DEVELOPMENT OF AN ARTIFICIAL INTELLIGENCE-BASED MEDICAL DIAGNOSTIC SYSTEM FOR THE EARLY DETECTION AND PREDICTION OF LYMPHATIC FILARIASIS
Abstract
Lymphatic filariasis (LF) is a serious tropical disease affecting millions of people, especially in endemic regions where diagnosis remains a problem due to slow detection, manual microscopic examination, inadequate diagnostic skills, and human errors. This study proposed an Artificial Intelligence (AI)-based medical diagnostic system for the early detection and prediction of lymphatic filariasis. The study used a combination of the Design Science Research Methodology (DSRM) and the Object-Oriented Analysis and Design Methodology (OOADM) to conduct the research, analyze system requirements, and design, develop, implement, and evaluate the system. The study used structured lymphatic filariasis data and microscopic blood smear images to develop the diagnostic system and evaluate its performance. The diagnostic system was trained using Logistic Regression, Random Forest, XGBoost, and VGG16 transfer learning to classify input data, such as infected, noninfected, early-stage, and late-stage lymphatic filariasis, and make predictions. The data were preprocessed, cleaned, normalized, encoded, reduced, resized, and augmented to meet the standard input requirements of the convolutional neural networks (CNN). Accuracy, precision, recall, F1-score, and a confusion matrix were used to assess the performance of the diagnostic system. A webbased diagnostic interface using the Flask framework and a SQLite database was developed and used to register and process patient information, upload and analyze patient scans, predict, and store patient diagnosis results. The experiment results demonstrated that the AI-based diagnostic system accurately predicted and classified the different stages of Lymphatic filariasis with minimal prediction error. The study concluded that AI-based diagnostic systems could be used to enhance the early prediction and detection of Lymphatic filariasis and other similar diseases in medical diagnostics while reducing diagnostic time and errors associated with manual processing of patient scans