Monitoring, Tracking and Diagnosing Dengue Fever using Smartphone, GPS and Machine Learning

Abstract

Diagnosing infectious disease using smartphone and ML has emerged as popular research area. Many tropical nations including Pakistan are suffering from a viral disease i.e. Dengue. It can be recognized by its symptoms. Due to exhausted pressure of patients i.e. Covid-19 in hospitals and early monitoring, tracking and diagnosing of dengue epidemic is a real challenge to the authorities. Moreover, currently there does not exit any application to diagnose DF and SDF. Hence, we proposed a model, developed an android application, conducted pilot testing and apply ML. Whereas, WHO recommended symptoms of dengue are adopted. A pilot study is conducted on 80 participants. It revealed that the smartphone technology along with GPS on particular symptoms is helpful for early detection. Furthermore, the incorporation of GPS technology is useful for the surveillance during an epidemic or pandemic. Moreover, we also collected data of the last six-year dengue infection from hospitals for applying ML classification techniques using WEKA on clinical features of the patients. The results are compared in terms of Precision, Recall, F-measures and Accuracy to evaluate the performance of SMO, J48, Naïve Bayes, Random Forest and ZeroR classifiers. The performance of the Random Forest classifier has been achieved 98.8% using 10-folds cross-validation and 66% percentage split techniques.