FINANCIAL FRAUD DETECTION IN

HEALTHCARE USING MACHINE AND DEEP

LEARNING

SUBJECT:- RECENT TOPICS

Submitted To :- Submitted By :-
Harish Sharma Sir Krish Khandelwal (22/305)
Kritika Sharma Mam Lakshya Bhati (22/306)
Manish Gupta (22/309)
Mayank Yadav (22/313
 FINANCIAL FRAUD DETECTION IN
HEALTHCARE USING MACHINE AND DEEP
LEARNING
 1 Introduction

2 Research Papers

CONTENT 3 Real World Problem

4 Experiment Methods

5 Result And Conclusion

INTRODUCTION
• The system for detecting the fraud might be composed of a manual
process and the expertise algorithm for detecting the fraud automatically.
The automatic operation can be based upon all previous ways of fraud
transactions happened . Detection of fraud is the process of analyzing the
behavior of card holders’ transactions to know whether the conducted
transaction is genuine.
Different frauds in a credit card can be categorized as fraud in external card or
inner card. Fraud in inner card happens due to commitment of false identity
between the bank and the cardholders, and fraud in external card includes the
usage of a stolen credit card to withdraw the cash by dubious means.

Credit card fraud detection is associated with many challenges, such as

dynamic or the fraudulent behavior of credit cardholders. Such kinds of
activities can be identified using the popular technology called artificial
intelligence through machine learning and deep learning algorithms , such as
KNN , Random Forest , Decision Tree , Logistic Regression , Neural
Networks.
CREDIT CARD FRAUD :-
The most popular algorithm for detecting frauds in credit is inspired by nature.

•Application fraud
•It relates to the criminal who owns a credit card from different issuing
companies by spreading false data related to the cardholder .

•Behavior fraud
• In behavior frauds, the criminal thieves the detail related to the account and
the password related to that account and uses that for with-drawing the money.
Credit card fraud is more accessible as more money can be earned with less
amount of risk in less duration of time.
The sequence pattern of credit card transactions mainly relates
to the Hidden Markov Model (HMM), which identifies the
effectiveness based on credit card fraud.

They evaluate the data with various techniques such as

Random Forest, Logistic Regression, and Support Vector
Machine to predict the different frauds related to the credit
card with the aggregation technique. However, this
aggregation method fails to detect the real-time fraud that
happens in the transaction with the credit card
FEATURE SELECTION :-

Payment Via Credit Card has become more common in both online and offline settings.
As result, the rate of fraud increases, resulting in massive losses for financial
and e-commerce companies. Traditional fraud detection takes a long time; thus,
some artificial intelligence models were required for detecting and tracking
down credit card fraud.

These intelligence techniques include many techniques based on

computational intelligence. The supervised and unsupervised learning methods
are used in the fraud detection system.
Supervised Learning :-
The supervised technique of fraud detection relies on the transaction based
on fraudulent and legitimate and then newly occurred transaction classified
based on the learned model, whereas in

Unsupervised Learning :-
An unsupervised model of fraud detection, the transactions that lie in
outliers are the mainly considered transactions related to the fraud.

For Fraud Detection Algorithms, such as Backpropagation of error

signals with forward and backward passes are used.
EXPERIMENTAL SETUP AND METHODS :-
This section explains how a dataset and various deep learning and machine
learning classifiers, including Logistic Regression, Naive Bayes, Decision Tree,
KNN, and the Sequential Model, were used in the experiment.
Before creating the classifier, all of these algorithms go through various
stages such as data collection, data preprocessing, data analysis, data training
with various classifiers, and later data testing. Preprocessing involves
converting all of the data into a format that can be used.

Preprocessed data are fed into the classifier algorithm during the training
phase. The accuracy of identifying credit card fraud is later determined by
evaluating the test data. Finally, accuracy and best performance are evaluated
for each of the various models.
Dataset :-
The dataset holds the information related transaction conducted
through credit cards as a default payment gateway of the different
customers .
Sequential Model :-
The sequential model generates its sequential value by estimating the
input values for the series which can be time-series data. It is easier to
train the dataset through a sequential model as it requires minimum
computation complexity and generates a better result.

