AICARE: DEVELOPMENT OF A MACHINE LEARNING BASED WEB APPLICATION FOR CHRONIC DISEASE PREDICT USING STREAMLIT

• Main Author: Nurmadani, Vina
• Format: Theses
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/71966
• Institution: Institut Teknologi Bandung

Chronic disease is one of the important problems in the health field. Management of chronic diseases must be done quickly to minimize patient risk. It is important to know and predict from the start so that the patient's condition does not worsen. The chronic diseases discussed in this study are breast cancer, diabetes disease, heart disease, and liver disease. Machine learning is used to help doctors make a fast and precise diagnosis. Machine learning models used in this research are Random Forest Algorithm, Logistic Regression, K-Nearest Neighbor, Decision Tree, and Gaussian Naïve Bayes and the model is evaluated with accuracy, precision, sensitivity, and F1-Score. The results of machine learning on classifying breast cancer with seven parameters show that the Random Forest algorithm has an accuracy of 0.97, precision of 0.92, sensitivity of 1, and F1-Score of 0.96. The Random Forest algorithm model is the best model for classifying diabetes with all parameters and using random oversampling with an accuracy of 0.86, precision of 0.83, sensitivity of 0.93, and F1-Score 0.88. Classification of heart disease with all parameters shows that the Logistic Regression algorithm model has an accuracy of 0.79, precision of 0.68, sensitivity of 0.96, and F1-Score 0.8. In classifying liver disease with nine parameters and using random oversampling, it shows that the Random Forest algorithm model has an accuracy of 0.84, precision of 0.9, sensitivity of 0.78, and F1-Score of 0.84. The results of each disease classification are implemented in the AICare web application developed with streamlit which can be accessed at https://aicaree.herokuapp.com/.