghajani
15;43(11):1952-1962. doi: 10.1523/JNEUROSCI.0608- R, Maleki Ghorbani Z, Menbari-Oskouie I, Aghajani F,
22.2023. Mirzamohamadi A, Ghafouri M, Faghani S, Memari AH.
[28] Xiao X, Fang H, Wu J, Xiao C, Xiao T, Qian L, Liang F, Automated diagnosis of autism with artificial intelligence:
Xiao Z, Chu KK, Ke X. Diagnostic model generated by State of the art. Rev Neurosci. 2023 Sep 8;35(2):141-163.
MRI-derived brain features in toddlers with autism doi: 10.1515/revneuro-2023-0050.
spectrum disorder. Autism Res. 2017 Apr;10(4):620-630. [40] Zhao Z, Tang H, Zhang X, Qu X, Hu X, Lu J.
doi: 10.1002/aur.1711. Classification of Children With Autism and Typical
[29] Kaur P, Kaur A. Review of Progress in Diagnostic Studies Development Using Eye-Tracking Data From Face-to-
of Autism Spectrum Disorder Using Neuroimaging. Face Conversations: Machine Learning Model
Interdiscip Sci. 2023 Mar;15(1):111-130. doi: Development and Performance Evaluation. J Med Internet
10.1007/s12539-022-00548-6. Res. 2021 Aug 26;23(8):e29328. doi: 10.2196/29328.
[30] Mertz L. AI, Virtual Reality, and Robots Advancing [41] Menaka R, Karthik R, Saranya S, Niranjan M, Kabilan S.
Autism Diagnosis and Therapy. IEEE Pulse. 2021 Sep- An Improved AlexNet Model and Cepstral Coefficient-
Oct;12(5):6-10. doi: 10.1109/MPULS.2021.3113092. Based Classification of Autism Using EEG. Clin EEG
[31] Al-Shaban FA, Ghazal I, Thompson IR, Klingemier EW, Neurosci. 2024 Jan;55(1):43-51. doi:
Aldosari M, Al-Shammari H, Al-Faraj F, El-Hag S, Tolefat 10.1177/15500594231178274.
M, Ali M, Nasir B, Frazier TW. Development and [42] Elliott SJ, Marshall D, Morley K, Uphoff E, Kumar M,
validation of an Arabic language eye-tracking paradigm Meader N. Behavioural and cognitive behavioural therapy
for the early screening and diagnosis of autism spectrum for obsessive compulsive disorder (OCD) in individuals
disorders in Qatar. Autism Res. 2023 Dec;16(12):2291- with autism spectrum disorder (ASD). Cochrane Database
2301. doi: 10.1002/aur.3046. Syst Rev. 2021 Sep 3;9(9):CD013173. doi:
[32] Tebartz van Elst L, Fangmeier T, Schaller UM, Hennig O, 10.1002/14651858.CD013173.pub2.
Kieser M, Koelkebeck K, Kuepper C, Roessner V,
Wildgruber D, Dziobek I. FASTER and SCOTT&EVA
www.ijirms.in 549
[43] Yu Y, Ozonoff S, Miller M. Assessment of Autism children with autism spectrum disorder. Health Technol
Spectrum Disorder. Assessment. 2024 Jan;31(1):24-41. Assess. 2015 Jun;19(41):1-506. doi: 10.3310/hta19410.
doi: 10.1177/10731911231173089. [54] Riddiford JA, Enticott PG, Lavale A, Gurvich C. Gaze and
[44] Zampella CJ, Wang LAL, Haley M, Hutchinson AG, de social functioning associations in autism spectrum
Marchena A. Motor Skill Differences in Autism Spectrum disorder: A systematic review and meta-analysis. Autism
Disorder: a Clinically Focused Review. Curr Psychiatry Res. 2022 Aug;15(8):1380-1446. doi: 10.1002/aur.2729.
Rep. 2021 Aug 13;23(10):64. doi: 10.1007/s11920-021- [55] Cannon J, O'Brien AM, Bungert L, Sinha P. Prediction in
01280-6. Autism Spectrum Disorder: A Systematic Review of
[45] Liu G, Shi L, Qiu J, Lu W. Two neuroanatomical subtypes Empirical Evidence. Autism Res. 2021 Apr;14(4):604-
of males with autism spectrum disorder revealed using 630. doi: 10.1002/aur.2482.
semi-supervised machine learning. Mol Autism. 2022 Feb [56] Ullah F, AbuAli NA, Ullah A, Ullah R, Siddiqui UA,
23;13(1):9. doi: 10.1186/s13229-022-00489-3. Siddiqui AA. Fusion-Based Body-Worn IoT Sensor