Naive Bayes Classifier:-

Naive Bayes is the statistical method that relies on Bayesian theory,
where the result is obtained based on the highest probability. It estimates
the probability of the unknown value based upon the known value.
Where, m indicates the maximum amount of features,
prob(feature k | class j) indicates the probability of generating
feature value feature k provided in class j.
RESULT:-
All performance metrics’ comparison graph is represented in Figure 3.
RESULT:-
• Evaluating all these classifier models, training is conducted using 70% of the
entire dataset, while for testing and validating, 30% of the dataset is used.
Accuracy, specificity, sensitivity, precision, and the Matthews correlation
coefficient (MCC) with the rate of balance classification are applied for
measuring the performance of all these classifier models. The performance
of all these classifier models is evaluated. The sequential model visualizes
the better performance. e technique of the sequential model generates
superior performance for the evaluation metrics applied. It generates the
highest value for precision and specificity. The obtained performance metrics
are presented in Table 2.
CONCLUSION :-

• The suggested methodology reveals that Sequential CNN performs

worse than Random Forest. The weakness of this methodology is that
Sequential CNN should outperform all other traditional ML
approaches, but this is not the case. It might occur because the dataset
is insufficient for training and identifying hidden patterns to predict
upcoming or future data, and the initialization of weights was
extremely random, which might have an impact on training.
 References:-
[[1] J. O. Awoyemi, A. O. Adetunmbi, and S. A. Oluwadare,
“Credit card fraud detection using machine learning techniques: a comparative analysis,” in Proceedings of the 2017
International Conference on Computing Networking and Informatics
(ICCNI), pp. 1–9, IEEE, Lagos, Nigeria, Oct. 2017.
[2] A. Dal Pozzolo, G. Boracchi, O. Caelen, C. Alippi, and G. Bontempi, “Credit card fraud detection: realistic modeling and a
novel learning strategy,” IEEE transactions on neural networks and learning systems, vol. 29, no. 8, pp. 3784–3797,2017.
[3] S. Xuan, G. Liu, Z. Li, L. Zheng, S. Wang, and C. Jiang,
“Random forest for credit card fraud detection,” in Proceedings of the 2018 IEEE 15th International Conference on Networking,
Sensing and Control (ICNSC), pp. 1–6, IEEE, Zhuhai, China, March 2018.
[4] J. Jurgovsky, M. Granitzer, K. Ziegler et al., “Sequence classification for credit-card fraud detection,” Expert Systems with
Applications, vol. 100, pp. 234–245, 2018.
[5] D. Varmedja, M. Karanovic, S. Sladojevic, M. Arsenovic, and A. Anderla,
“Credit card fraud detection-machine learning methods,” in Proceeding of the 2019 18th International Symposium INFOTEH-
JAHORINA (INFOTEH), pp. 1–5,
IEEE, East Sarajevo, Bosnia and Herzegovina, March 2019.
[6] F. Carcillo, Y.-A. Le Borgne, O. Caelen, Y. Kessaci, F. Obl´e, and G. Bontempi,
“Combining unsupervised and supervised learning in credit card fraud detection,” Information Sciences, vol. 557, pp. 317–331,
2021.
[7] K. Randhawa, C. K. Loo, M. Seera, C. P. Lim, and A. K. Nandi,
“Credit card fraud detection using AdaBoost and majority voting,” IEEE access, vol. 6, Article ID 14277, 2018.
[8] A. G. C. de S´a, A. C. M. Pereira, and G. L. Pappa,
“A customized classification algorithm for credit card fraud detection,” Engineering Applications of Artificial Intelligence,vol.
72, pp. 21–29, 2018.
[9] R. Sailusha, V. Gnaneswar, R. Ramesh, and G. R. Rao,
“Credit card fraud detection using machine learning,” in Proceeding of the 2020 4th International Conference on Intelligent
Computing and Control Systems (ICICCS), pp. 1264–1270, IEEE, Madurai, India, May 2020.
[10] S. Bagga, A. Goyal, N. Gupta, and A. Goyal,
“Credit card frauddetection using pipeling and ensemble learning,” Procedia Computer Science, vol. 173, pp. 104–112, 2020.
THANK YOU
Short Answer Type:
What are the key challenges associated with credit card fraud detection in the healthcare sector, and how do machine learning
and deep learning techniques address these challenges?
 Credit card fraud detection in the healthcare sector faces several challenges, including the dynamic nature of fraudulent behaviors,
the large volume of transactions, and the need to distinguish between genuine and fraudulent transactions accurately. Traditional
methods often struggle with these challenges due to the high variability and complexity of fraud patterns. Machine learning and deep
learning techniques, such as K-Nearest Neighbor (KNN), Random Forest, and Sequential Convolutional Neural Networks (CNNs),
address these challenges by analyzing vast amounts of transaction data to identify patterns and anomalies. These algorithms learn
from historical data, enabling them to detect even subtle deviations that might indicate fraud. For instance, KNN and Random Forest
can classify transactions based on past behaviors, while CNNs can detect complex patterns in transaction sequences, improving the
accuracy and speed of fraud detection in the healthcare sector.

