INTEGRATING MACHINE LEARNING AND CLINICAL DATA FOR ACCURATE FOOT PAIN DIAGNOSIS AND PREDICTION

Abstract

Foot pain and problems directly affect the quality of human life. The functionality of the foot and associated functions are also affected. The foot problems can be caused due to multiple factors. The use of predictive algorithms utilizing the electronic health records of the patients and data acquired from other sources is helping the healthcare industry to develop tools that are very helpful in diagnosis and prognosis. Professional physiotherapists have been consulted for restructuring of Foot Health Status Questionnaire (FHSQ). Machine learning model and foot disorder predictive algorithm based on the real-world data collected from the foot disorder patients has been developed that is able to accurately diagnose the foot problem of patient. The machine learning methods GaussianNB, Random Forest, XGBoost, Logistic Regression, KNN, Decision Tree and AdaBoost have been applied to predict foot disorders. The statistical analysis has been performed for finding the association between foot disorders and other variables. The results of the study have provided useful results about association of variables and the frequency charts and tables provide the useful description of general trends of the society towards the foot functionality and problems as well. Strong association is found among various factors and foot pain.