Development Of An Intelligent Decision Support System For Prompt Diagnosis Of Ebola And Lassa Fever Diseases.

ABSTRACT

Ebola Virus Disease (EVD) and Lassa fever are infectious viral diseases that are very deadly to mankind. These diseases, when handled lightly can quickly degenerate into deadly epidemics. Accurate and prompt diagnosis, and effective treatment of these infectious diseases is a very critical factor in their prevention and containment. The difficulty in differentiating between EVD and Lassa fever at their initial phase can result in wrong diagnosis which can be catastrophic. An intelligent decision support system can help make faster and more accurate diagnosis of these diseases.

In this research, a decision support system for diagnosis of EVD and Lassa fever was developed. Patient clinical history, demographic data, treatment data, and prognosis were obtained from Lagos State University Teaching Hospital (LASUTH) and Irrua Specialist Teaching Hospital, Irrua, Nigeria for 1000 cases of Ebola virus disease and Lassa fever respectively. The data was encoded in a way suitable for input into the learning algorithm. Conditional inference tree and Support vector classifier with sigmoid, polynomial and radial basis function kernel were used to build a classification model for the diseases. 10-fold cross validation technique was used for model validation. Evaluation was based on accuracy, precision, sensitivity, and recall. Results show that the Conditional inference tree performed best with 99% accuracy and 367 true positive classifications. Support vector machine with Sigmoid and RBF kernels also achieved 99% accuracy, but with lower sensitivity scores.A web based decision support system was built based on the Conditional inference tree. The developed system can enhance and facilitate decision making process by leveraging on stored knowledge to correctly classify EVD and Lassa fever and assign necessary treatment to the patients.