[46] Yap CX, Alvares GA, Henders AK, Lin T, Wallace L, Platform for Gesture Recognition of Autism Spectrum
Farrelly A, McLaren T, Berry J, Vinkhuyzen AAE, Disorder Children. Sensors (Basel). 2023 Feb
Trzaskowski M, Zeng J, Yang Y, Cleary D, Grove R, 3;23(3):1672. doi: 10.3390/s23031672.
Hafekost C, Harun A, Holdsworth H, Jellett R, Khan F, [57] Ahmed ZAT, Aldhyani THH, Jadhav ME, Alzahrani MY,
Lawson L, Leslie J, Levis Frenk M, Masi A, Mathew NE, Alzahrani ME, Althobaiti MM, Alassery F, Alshaflut A,
Muniandy M, Nothard M, Visscher PM, Dawson PA, Alzahrani NM, Al-Madani AM. Facial Features Detection
Dissanayake C, Eapen V, Heussler HS, Whitehouse AJO, System To Identify Children With Autism Spectrum
Wray NR, Gratten J. Analysis of common genetic Disorder: Deep Learning Models. Comput Math Methods
variation and rare CNVs in the Australian Autism Med. 2022 Apr 4;2022:3941049. doi:
Biobank. Mol Autism. 2021 Feb 10;12(1):12. doi: 10.1155/2022/3941049.
10.1186/s13229-020-00407-5. [58] Karaminis T, Stavrakaki S. The psychometric properties
[47] Di Renzo M, di Castelbianco FB, Alberto V, Antonio DV, of the Greek version of the Social Communication
Giovanni C, Vanadia E, Petrillo M, Davide T, Racinaro L, Questionnaire. Autism Res. 2022 Sep;15(9):1768-1780.
Rea M. Prognostic factors and predictors of outcome in doi: 10.1002/aur.2790.
children with autism spectrum disorder: the role of the [59] Andrews DS, Marquand A, Ecker C, McAlonan G. Using
paediatrician. Ital J Pediatr. 2021 Mar 18;47(1):67. doi: Pattern Classification to Identify Brain Imaging Markers
10.1186/s13052-021-01008-5. in Autism Spectrum Disorder. Curr Top Behav Neurosci.
[48] Minissi ME, Altozano A, Marin-Morales J, Centelles N, 2018;40:413-436. doi: 10.1007/7854_2018_47.
Sirera M, Abad L, Alcañiz M. Realidad virtual y [60] Abdel Hameed M, Hassaballah M, Hosney ME, Alqahtani
biomarcadores digitales: una herramienta clínica para el A. An AI-Enabled Internet of Things Based Autism Care
diagnóstico del autismo [Virtual reality and digital System for Improving Cognitive Ability of Children with
biomarkers: a clinical tool for early autism diagnosis]. Autism Spectrum Disorders. Comput Intell Neurosci.
Medicina (B Aires). 2024 Mar;84 Suppl 1:57-64. Spanish. 2022 May 23;2022:2247675. doi: 10.1155/2022/2247675.
[49] Mujeeb Rahman KK, Monica Subashini M. A Deep [61] Nag A, Haber N, Voss C, Tamura S, Daniels J, Ma J,
Neural Network-Based Model for Screening Autism Chiang B, Ramachandran S, Schwartz J, Winograd T,
Spectrum Disorder Using the Quantitative Checklist for Feinstein C, Wall DP. Toward Continuous Social
Autism in Toddlers (QCHAT). J Autism Dev Disord. 2022 Phenotyping: Analyzing Gaze Patterns in an Emotion
Jun;52(6):2732-2746. doi: 10.1007/s10803-021-05141-2. Recognition Task for Children With Autism Through
[50] Peck FC, Gabard-Durnam LJ, Wilkinson CL, Bosl W, Wearable Smart Glasses. J Med Internet Res. 2020 Apr
Tager-Flusberg H, Nelson CA. Prediction of autism 22;22(4):e13810. doi: 10.2196/13810.
spectrum disorder diagnosis using nonlinear measures of [62] Rahman MM, Usman OL, Muniyandi RC, Sahran S,
language-related EEG at 6 and 12 months. J Neurodev Mohamed S, Razak RA. A Review of Machine Learning
Disord. 2021 Nov 30;13(1):57. doi: 10.1186/s11689-021- Methods of Feature Selection and Classification for
09405-x. Autism Spectrum Disorder. Brain Sci. 2020 Dec
[51] Nahas LD, Datta A, Alsamman AM, Adly MH, Al-Dewik 7;10(12):949. doi: 10.3390/brainsci10120949.
