ELECTRONIC HEALTH RECORDS ANALYSIS OF LEPROSY PATIENTS USING MACHINE LEARNING TECHNIQUES
Electronic Health Records (EHRs) are rapidly being implemented by health care providers in the recent years. This has given rise to increase in the availability and quality of EHR data. Leprosy is one of the main public health problems and listed among the neglected tropical diseases in India. It is also called Hansen's Diseases (HD), which could be a long-term contamination by microorganisms, mycobacterium leprae. The delay in the diagnosis of leprosy can lead to increase disability rate among various patients. This paper intends to identify type of leprosy by applying Machine Learning based classification techniques on various leprosy cases from the first sign of symptoms recorded in clinical text included in Electronic Health Records (EHRs). Electronic Health Records (EHRs) of Leprosy patients from verified sources have been generated. The clinical notes included in EHRs have been processed through various Natural Language Processing techniques. In order to predict type of leprosy, Rule based classification method has been applied in this paper. Further the classification results of various Machine Learning (ML) algorithms like Support Vector Machine (SVM), Logistic regression (LR), K-nearest neighbor (KNN) and Random Forest (RF) are compared and their performance parameters are analyzed.
Electronic Health Records, Natural Language Processing, Clinical notes, Machine Learning, Support Vector Machine, Logistic Regression, K-nearest neighbor, Rule based classification, SNOMED-CT, Random Forest