With her strong background in Dental Surgery and Computer Science and interdisciplinary style of addressing problems, Prof. Suebnukarn has carved out a unique and important niche for herself in Thailand and has established herself as a recognized researcher globally in the application of computing in Dentistry.
Situation awareness is known to be a critical skill in surgical decision making. While a few simu... more Situation awareness is known to be a critical skill in surgical decision making. While a few simulators have been developed to teach surgical decision making, none explicitly address teaching situation awareness skills. In this paper we present a knowledge representation framework that captures the key elements in reasoning about situation awareness. The framework makes use of concepts from AI planning and uses PDDL to represent surgical procedures. We describe tutorial feedback strategies identified in a preliminary observational study of endodontic surgery. We then present algorithms that implement these strategies using the knowledge representation framework. We show how the representation supports generating a number of tutorial interventions observed in teaching sessions by expert endodontic surgeons. We finally describe the contributions of our work.
International Journal of Environmental Research and Public Health, 2021
This study aims to analyze the patient characteristics and factors related to clinical outcomes i... more This study aims to analyze the patient characteristics and factors related to clinical outcomes in the crisis management of the COVID-19 pandemic in a field hospital. We conducted retrospective analysis of patient clinical data from March 2020 to August 2021 at the first university-based field hospital in Thailand. Multivariable logistic regression models were used to evaluate the factors associated with the field hospital discharge destination. Of a total of 3685 COVID-19 patients, 53.6% were women, with the median age of 30 years. General workers accounted for 97.5% of patients, while 2.5% were healthcare workers. Most of the patients were exposed to coronavirus from the community (84.6%). At the study end point, no patients had died, 97.7% had been discharged home, and 2.3% had been transferred to designated high-level hospitals due to their condition worsening. In multivariable logistic regression analysis, older patients with one or more underlying diseases who showed symptoms ...
Proceedings of the 22nd International Conference on Intelligent User Interfaces Companion, 2017
We present a virtual reality simulator for teaching emergency management decision-making in endod... more We present a virtual reality simulator for teaching emergency management decision-making in endodontic surgery. Objectives of the simulator are to 1) teach how to correctly respond to a variety of emergency situations, 2) acclimate students to making decisions in stressful emergency situations and 3) teach students the situation awareness skills required to rapidly recognize and respond to emergencies. To meet these objectives, we present a simulator that permits emergency situations to be dynamically inserted at various points in the procedure and that is immersive. The simulator also allows a teacher to observe and review a session in real-time or post session. Preliminary evaluation of face and content validity shows that the simulation is sufficiently realistic and the system is a promising teaching tool.
Background: Artificial intelligence (AI) applications in oncology have been developed rapidly wit... more Background: Artificial intelligence (AI) applications in oncology have been developed rapidly with reported successes in recent years. This work aims to evaluate the performance of deep convolutional neural network (CNN) algorithms for the classification and detection of oral potentially malignant disorders (OPMDs) and oral squamous cell carcinoma (OSCC) in oral photographic images.Methods: A dataset comprising 980 oral photographic images was divided into 365 images of OSCC, 315 images of OPMDs and 300 images of non-pathological images. Multiclass image classification models were created by using DenseNet-169, ResNet-101, SqueezeNet and Swin-S. And multiclass object detection models were fabricated by using faster R-CNN, YOLOv5, RetinaNet and CenterNet2.Results: The AUC of multiclass image classification of CNN models was 0.71-1.00 and 0.80- 0.98 on OSCC and OPMDs, respectively. The AUC of multiclass CNN-base object detection models was 0.81-0.91 and 0.34-0.64 on OSCC and OPMDs, re...
2021 18th International Joint Conference on Computer Science and Software Engineering (JCSSE), 2021
Clinical training is one of the most challenging areas for education especially during the COVID-... more Clinical training is one of the most challenging areas for education especially during the COVID-19 pandemic. There are limited access to apprenticeship training in the complex scenarios with corresponding difficulty training in a time-effective manner. Our work on intelligent clinical training systems provides one effective solution to this problem by introducing intelligent clinical training systems that can supplement tutoring sessions by expert clinical instructors. The Bayesian representation techniques and algorithms for generating tutoring feedback in medical problem-based learning group problem solving made important contributions to the field of Intelligent Tutoring Systems. In particular, it was one of the first systems for tutoring groups of students and the first intelligent tutoring systems for medical problem-based learning. The virtual reality simulator we developed is one of the most sophisticated dental simulators. It stands out as the first dental simulator to integrate sophisticated analysis of the surgical procedure. Particularly noteworthy is also the creative way to understand important issues such as differences in expert and novice performance, the effectiveness of virtual pre-operative practice, and the teaching effectiveness of the simulator. The systems have been implemented in undergrad pre-clinical training and postgrad pre-surgical training with strong scientific evidence of their effectiveness.