How does the Sequential Convolutional Neural Network (CNN) compare to other machine learning algorithms in detecting credit
card fraud, according to the research findings?
 The research compares various machine learning algorithms, including Naive Bayes, Logistic Regression, K-Nearest Neighbor
(KNN), Random Forest, and Sequential Convolutional Neural Network (CNN), in detecting credit card fraud. According to the
findings, the Sequential CNN achieved an accuracy of 92.3%, which, while substantial, was slightly lower than some other
algorithms like Random Forest, which had an accuracy of 97.58%, and KNN, with 95.89%. The CNN's slightly lower performance
might be due to the complexity and specificity of the patterns it identifies, which could be less effective on the particular dataset
used. However, CNNs are typically strong in detecting complex, sequential patterns, making them potentially more effective in
scenarios where transaction sequences exhibit intricate temporal dependencies, which might not have been fully captured in this
specific study.
What role does the classification of transactions play in fraud detection, and what methodologies are commonly used for this
purpose?
 Classification of transactions is a crucial aspect of fraud detection, as it helps determine whether a transaction is legitimate or
fraudulent. This process involves analyzing transaction data to identify patterns that distinguish normal behavior from
fraudulent activity. Common methodologies used for transaction classification include machine learning algorithms like K-
Nearest Neighbor (KNN), Random Forest, Logistic Regression, Naive Bayes, and Sequential Convolutional Neural Networks
(CNN). These algorithms classify transactions based on various features, such as transaction amount, frequency, location, and
merchant details. By learning from historical data, these models can predict the likelihood of a new transaction being
fraudulent. For example, KNN classifies transactions by comparing them to similar past transactions, while Random Forest
uses decision trees to assess the probability of fraud. Effective classification helps in minimizing false positives and negatives,
thereby enhancing the accuracy and efficiency of fraud detection systems.
How does the integration of anomaly detection enhance the process of identifying fraudulent transactions in credit card
usage?
 Anomaly detection enhances the identification of fraudulent transactions by focusing on transactions that deviate from
established patterns of legitimate behavior. In the context of credit card usage, anomaly detection involves comparing current
transactions against historical data to identify irregularities that may indicate fraud. This method is particularly effective in
detecting new or evolving fraud strategies that may not be captured by traditional rule-based systems. Anomaly detection can
be implemented through machine learning algorithms that learn from historical transaction data to establish what constitutes
'normal' behavior for a given cardholder. When a transaction deviates significantly from this learned behavior, it is flagged as a
potential fraud. This approach is crucial in detecting both known and unknown types of fraud, reducing the risk of financial
loss. Moreover, by incorporating anomaly detection into fraud detection systems, organizations can improve the accuracy of
their fraud prevention measures, reducing false positives and ensuring legitimate transactions are not unnecessarily blocked.
 What are the advantages of using machine learning algorithms like K-Nearest Neighbor (KNN) and Random Forest in
detecting credit card fraud over traditional methods?
 Machine learning algorithms such as K-Nearest Neighbor (KNN) and Random Forest offer several advantages over traditional
methods in detecting credit card fraud. Traditional methods, which often rely on predefined rules and manual processes, can
be inflexible and slow to adapt to new fraud patterns. In contrast, machine learning algorithms can learn from vast amounts of
transaction data, enabling them to identify complex patterns and adapt to evolving fraudulent behaviors. KNN, for instance,
classifies transactions based on the similarity to previous transactions, making it effective in identifying subtle differences
between legitimate and fraudulent activities. Random Forest, on the other hand, uses multiple decision trees to evaluate
transactions from various perspectives, improving the accuracy and robustness of fraud detection. These algorithms can
handle large datasets and perform real-time analysis, making them more efficient and scalable than traditional methods.
Additionally, machine learning models can be continuously updated with new data, ensuring that fraud detection systems
remain effective in the face of changing threats.
Long Answer Questions
1. How does the application of machine learning and deep learning enhance the detection of financial fraud in the
healthcare sector?
Machine learning (ML) and deep learning (DL) significantly enhance financial fraud detection in the healthcare sector by offering
sophisticated tools for analyzing complex and large datasets. The healthcare sector generates a massive amount of data, not only
related to patient health but also involving financial transactions, which are prone to fraud. Traditional methods often fall short in
identifying fraudulent activities due to their reliance on static rules and limited capacity to process large volumes of data. ML and
DL techniques, however, can dynamically learn from historical fraud patterns and adapt to new, emerging threats. Algorithms such
as Naive Bayes, Logistic Regression, K-Nearest Neighbor (KNN), and Sequential Convolutional Neural Networks (CNNs) are
employed to detect anomalies in transactions that may indicate fraud. These algorithms work by analyzing transaction data in real-
time, identifying deviations from normal behavior that could signify fraudulent activities. The application of these technologies in
fraud detection helps in reducing false positives, improving the accuracy of detection, and enabling timely intervention. For
instance, the research paper reports high accuracy rates for several algorithms, with KNN achieving 97.58% accuracy, which
underscores the potential of ML and DL in mitigating financial fraud in healthcare.
2. What challenges are associated with credit card fraud detection in the healthcare sector, and how do machine learning
techniques address these challenges?
Credit card fraud detection in the healthcare sector faces several challenges, including the dynamic nature of fraudulent behavior, the
large volume of transactions, and the need for real-time analysis. Fraudsters constantly evolve their tactics, making it difficult for
static, rule-based systems to keep up. Additionally, the healthcare sector deals with a vast amount of data, making manual fraud
detection processes impractical and inefficient. The need for real-time fraud detection further complicates the situation, as any delay
