RISK FACTOR PREDICTION OF HEPATITIS PATIENT USING MACHINE LEARNING APPROACH

SEBLE, AZENE ALEMU

dc.contributor.author	SEBLE, AZENE ALEMU
dc.date.accessioned	2024-12-05T07:52:23Z
dc.date.available	2024-12-05T07:52:23Z
dc.date.issued	2024-08
dc.identifier.uri	http://ir.bdu.edu.et/handle/123456789/16286
dc.description.abstract	Hepatitis is a severe and potentially fatal disease characterized by liver cell inflammation and reduced liver function. It progress to cirrhosis, fibrosis, or even liver cancer. Hepatitis viruses are the leading cause of liver-related illnesses worldwide. In this study used Particle Swarm Optimization (PSO) and Genetic Algorithm (GA) for hyperparameter optimization, model explainability techniques and the incorporation of data balancing methods. This study aims to develop a machine learning model for predicting the risk factor of hepatitis patients and elucidate the predictive mechanisms of the models. Patient data were collected from Felege Hiwot, Tibebe Gion, and Addis Alem hospitals. Six machine learning techniques, namely Decision Tree (DT), Random Forest (RF), Deep Neural Network (DNN), Extreme Grahigh risknt Boosting (XGBoost), Adaptive Boosting (AdaBoost), and Light Grahigh risknt Boosting Machine (LightGBM), were combined with PSO and GA for hyperparameter optimization to identify the best-performing model. Additionally, model explainability techniques, such as SHAP and LIME, were employed to gain insights into the predictive mechanisms of the models. Among these techniques, LightGBM exhibited superior performance, achieving an accuracy of 97% and an AUC value of 99% when utilizing the PSO hyperparameter optimization with selected features and applying data balancing with SMOTE techniques. Additionally, employing the GA hyperparameter optimization resulted in an accuracy of 94% and an AUC of 99%. Comparing the two hyperparameter optimization techniques, PSO demonstrated the best performance. Furthermore, the model explainability techniques provided valuable insights into the factors driving the predictions, enabling a better understanding of the underlying mechanisms. The data for this model were collected from only three hospitals, but for better model performance, adding more data and developing healthcare and medical recommendation models using the collected data are important suggestions for future investigation. Keywords: Hepatitis, Risk factor prediction, machine learning, Model Explainability	en_US
dc.language.iso	en_US	en_US
dc.subject	Computer Science	en_US
dc.title	RISK FACTOR PREDICTION OF HEPATITIS PATIENT USING MACHINE LEARNING APPROACH	en_US
dc.type	Thesis	en_US