British Journal of Oral and Maxillofacial Surgery, 2015
An understanding of the processes of clinical decision-making is essential for the development of... more An understanding of the processes of clinical decision-making is essential for the development of health information technology. In this study we have analysed the acquisition of information during decision-making in oral surgery, and analysed cognitive tasks using a "think-aloud" protocol. We studied the techniques of processing information that were used by novices and experts as they completed 4 oral surgical cases modelled from data obtained from electronic hospital records. We studied 2 phases of an oral surgeon's preoperative practice including the "diagnosis and planning of treatment" and "preparing for a procedure". A framework analysis approach was used to analyse the qualitative data, and a descriptive statistical analysis was made of the quantitative data. The results showed that novice surgeons used hypotheticodeductive reasoning, whereas experts recognised patterns to diagnose and manage patients. Novices provided less detail when they prepared for a procedure. Concepts regarding "signs", "importance", "decisions", and "process" occurred most often during acquisition of information by both novices and experts. Based on these results, we formulated recommendations for the design of clinical information technology that would help to improve the acquisition of clinical information required by oral surgeons at all levels of expertise in their clinical decision-making.
This paper describes COMET, a collaborative intelligent tutoring system for medical problembased ... more This paper describes COMET, a collaborative intelligent tutoring system for medical problembased learning. COMET uses Bayesian networks to model individual student knowledge and activity, as well as that of the group. Generic domainindependent tutoring algorithms use the models to generate tutoring hints. We present an overview of the system and then the results of two evaluation studies. The validity of the modeling approach is evaluated in the areas of head injury, stroke and heart attack. Receiver operating characteristic (ROC) curve analysis indicates that, the models are accurate in predicting individual student actions. Comparison of learning outcomes shows that student
BACKGROUND Oral cancer is a deadly disease among the most common malignant tumors worldwide, and ... more BACKGROUND Oral cancer is a deadly disease among the most common malignant tumors worldwide, and it has become an increasingly important public health problem in developing and low to middle income countries. This study aims to use the convolutional neural network (CNN) deep learning algorithms to develop an automated classification and detection model for oral cancer screening. METHODS The study included 700 clinical oral photographs, collected retrospectively from the oral and maxillofacial center, which were divided into 350 images of oral squamous cell carcinoma and 350 images of normal oral mucosa. The classification and detection models were created by using DenseNet121 and faster R-CNN, respectively. Four hundred ninety images were randomly selected as training data. In addition, 70 and 140 images were assigned as validating and testing data, respectively. RESULTS The classification accuracy of DenseNet121 model achieved a precision of 99%, a recall of 100%, an F1 score of 99%, a sensitivity of 98.75%, a specificity of 100% and an area under the receiver operating characteristic curve of 99%. The detection accuracy of a faster R-CNN model achieved a precision of 76.67%, a recall of 82.14%, an F1 score of 79.31% and an area under the precision-recall curve of 0.79. CONCLUSION The DenseNet121 and faster R-CNN algorithm were proved to offer the acceptable potential for classification and detection of cancerous lesions in oral photographic images.
ABSTRACT Most surgical simulations focus on enhancing the learning of technical skill; whereas, t... more ABSTRACT Most surgical simulations focus on enhancing the learning of technical skill; whereas, teaching of non-technical skills particularly decision making skills has received significantly less attention. While some computer-based system for teaching decision making have been developed, they lack the richness of interaction that occurs between student and expert in the operating room. With the end objective of developing an automated system to teach decision making skills, this paper takes a first step by carrying out an observational study of expert tutorial interventions in teaching intraoperative decision making in dental surgery. Actions and discussions were transcribed. Decisions made by novice, assistant, and interventions by expert were identified. The expert interventions were clustered into types. The situation triggering each intervention type was determined. Preliminary analysis of expert interventions identified seven types of expert intervention strategies, in which five of them were found for teaching decision making. The analysis also identified the triggering situation of each, intervention strategies usage comparison in teaching technical and decision making skill, as well as the interaction patterns of among expert, novice, and assistant. The results provide a foundation for designing the pedagogical strategies for an intelligent tutoring system for decision making in dental surgery.