N, Sekaran K, Sasikumar K, Verma K, Doss GPC, Zayed [63] Rehman IU, Sobnath D, Nasralla MM, Winnett M, Anwar
H. Genomic insights and advanced machine learning: A, Asif W, Sherazi HHR. Features of Mobile Apps for
characterizing autism spectrum disorder biomarkers and People with Autism in a Post COVID-19 Scenario:
genetic interactions. Metab Brain Dis. 2024 Jan;39(1):29- Current Status and Recommendations for Apps Using AI.
42. doi: 10.1007/s11011-023-01322-3. Diagnostics (Basel). 2021 Oct 17;11(10):1923. doi:
[52] Pagnozzi AM, Conti E, Calderoni S, Fripp J, Rose SE. A 10.3390/diagnostics11101923.
systematic review of structural MRI biomarkers in autism [64] Rochette AC, Soulières I, Berthiaume C, Godbout R.
spectrum disorder: A machine learning perspective. Int J NREM sleep EEG activity and procedural memory: A
Dev Neurosci. 2018 Dec;71:68-82. doi: comparison between young neurotypical and autistic
10.1016/j.ijdevneu.2018.08.010. adults without sleep complaints. Autism Res. 2018
[53] McConachie H, Parr JR, Glod M, Hanratty J, Livingstone Apr;11(4):613-623. doi: 10.1002/aur.1933.
N, Oono IP, Robalino S, Baird G, Beresford B, Charman [65] Rafiee F, Rezvani Habibabadi R, Motaghi M, Yousem
T, Garland D, Green J, Gringras P, Jones G, Law J, Le DM, Yousem IJ. Brain MRI in Autism Spectrum Disorder:
Couteur AS, Macdonald G, McColl EM, Morris C, Narrative Review and Recent Advances. J Magn Reson
Rodgers J, Simonoff E, Terwee CB, Williams K. Imaging. 2022 Jun;55(6):1613-1624. doi:
Systematic review of tools to measure outcomes for young 10.1002/jmri.27949.
www.ijirms.in 550
[66] Lu Z, Wang J, Mao R, Lu M, Shi J. Jointly Composite Disorders. Autism Res. 2021 Mar;14(3):551-559. doi:
Feature Learning and Autism Spectrum Disorder 10.1002/aur.2442.
Classification Using Deep Multi-Output Takagi-Sugeno- [77] Le Menn-Tripi C, Vachaud A, Defas N, Malvy J, Roux S,
Kang Fuzzy Inference Systems. IEEE/ACM Trans Bonnet-Brilhault F. L’évaluation sensori-psychomotrice
Comput Biol Bioinform. 2023 Jan-Feb;20(1):476-488. dans l’autisme : un nouvel outil d’aide au diagnostic
doi: 10.1109/TCBB.2022.3163140. fonctionnel [Sensory-psychomotor evaluation in Autism:
[67] Wang H, Avillach P. Diagnostic Classification and A new tool for functional diagnosis]. Encephale. 2019
Prognostic Prediction Using Common Genetic Variants in Sep;45(4):312-319. French. doi:
Autism Spectrum Disorder: Genotype-Based Deep 10.1016/j.encep.2018.12.003.
Learning. JMIR Med Inform. 2021 Apr 7;9(4):e24754. [78] Deng Z, Wang S. Sex differentiation of brain structures in
doi: 10.2196/24754. autism: Findings from a gray matter asymmetry study.
[68] Pauly R, Ziats CA, Abenavoli L, Schwartz CE, Boccuto L. Autism Res. 2021 Jun;14(6):1115-1126. doi:
New Strategies for Clinical Trials in Autism Spectrum 10.1002/aur.2506.
Disorder. Rev Recent Clin Trials. 2021;16(2):131-137. [79] Ranaut A, Khandnor P, Chand T. Identifying autism using
doi: 10.2174/1574887115666201120093634. EEG: unleashing the power of feature selection and
[69] Baio J, Wiggins L, Christensen DL, Maenner MJ, Daniels machine learning. Biomed Phys Eng Express. 2024 Mar
J, Warren Z, Kurzius-Spencer M, Zahorodny W, Robinson 19;10(3). doi: 10.1088/2057-1976/ad31fb.
Rosenberg C, White T, Durkin MS, Imm P, Nikolaou L, [80] Wolff N, Kohls G, Mack JT, Vahid A, Elster EM, Stroth S,
Yeargin-Allsopp M, Lee LC, Harrington R, Lopez M, Poustka L, Kuepper C, Roepke S, Kamp-Becker I,
