A Single Soft Actor Critic Reinforcement Learning Technique for Prescription Patterns Prediction and Analysis in South Indian Region

Abstract

            The healthcare professionals prescribe the inappropriate antibiotics to the patients is a worldwide concern. Therefore, analysis and prediction of prescription patterns would be interesting and useful goal from multiple perspectives. Subject-matter experts evaluate the identified prescription patterns for further investigation, i.e. identification of disease causing conditions and provides better prevention to end users about spreading of diseases. This work proposes a single shot soft actor critic reinforcement learning technique to prescription pattern and disease spread nature. The purpose of this study was to determine and investigate the pattern of disease spread among people and healthcare providers in South Indian Regions. In this research study, the data was collected from various healthcare clinics in major three metro cities of South India (i.e.100 clinics for each city) for a period of 3 years. Proposed solution was evaluated to perform better in terms of predicting the disease and suggesting the suitable treatment to user with 0.415 Root Mean Square Error (RMSE) and 0.32 Mean Absolute Error (MAE) in comparative to other existing artificial intelligence techniques.