Studies in health technology and informatics, 2015
Dentists are subject to staying in static or awkward postures for long periods due to their highl... more Dentists are subject to staying in static or awkward postures for long periods due to their highly concentrated work. This study describes a real-time personalized biofeedback system developed for dental posture training with the use of vibrotactile biofeedback. The real-time personalized biofeedback system was an integrated solution that comprised of two components: 1) a wearable device that contained an accelerometer sensor for measuring the tilt angle of the body (input) and provided real-time vibrotactile biofeedback (output); and 2) software for data capturing, processing, and personalized biofeedback generation. The implementation of real-time personalized vibrotactile feedback was computed using Hidden Markov Models (HMMs). For the test case, we calculated the probability and log-likelihood of the test movements under the Work related Musculoskeletal Disorders (WMSD) and non-WMSD HMMs. The vibrotactile biofeedback was provided to the user via a wearable device for a WMSD-pred...
Studies in health technology and informatics, 2015
Recognizing clinical style is essential for generating intelligent guidance in virtual reality si... more Recognizing clinical style is essential for generating intelligent guidance in virtual reality simulators for dental skill acquisition. The aim of this study was to determine the potential of Dynamic Time Warping (DTW) in matching novices' tooth cutting sequences with those of experts. Forty dental students and four expert dentists were enrolled to perform access opening to the root canals with a simulator. Four experts performed in manners that differed widely in the tooth preparation sequence. Forty students were randomly allocated into four groups and were trained following each expert. DTW was performed between each student's sequence and all the expert sequences to determine the best match. Overall, the accuracy of the matching was high (95%). The current results suggest that the DTW is a useful technique to find the best matching expert for a student so that feedback based on that expert's performance can be given to the novice in clinical skill training.
The traditional apprenticeship approach to dental surgical skill training has known limitations i... more The traditional apprenticeship approach to dental surgical skill training has known limitations including subjectivity of evaluation, scarcity of available experts, and lack of standardization. As an attempt to address these limitations, dentistry schools have begun to incorporate virtual reality (VR) simulators into surgical curricula. However, automated outcome scoring is not fully supported in existing dental VR simulators. Without automatic outcome analysis, students must still depend on human experts for evaluation of the outcome. With the limited availability of expert supervision, students often end up in unsupervised training with delayed feedback. In this study, we present an approach to automate the process of outcome scoring in dental simulators. Automated outcome scoring is an initial step toward our larger endeavor of automated objective assessment and real-time feedback generation for surgical skill training.
Situation awareness is known to be a critical skill in surgical decision making. While a few simu... more Situation awareness is known to be a critical skill in surgical decision making. While a few simulators have been developed to teach surgical decision making, none explicitly address teaching situation awareness skills. In this paper we present a knowledge representation framework that captures the key elements in reasoning about situation awareness. The framework makes use of concepts from AI planning and uses PDDL to represent surgical procedures. We describe tutorial feedback strategies identified in a preliminary observational study of endodontic surgery. We then present algorithms that implement these strategies using the knowledge representation framework. We show how the representation supports generating a number of tutorial interventions observed in teaching sessions by expert endodontic surgeons. We finally describe the contributions of our work.
International Journal of Environmental Research and Public Health, 2021
This study aims to analyze the patient characteristics and factors related to clinical outcomes i... more This study aims to analyze the patient characteristics and factors related to clinical outcomes in the crisis management of the COVID-19 pandemic in a field hospital. We conducted retrospective analysis of patient clinical data from March 2020 to August 2021 at the first university-based field hospital in Thailand. Multivariable logistic regression models were used to evaluate the factors associated with the field hospital discharge destination. Of a total of 3685 COVID-19 patients, 53.6% were women, with the median age of 30 years. General workers accounted for 97.5% of patients, while 2.5% were healthcare workers. Most of the patients were exposed to coronavirus from the community (84.6%). At the study end point, no patients had died, 97.7% had been discharged home, and 2.3% had been transferred to designated high-level hospitals due to their condition worsening. In multivariable logistic regression analysis, older patients with one or more underlying diseases who showed symptoms ...
Proceedings of the 22nd International Conference on Intelligent User Interfaces Companion, 2017
We present a virtual reality simulator for teaching emergency management decision-making in endod... more We present a virtual reality simulator for teaching emergency management decision-making in endodontic surgery. Objectives of the simulator are to 1) teach how to correctly respond to a variety of emergency situations, 2) acclimate students to making decisions in stressful emergency situations and 3) teach students the situation awareness skills required to rapidly recognize and respond to emergencies. To meet these objectives, we present a simulator that permits emergency situations to be dynamically inserted at various points in the procedure and that is immersive. The simulator also allows a teacher to observe and review a session in real-time or post session. Preliminary evaluation of face and content validity shows that the simulation is sufficiently realistic and the system is a promising teaching tool.