in identifying fraud can lead to significant financial losses. Machine learning (ML) techniques address these challenges by leveraging
historical data to predict and identify fraudulent activities. Algorithms such as K-Nearest Neighbor (KNN), Random Forest, and
Neural Networks are capable of learning from past transaction patterns, enabling them to detect anomalies that may indicate fraud.
These algorithms can process large datasets quickly and efficiently, providing real-time analysis and reducing the likelihood of
undetected fraud. Moreover, ML techniques can adapt to new fraud patterns, making them more effective than traditional methods in
dealing with the ever-changing landscape of financial fraud. The research paper highlights the effectiveness of these techniques, with
ML models achieving high accuracy rates, demonstrating their potential in overcoming the challenges associated with credit card
fraud detection in the healthcare sector.
3. Discuss the comparative performance of various machine learning algorithms used in the research for fraud detection in
credit cards. Which algorithm showed the highest accuracy, and why?
The research paper provides a comparative analysis of various machine learning algorithms used for fraud detection in credit card
transactions, particularly in the healthcare sector. The algorithms evaluated include Naive Bayes, Logistic Regression, K-Nearest
Neighbor (KNN), Random Forest, and Sequential Convolutional Neural Network (CNN). Among these, the K-Nearest Neighbor
(KNN) algorithm demonstrated the highest accuracy, with a reported accuracy rate of 97.58%. This high accuracy can be attributed to
KNN’s ability to classify transactions based on their proximity to known fraudulent and legitimate transactions in a multi-
dimensional space. KNN is particularly effective in scenarios where the data points are close to each other in clusters, making it
easier to detect outliers, which in this case are fraudulent transactions. The research also highlighted the strengths of other algorithms,
such as Random Forest, which combines multiple decision trees to improve classification accuracy, and Sequential CNN, which is
effective in processing sequential data. However, the superior performance of KNN in this study suggests that its simplicity and
effectiveness in dealing with imbalanced datasets, where fraudulent transactions are much rarer than legitimate ones, make it a
preferred choice for credit card fraud detection in healthcare. This comparison underscores the importance of selecting the
appropriate algorithm based on the specific characteristics of the data and the nature of the fraud detection task
4. How does the integration of deep learning techniques, such as Convolutional Neural Networks (CNNs), contribute to
the accuracy of fraud detection models in the study?
The integration of deep learning techniques, particularly Convolutional Neural Networks (CNNs), significantly contributes to
the accuracy of fraud detection models by enabling the analysis of complex patterns in transaction data that are often difficult to
capture with traditional machine learning algorithms. In the study, Sequential CNNs were employed to process transaction
sequences, allowing the model to learn temporal patterns that indicate fraudulent behavior. This is particularly important in the
context of credit card fraud detection, where the timing and sequence of transactions can provide crucial clues about potential
fraud. CNNs are highly effective at recognizing these patterns due to their ability to capture spatial hierarchies in data through
the use of convolutional layers. These layers apply filters to the input data to detect features at various levels of abstraction,
which are then combined to form a comprehensive understanding of the transaction patterns. The study reported that while K-
Nearest Neighbor (KNN) showed the highest accuracy overall, the CNN model also achieved a strong accuracy rate of 92.3%.
This performance demonstrates that CNNs are a powerful tool in fraud detection, particularly when dealing with sequential data.
Their ability to automatically learn and extract features from raw data, without the need for extensive manual feature
engineering, makes them a valuable addition to the suite of tools used in detecting financial fraud in the healthcare sector.
5. What role do publicly available datasets play in the development and validation of fraud detection models in the
healthcare sector?
Publicly available datasets play a crucial role in the development and validation of fraud detection models, particularly in the
healthcare sector, where access to real-world data may be restricted due to privacy concerns. These datasets provide researchers
with the necessary data to train and test their models, ensuring that the algorithms can effectively detect fraudulent activities. In
the context of the research paper, publicly available datasets were used to evaluate the performance of various machine learning
and deep learning algorithms, such as Naive Bayes, Logistic Regression, K-Nearest Neighbor (KNN), Random Forest, and
Sequential Convolutional Neural Networks (CNNs). These datasets often contain labeled examples of both legitimate and
fraudulent transactions, allowing the models to learn the distinguishing features of fraud. By using these datasets, researchers
can benchmark their models against existing solutions and ensure that their approaches are robust and generalizable. Moreover,
publicly available datasets facilitate the replication of studies, enabling other researchers to validate findings and contribute to
the advancement of fraud detection technologies. The use of these datasets in the study underscores their importance in building
effective and reliable fraud detection systems, which are essential for mitigating financial losses in the healthcare sector due to
fraudulent activities.
Q1: What is HMM?
Ans:HMM, or Hidden Markov Model, is a statistical model used in various fields, including speech recognition, natural language
processing, and bioinformatics. It is a type of probabilistic model that represents a system as a sequence of hidden states, each
associated with observable data.
Q2: What is supervised learning?
Ans:Supervised learning is a machine learning technique where an algorithm learns to make predictions or classifications based on
labeled data. It involves training the algorithm on a dataset that includes input features and corresponding target output labels.
Q3: What is unsupervised learning?
Ans: Unsupervised learning is a machine learning technique where the algorithm is given data without labeled outcomes or guidance. It
aims to discover patterns, structures, or relationships in the data without explicit supervision, allowing the model to learn and make
sense of the data on its own.
Q4: What is dataset?
Ans: A dataset in machine learning is a structured collection ofdata used for training, testing, or validating models. It typically
consists of input features and corresponding target outputs orlabels.