Fitzgerald RT, Hewitt A, Pettygrove S, Constantino JN, Roessner V. A data driven machine learning approach to
Vehorn A, Shenouda J, Hall-Lande J, Van Naarden Braun differentiate between autism spectrum disorder and
K, Dowling NF. Prevalence of Autism Spectrum Disorder attention-deficit/hyperactivity disorder based on the best-
Among Children Aged 8 Years - Autism and practice diagnostic instruments for autism. Sci Rep. 2022
Developmental Disabilities Monitoring Network, 11 Sites, Nov 5;12(1):18744. doi: 10.1038/s41598-022-21719-x.
United States, 2014. MMWR Surveill Summ. 2018 Apr [81] Minissi ME, Chicchi Giglioli IA, Mantovani F, Alcañiz
27;67(6):1-23. doi: 10.15585/mmwr.ss6706a1. Raya M. Assessment of the Autism Spectrum Disorder
[70] Ko C, Kim N, Kim E, Song DH, Cheon KA. The effect of Based on Machine Learning and Social Visual Attention:
epilepsy on autistic symptom severity assessed by the A Systematic Review. J Autism Dev Disord. 2022
social responsiveness scale in children with autism May;52(5):2187-2202. doi: 10.1007/s10803-021-05106-
spectrum disorder. Behav Brain Funct. 2016 Jun 5.
27;12(1):20. doi: 10.1186/s12993-016-0105-0. [82] Okoye C, Obialo-Ibeawuchi CM, Obajeun OA, Sarwar S,
[71] Kalantarian H, Jedoui K, Dunlap K, Schwartz J, Tawfik C, Waleed MS, Wasim AU, Mohamoud I,
Washington P, Husic A, Tariq Q, Ning M, Kline A, Wall Afolayan AY, Mbaezue RN. Early Diagnosis of Autism
DP. The Performance of Emotion Classifiers for Children Spectrum Disorder: A Review and Analysis of the Risks
with Parent-Reported Autism: Quantitative Feasibility and Benefits. Cureus. 2023 Aug 9;15(8):e43226. doi:
Study. JMIR Ment Health. 2020 Apr 1;7(4):e13174. doi: 10.7759/cureus.43226.
10.2196/13174. [83] Awaji B, Senan EM, Olayah F, Alshari EA, Alsulami M,
[72] Haebich KM, Pride NA, Walsh KS, Chisholm A, Rouel M, Abosaq HA, Alqahtani J, Janrao P. Hybrid Techniques of
Maier A, Anderson V, Barton B, Silk T, Korgaonkar M, Facial Feature Image Analysis for Early Detection of
Seal M, Lami F, Lorenzo J, Williams K, Dabscheck G, Rae Autism Spectrum Disorder Based on Combined CNN
CD, Kean M, North KN, Payne JM. Understanding autism Features. Diagnostics (Basel). 2023 Sep 14;13(18):2948.
spectrum disorder and social functioning in children with doi: 10.3390/diagnostics13182948.
neurofibromatosis type 1: protocol for a cross-sectional [84] Tachibana Y, Miyazaki C, Ota E, Mori R, Hwang Y,
multimodal study. BMJ Open. 2019 Sep 26;9(9):e030601. Kobayashi E, Terasaka A, Tang J, Kamio Y. A systematic
doi: 10.1136/bmjopen-2019-030601. review and meta-analysis of comprehensive interventions
[73] Wang G, Chen J, Zhang K, Tang S, Wang G. The for pre-school children with autism spectrum disorder
mediating role of gaze patterns in the association of child (ASD). PLoS One. 2017 Dec 6;12(12):e0186502. doi:
sleep disturbances and core symptoms of autism spectrum 10.1371/journal.pone.0186502.
disorder. Autism Res. 2022 Sep;15(9):1719-1731. doi: [85] Garcés P, Baumeister S, Mason L, Chatham CH, Holiga S,
10.1002/aur.2737. Dukart J, Jones EJH, Banaschewski T, Baron-Cohen S,
[74] Al-Hiyali MI, Yahya N, Faye I, Hussein AF. Identification Bölte S, Buitelaar JK, Durston S, Oranje B, Persico AM,
of Autism Subtypes Based on Wavelet Coherence of Beckmann CF, Bougeron T, Dell'Acqua F, Ecker C,
BOLD FMRI Signals Using Convolutional Neural Moessnang C, Charman T, Tillmann J, Murphy DGM,
Network. Sensors (Basel). 2021 Aug 4;21(16):5256. doi: Johnson M, Loth E, Brandeis D, Hipp JF; EU-AIMS