Background: Artificial intelligence (AI) applications in oncology have been developed rapidly wit... more Background: Artificial intelligence (AI) applications in oncology have been developed rapidly with reported successes in recent years. This work aims to evaluate the performance of deep convolutional neural network (CNN) algorithms for the classification and detection of oral potentially malignant disorders (OPMDs) and oral squamous cell carcinoma (OSCC) in oral photographic images.Methods: A dataset comprising 980 oral photographic images was divided into 365 images of OSCC, 315 images of OPMDs and 300 images of non-pathological images. Multiclass image classification models were created by using DenseNet-169, ResNet-101, SqueezeNet and Swin-S. And multiclass object detection models were fabricated by using faster R-CNN, YOLOv5, RetinaNet and CenterNet2.Results: The AUC of multiclass image classification of CNN models was 0.71-1.00 and 0.80- 0.98 on OSCC and OPMDs, respectively. The AUC of multiclass CNN-base object detection models was 0.81-0.91 and 0.34-0.64 on OSCC and OPMDs, re...
2021 18th International Joint Conference on Computer Science and Software Engineering (JCSSE), 2021
Clinical training is one of the most challenging areas for education especially during the COVID-... more Clinical training is one of the most challenging areas for education especially during the COVID-19 pandemic. There are limited access to apprenticeship training in the complex scenarios with corresponding difficulty training in a time-effective manner. Our work on intelligent clinical training systems provides one effective solution to this problem by introducing intelligent clinical training systems that can supplement tutoring sessions by expert clinical instructors. The Bayesian representation techniques and algorithms for generating tutoring feedback in medical problem-based learning group problem solving made important contributions to the field of Intelligent Tutoring Systems. In particular, it was one of the first systems for tutoring groups of students and the first intelligent tutoring systems for medical problem-based learning. The virtual reality simulator we developed is one of the most sophisticated dental simulators. It stands out as the first dental simulator to integrate sophisticated analysis of the surgical procedure. Particularly noteworthy is also the creative way to understand important issues such as differences in expert and novice performance, the effectiveness of virtual pre-operative practice, and the teaching effectiveness of the simulator. The systems have been implemented in undergrad pre-clinical training and postgrad pre-surgical training with strong scientific evidence of their effectiveness.
British Journal of Oral and Maxillofacial Surgery, 2015
An understanding of the processes of clinical decision-making is essential for the development of... more An understanding of the processes of clinical decision-making is essential for the development of health information technology. In this study we have analysed the acquisition of information during decision-making in oral surgery, and analysed cognitive tasks using a "think-aloud" protocol. We studied the techniques of processing information that were used by novices and experts as they completed 4 oral surgical cases modelled from data obtained from electronic hospital records. We studied 2 phases of an oral surgeon's preoperative practice including the "diagnosis and planning of treatment" and "preparing for a procedure". A framework analysis approach was used to analyse the qualitative data, and a descriptive statistical analysis was made of the quantitative data. The results showed that novice surgeons used hypotheticodeductive reasoning, whereas experts recognised patterns to diagnose and manage patients. Novices provided less detail when they prepared for a procedure. Concepts regarding "signs", "importance", "decisions", and "process" occurred most often during acquisition of information by both novices and experts. Based on these results, we formulated recommendations for the design of clinical information technology that would help to improve the acquisition of clinical information required by oral surgeons at all levels of expertise in their clinical decision-making.
