On the Efficiency of Machine Learning Models in Malaria Prediction

Area of Science:

• Medical Informatics
• Public Health
• Machine Learning

Background:

• Malaria remains a significant public health challenge in Sub-Saharan Africa, particularly Senegal, accounting for 35% of hospital consultations.
• Late and inaccurate diagnoses, coupled with unreliable diagnostic tools like Rapid Diagnosis Tests (RDTs), hinder effective malaria management.
• There is a critical need for improved diagnostic accuracy to support healthcare providers and reduce disease burden.

Purpose of the Study:

• To evaluate the efficiency and accuracy of popular machine learning models for predicting malaria occurrence.
• To compare the performance of these models against established diagnostic methods using patient sign and symptom data.
• To identify the most effective machine learning algorithm for malaria diagnosis in the Senegalese context.

Main Methods:

• Utilized sign and symptom data from patients in Senegal to train and test various machine learning algorithms.
• Focused on popular and widely applicable machine learning models for predictive analysis.
• Evaluated model performance based on key metrics including accuracy, precision, and recall.

Main Results:

• Machine learning algorithms demonstrate significant potential for malaria prediction.
• The Naive Bayesian model exhibited a recall rate comparable to Rapid Diagnosis Tests.
• Notably, the Naive Bayesian model achieved a 9% improvement in precision over RDTs.

Conclusions:

• Machine learning, particularly the Naive Bayesian algorithm, offers a promising advancement in malaria diagnosis.
• Improved precision in diagnosis can lead to more effective treatment and better patient outcomes.
• These findings support the integration of machine learning tools into healthcare systems for enhanced malaria control in resource-limited settings.