Related Experiment Video
Updated: Aug 6, 2026
08:20
Superior Auto-Identification of Trypanosome Parasites by Using a Hybrid Deep-Learning Model
Published on: October 27, 2023
Ensemble Machine Learning for Malaria Diagnosis in Resource-Limited Settings Using Clinical and Demographic Features
Panashe Nyengera1, Hilary Takunda Takawira1, Farai Fredric Mlambo2
1Department of Applied Biosciences and Biotechnology, Midlands State University, Private Bag 9055, Gweru, Zimbabwe.
Infectious Disease Reports
|July 24, 2026
Summary
Machine learning models accurately diagnose malaria using common symptoms and patient data in Sub-Saharan Africa. This approach offers a cost-effective alternative to traditional methods, improving healthcare in resource-limited settings.
Area of Science:
- Medical Informatics
- Computational Epidemiology
- Public Health Technology
Background:
- Sub-Saharan Africa faces a disproportionate malaria burden, accounting for 94% of global cases and 95% of deaths.
- Weak health systems and limitations of current diagnostics (microscopy, mRDTs) hinder malaria elimination efforts in resource-limited settings.
- This research addresses diagnostic gaps by developing a machine learning (ML) framework using readily available clinical and demographic data.
Purpose of the Study:
- To develop and evaluate a machine learning (ML) framework for malaria diagnosis.
- To leverage routine clinical symptoms and demographic information for improved diagnostic accuracy.
- To provide a scalable and cost-effective diagnostic tool for resource-limited environments.
Main Methods:
- Analysis of 637 patient records from Zimbabwe, including clinical symptoms (fever, chills, etc.) and demographic data.
- Data preprocessing techniques: Synthetic Minority Oversampling Technique (SMOTE) for class imbalance and Recursive Feature Elimination (RFE) for feature selection.
- Training and ensemble modeling (Bagging, Stacking, Voting, AdaBoost) of seven ML algorithms (Logistic Regression, Random Forest, etc.) with performance evaluation using accuracy, precision, recall, F1 score, and AUC-ROC.
Main Results:
- Chills, fever, diarrhea, and abdominal pain were significant clinical predictors of malaria.
- Travel history emerged as a significant demographic predictor.
- The stacking ensemble model achieved high performance: 0.96 accuracy, 0.95 precision, 0.98 recall, 0.96 F1 score, and 0.98 AUC-ROC.
Conclusions:
- Machine learning, especially ensemble techniques, shows significant potential for enhancing malaria diagnosis and management.
- This ML framework offers a scalable, cost-effective diagnostic alternative utilizing accessible data.
- The approach supports healthcare workers and malaria control programs in areas with inadequate traditional diagnostic methods.
