MALrisk: a machine-learning-based tool to predict imported malaria in returned travellers with fever

Area of Science:

• Infectious Disease Epidemiology
• Artificial Intelligence in Healthcare
• Travel Medicine

Background:

• Early diagnosis of Plasmodium falciparum malaria is crucial for reducing mortality in international travelers.
• Limited access to diagnostic tests in some healthcare settings hinders timely malaria detection.
• Artificial Intelligence (AI) offers potential to enhance the management of febrile travelers.

Purpose of the Study:

• To develop and validate a machine-learning model for predicting malaria cases in febrile travelers.
• To identify key clinical and demographic features predictive of malaria in this population.
• To create a tool for improved malaria diagnosis in resource-limited settings.

Main Methods:

• A multicentric prospective study collected data from febrile travelers.
• Eleven machine-learning models were trained and validated using cross-validation.
• The best-performing model (XGBoost) was optimized and a reduced model (MALrisk) was developed.

Main Results:

• XGBoost achieved the highest performance with an AUC of 0.98.
• The reduced MALrisk model, using six features, demonstrated 100% sensitivity and 72% specificity.
• Key predictive features included travel to Africa, low platelet count, rash, respiratory symptoms, hyperbilirubinemia, and chemoprophylaxis.

Conclusions:

• The MALrisk model can facilitate early malaria identification in non-endemic areas.
• It supports initiating empiric antimalarials and urgent transfers when diagnostic tests are unavailable.
• Scalable as a digital tool, MALrisk can reduce malaria-related morbidity from delayed diagnosis.