MACHINE LEARNING-BASED APPROACH FOR DIAGNOSING INTESTINAL PARASITIC INFECTIONS IN NORTHERN NIGERIA

Abstract

Intestinal parasitic infections (IPIs) present a significant health challenge in many developing regions, including Northern Nigeria. Traditional diagnostic methods are often inadequate due to their labor-intensive nature and requirement for specialized expertise. This study explores the application of machine learning (ML) to improve the management of IPIs, by utilizing demographic information from 651 fecal samples collected from school-aged children. Two neural network techniques, Multi-layer Perceptron (MLP) and Radial Basis Function Network (RBFN), were employed. Significant Risk Factors assessment were conducted using Recursive Feature Elimination (RFE) and Lasso regression. The MLP-Lasso model demonstrated higher performance with an accuracy score of 0.83, a recall score of 0.87, and an AUC score of 0.92. These findings suggest that ML can significantly enhance diagnostic accuracy and efficiency, providing a valuable tool for public health interventions in resource-constrained settings.