Optimization of a hybrid forward chaining and certainty factor model for malaria diagnosis based on clinical and laboratory data

Abstract

Malaria remains a serious public health problem in Indonesia, particularly in Papua Province, which accounts for 89% of national malaria cases. The similarity of malaria symptoms with other infectious diseases and limited laboratory facilities often lead to delays and inaccuracies in diagnosis. The study proposes an optimized hybrid model that combines forward chaining and certainty factor (CF) by integrating clinical and laboratory data to improve the accuracy of malaria diagnosis. The research design includes acquiring knowledge from medical experts, developing a rule-based system using forward chaining, and applying CFs to overcome uncertainty in symptom interpretation. The system is implemented using Python with support from libraries such as NumPy and PyKnow. The test results showed that the integration of laboratory data significantly improved diagnostic performance, with accuracy increasing from 81% malaria-positive using clinical data alone to 98% malaria-positive after combining with laboratory data. Expert testing to validate the accuracy of clinical and laboratory data results compared to expert validation results in an accuracy score of 98%. These findings show that the optimization of the hybrid forward chaining model and CF for malaria diagnosis based on clinical and laboratory data as a recommendation tool for early diagnosis of malaria in endemic areas.