Q5: Write formula of Naive Bayes classifier?

Ans: P(A|B) = P(B|A) * P(A) / P(B)

Q6: Explain random forest technique. Ans: Random Forest is a powerful ensemble learning technique in machine learning. It creates a
"forest" of decision trees, where each tree is trained on a random subset of the data and a random subset of features. This randomness
makes it resistant to overfitting, enhancing predictive accuracy. When making predictions, Random Forest combines the results from all
the individual trees, typically by majority voting for classification or averaging for regression. It excels in handling complex datasets,
provides feature importance insights, and is widely used in applications like image recognition, anomaly detection, and recommendation
systems due to its robustness and versatility.
Q8: Explain k nearest neighbour.
Ans: K-Nearest Neighbours (KNN) is a simple yet effective supervised machine learning algorithm. It operates on the principle
that similar data points share similar attributes. In KNN, the "K“ represents the number of neighbouring data points to consider
when making a prediction. For classification, KNN identifies the K nearest neighbours to a new data point and assigns the
majority
class among these neighbors as the predicted class. In regression, it computes the average of the K nearest neighbors' values for a
numeric prediction. While KNN is intuitive and suitable for small to medium-sized datasets, it can be computationally expensive
for large datasets. Choosing an appropriate K value and distance metric is crucial for optimal performance.
Q9:Explain credit card fraud and it's types.
Ans :Credit card fraud is a criminal act in which someone uses another person's credit card or card information without their
authorization to make unauthorized purchases or transactions.