This paper describes COMET, a collaborative intelligent tutoring system for medical problembased ... more This paper describes COMET, a collaborative intelligent tutoring system for medical problembased learning. COMET uses Bayesian networks to model individual student knowledge and activity, as well as that of the group. Generic domainindependent tutoring algorithms use the models to generate tutoring hints. We present an overview of the system and then the results of two evaluation studies. The validity of the modeling approach is evaluated in the areas of head injury, stroke and heart attack. Receiver operating characteristic (ROC) curve analysis indicates that, the models are accurate in predicting individual student actions. Comparison of learning outcomes shows that student
BACKGROUND Oral cancer is a deadly disease among the most common malignant tumors worldwide, and ... more BACKGROUND Oral cancer is a deadly disease among the most common malignant tumors worldwide, and it has become an increasingly important public health problem in developing and low to middle income countries. This study aims to use the convolutional neural network (CNN) deep learning algorithms to develop an automated classification and detection model for oral cancer screening. METHODS The study included 700 clinical oral photographs, collected retrospectively from the oral and maxillofacial center, which were divided into 350 images of oral squamous cell carcinoma and 350 images of normal oral mucosa. The classification and detection models were created by using DenseNet121 and faster R-CNN, respectively. Four hundred ninety images were randomly selected as training data. In addition, 70 and 140 images were assigned as validating and testing data, respectively. RESULTS The classification accuracy of DenseNet121 model achieved a precision of 99%, a recall of 100%, an F1 score of 99%, a sensitivity of 98.75%, a specificity of 100% and an area under the receiver operating characteristic curve of 99%. The detection accuracy of a faster R-CNN model achieved a precision of 76.67%, a recall of 82.14%, an F1 score of 79.31% and an area under the precision-recall curve of 0.79. CONCLUSION The DenseNet121 and faster R-CNN algorithm were proved to offer the acceptable potential for classification and detection of cancerous lesions in oral photographic images.
ABSTRACT Most surgical simulations focus on enhancing the learning of technical skill; whereas, t... more ABSTRACT Most surgical simulations focus on enhancing the learning of technical skill; whereas, teaching of non-technical skills particularly decision making skills has received significantly less attention. While some computer-based system for teaching decision making have been developed, they lack the richness of interaction that occurs between student and expert in the operating room. With the end objective of developing an automated system to teach decision making skills, this paper takes a first step by carrying out an observational study of expert tutorial interventions in teaching intraoperative decision making in dental surgery. Actions and discussions were transcribed. Decisions made by novice, assistant, and interventions by expert were identified. The expert interventions were clustered into types. The situation triggering each intervention type was determined. Preliminary analysis of expert interventions identified seven types of expert intervention strategies, in which five of them were found for teaching decision making. The analysis also identified the triggering situation of each, intervention strategies usage comparison in teaching technical and decision making skill, as well as the interaction patterns of among expert, novice, and assistant. The results provide a foundation for designing the pedagogical strategies for an intelligent tutoring system for decision making in dental surgery.
Studies in health technology and informatics, 2015
Dentists are subject to staying in static or awkward postures for long periods due to their highl... more Dentists are subject to staying in static or awkward postures for long periods due to their highly concentrated work. This study describes a real-time personalized biofeedback system developed for dental posture training with the use of vibrotactile biofeedback. The real-time personalized biofeedback system was an integrated solution that comprised of two components: 1) a wearable device that contained an accelerometer sensor for measuring the tilt angle of the body (input) and provided real-time vibrotactile biofeedback (output); and 2) software for data capturing, processing, and personalized biofeedback generation. The implementation of real-time personalized vibrotactile feedback was computed using Hidden Markov Models (HMMs). For the test case, we calculated the probability and log-likelihood of the test movements under the Work related Musculoskeletal Disorders (WMSD) and non-WMSD HMMs. The vibrotactile biofeedback was provided to the user via a wearable device for a WMSD-pred...
Studies in health technology and informatics, 2015
Recognizing clinical style is essential for generating intelligent guidance in virtual reality si... more Recognizing clinical style is essential for generating intelligent guidance in virtual reality simulators for dental skill acquisition. The aim of this study was to determine the potential of Dynamic Time Warping (DTW) in matching novices' tooth cutting sequences with those of experts. Forty dental students and four expert dentists were enrolled to perform access opening to the root canals with a simulator. Four experts performed in manners that differed widely in the tooth preparation sequence. Forty students were randomly allocated into four groups and were trained following each expert. DTW was performed between each student's sequence and all the expert sequences to determine the best match. Overall, the accuracy of the matching was high (95%). The current results suggest that the DTW is a useful technique to find the best matching expert for a student so that feedback based on that expert's performance can be given to the novice in clinical skill training.
The traditional apprenticeship approach to dental surgical skill training has known limitations i... more The traditional apprenticeship approach to dental surgical skill training has known limitations including subjectivity of evaluation, scarcity of available experts, and lack of standardization. As an attempt to address these limitations, dentistry schools have begun to incorporate virtual reality (VR) simulators into surgical curricula. However, automated outcome scoring is not fully supported in existing dental VR simulators. Without automatic outcome analysis, students must still depend on human experts for evaluation of the outcome. With the limited availability of expert supervision, students often end up in unsupervised training with delayed feedback. In this study, we present an approach to automate the process of outcome scoring in dental simulators. Automated outcome scoring is an initial step toward our larger endeavor of automated objective assessment and real-time feedback generation for surgical skill training.
Uploads