| dc.description.abstract | This thesis explores the development of accurate predictive models for hepatitis C diagnosis by
leveraging various machine learning algorithms and pre-processing techniques. The primary
objectives of this study are to investigate and compare the impact of pre-processing methods on
handling missing data, outliers, noise, and imbalanced datasets, as well as to evaluate the importance
of each feature through feature extraction techniques. A wide range of machine learning algorithms,
including Decision Trees, Random Forest, Logistic Regression, Naive Bayes, SVM, and Multilayer
Perceptron, were optimized through hyperparameter tuning and assessed using metrics such as
accuracy, sensitivity, specificity, precision, F1 score, and AUC.
To address class imbalance, resampling techniques were employed. Additionally, a novel
deep learning model, the Multiscale Feature Extraction Neural Network (MSENN), was proposed and
developed, demonstrating a 25% increase in accuracy, recall, and precision compared to traditional
machine learning methods.
Furthermore, a user-friendly graphical user interface (GUI) was implemented to facilitate the
practical application of the developed models. This study also compares the performance of these
models to existing diagnostic methods and discusses challenges related to limited data quality and
medical expertise.
The aim is to improve the accuracy and efficiency of hepatitis C diagnosis, optimize resource
utilization, and enhance patient outcomes. The findings have the potential for implementation in realworld clinical settings, although further research is necessary to address the limitations in medical
data and expertise. Overall, this thesis leverages advanced predictive modeling algorithms to enhance
diagnostic techniques for hepatitis C. | en_US |
