An assistive bot for healthcare using deep learning: Conversation-as-a-service

Abstract

Gone are the days when software was used only for complex mathematical calculations or graphical motions alone. Today, it is software that has exponentially grown to become more powerful and more human—most obviously in applications such as ‘Chatbots.’ The year 2017 marks the Chatbot revolution in various industries like health, career, insurance, customer care support. Artificial intelligence (AI), which is the key player in enabling human-like behavior intelligently, is dramatically changing business. Chatbots, fueled by AI, are becoming a viable option for human–machine interaction. Deep learning algorithms have made it possible to build intelligent machine. In this research, we have developed a HealthBot using TensorFlow and Natural Language Processing (NLP) techniques. There is no denying that efficient patient engagement is a key challenge for all healthcare organizations and any company that can unravel this challenge can effectively earn high returns of investments. Chatbots are one of the major overhauls that hospitals can easily provide more customized care for patients while cutting down on the waiting period. The proposed HealthBot lists the common symptoms; then, based on user’s health issue it gets deeper into the conversations predicting the health problem of the user. Such bots are needed for today’s fast-moving population where they have no time to keep a tab on their health. Neural network implementation adds more accuracy to the responses. The proposed Chatbot model is a retrieval-based bot and of closed domain. Finally, the HealthBot is deployed on the Flask, a Python web development framework.