10.3390/s21165256. LEAP group authorship. Resting state EEG power
[75] Defresne P, Mottron L. Clinical Situations in Which the spectrum and functional connectivity in autism: a cross-
Diagnosis of Autism is Debatable: An Analysis and sectional analysis. Mol Autism. 2022 May 18;13(1):22.
Recommendations. Can J Psychiatry. 2022 doi: 10.1186/s13229-022-00500-x.
May;67(5):331-335. doi: 10.1177/07067437211041469. [86] Sigar P, Uddin LQ, Roy D. Altered global modular
[76] Hirota T, Bishop S, Adachi M, Shui A, Takahashi M, Mori organization of intrinsic functional connectivity in autism
H, Nakamura K. Utilization of the Maternal and Child arises from atypical node-level processing. Autism Res.
Health Handbook in Early Identification of Autism 2023 Jan;16(1):66-83. doi: 10.1002/aur.2840. Erratum in:
Spectrum Disorder and Other Neurodevelopmental Autism Res. 2023 Apr;16(4):868-869.
www.ijirms.in 551
[87] Das S, Zomorrodi R, Mirjalili M, Kirkovski M, of autism spectrum disorder. BMC Psychiatry. 2023 Mar
Blumberger DM, Rajji TK, Desarkar P. Machine learning 20;23(1):183. doi: 10.1186/s12888-023-04666-3.
approaches for electroencephalography and [96] Kitzerow J, Hackbusch M, Jensen K, Kieser M,
magnetoencephalography analyses in autism spectrum Noterdaeme M, Fröhlich U, Taurines R, Geißler J, Wolff
disorder: A systematic review. Prog N, Roessner V, Bast N, Teufel K, Kim Z, Freitag CM.
Neuropsychopharmacol Biol Psychiatry. 2023 Apr Study protocol of the multi-centre, randomised controlled
20;123:110705. doi: 10.1016/j.pnpbp.2022.110705. trial of the Frankfurt Early Intervention Programme A-
[88] Sohl K, Kilian R, Brewer Curran A, Mahurin M, FFIP versus early intervention as usual for toddlers and
Nanclares-Nogués V, Liu-Mayo S, Salomon C, Shannon J, preschool children with Autism Spectrum Disorder (A-
Taraman S. Feasibility and Impact of Integrating an FFIP study). Trials. 2020 Feb 24;21(1):217. doi:
Artificial Intelligence-Based Diagnosis Aid for Autism 10.1186/s13063-019-3881-7.
into the Extension for Community Health Outcomes [97] Ko C, Lim JH, Hong J, Hong SB, Park YR. Development
Autism Primary Care Model: Protocol for a Prospective and Validation of a Joint Attention-Based Deep Learning
Observational Study. JMIR Res Protoc. 2022 Jul System for Detection and Symptom Severity Assessment
19;11(7):e37576. doi: 10.2196/37576. of Autism Spectrum Disorder. JAMA Netw Open. 2023
[89] Aghdam MA, Sharifi A, Pedram MM. Diagnosis of May 1;6(5):e2315174. doi:
Autism Spectrum Disorders in Young Children Based on 10.1001/jamanetworkopen.2023.15174.
Resting-State Functional Magnetic Resonance Imaging [98] Dow D, Day TN, Kutta TJ, Nottke C, Wetherby AM.
Data Using Convolutional Neural Networks. J Digit Screening for autism spectrum disorder in a naturalistic
Imaging. 2019 Dec;32(6):899-918. doi: 10.1007/s10278- home setting using the systematic observation of red flags
019-00196-1. (SORF) at 18-24 months. Autism Res. 2020
[90] Dell'Osso L, Gesi C, Massimetti E, Cremone IM, Barbuti Jan;13(1):122-133. doi: 10.1002/aur.2226.
M, Maccariello G, Moroni I, Barlati S, Castellini G, [99] Rahman R, Kodesh A, Levine SZ, Sandin S, Reichenberg
Luciano M, Bossini L, Rocchetti M, Signorelli M, Aguglia A, Schlessinger A. Identification of newborns at risk for
E, Fagiolini A, Politi P, Ricca V, Vita A, Carmassi C, Maj autism using electronic medical records and machine
M. Adult Autism Subthreshold Spectrum (AdAS learning. Eur Psychiatry. 2020 Feb 26;63(1):e22. doi:
Spectrum): Validation of a questionnaire investigating 10.1192/j.eurpsy.2020.17.
subthreshold autism spectrum. Compr Psychiatry. 2017 [100] Fernandes JM, Soares S, Lopes R, Jerónimo R, Barahona-
Feb;73:61-83. doi: 10.1016/j.comppsych.2016.11.001. Corrêa JB. Attribution of intentions in autism spectrum
[91] Zhang X, Noah JA, Singh R, McPartland JC, Hirsch J. disorder: A study of event-related potentials. Autism Res.
Support vector machine prediction of individual Autism 2022 May;15(5):847-860. doi: 10.1002/aur.2702.
Diagnostic Observation Schedule (ADOS) scores based [101] Senarathne UD, Indika NR, Jezela-Stanek A, Ciara E,
on neural responses during live eye-to-eye contact. Sci Frye RE, Chen C, Stepien KM. Biochemical, Genetic and
Rep. 2024 Feb 8;14(1):3232. doi: 10.1038/s41598-024- Clinical Diagnostic Approaches to Autism-Associated
53942-z. Inherited Metabolic Disorders. Genes (Basel). 2023 Mar
[92] Uljarević M, Carrington SJ, Hardan AY, Leekam SR. 27;14(4):803. doi: 10.3390/genes14040803.
Subdomains of restricted and repetitive behaviors within
autism: Exploratory structural equation modeling using
the diagnostic interview for social and communication
disorders. Autism Res. 2022 May;15(5):861-869. doi:
10.1002/aur.2687. Open Access This article is licensed under a
[93] McClain MB, Harris B, Schwartz SE, Benallie KJ, Golson Creative Commons Attribution 4.0 International
ME, Benney CM. Brief Report: Development and License, which permits use, sharing, adaptation, distribution and
Validation of the Autism Spectrum Knowledge Scale reproduction in any medium or format, as long as you give
General Population Version: Preliminary Analyses. J appropriate credit to the original author(s) and the source, provide a
Autism Dev Disord. 2019 Jul;49(7):3007-3015. doi: link to the Creative Commons license, and indicate if changes were
10.1007/s10803-019-04019-8. made. The images or other third-party material in this article are
[94] Wu T, Duan Y, Zhang T, Tian W, Liu H, Deng Y. Research included in the article’s Creative Commons license, unless indicated
Trends in the Application of Artificial Intelligence in otherwise in a credit line to the material. If material is not included
Oncology: A Bibliometric and Network Visualization in the article’s Creative Commons license and your intended use is
Study. Front Biosci (Landmark Ed). 2022 Aug not permitted by statutory regulation or exceeds the permitted use,
31;27(9):254. doi: 10.31083/j.fbl2709254. you will need to obtain permission directly from the copyright
[95] Xiao L, Huo X, Wang Y, Li W, Li M, Wang C, Wang F, holder. To view a copy of this license, visit
Sun T. A bibliometric analysis of global research status https://creativecommons.org/licenses/by/4.0/.
and trends in neuromodulation techniques in the treatment
© The Author(s) 2024
www.ijirms.in 552

Ai & Asd

Uploaded by

Copyright:

Available Formats

Ai & Asd

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Ai & Asd

Uploaded by

Copyright:

Available Formats

International Journal of Innovative Research in Medical Science (IJIRMS)

Volume 09, Issue 09, September 2024,

Artificial Intelligence in Autism Spectrum Disorder:

Received 22 August 2024; Accepted 18 September 2024; Published 22 September 2024

Keywords: autism spectrum disorder, artificial intelligence, quality of life.

Introduction AI, defined as the capability of a machine to imitate

You might also like