Conversational Chatbot for Enhancing Healthcare Services

In healthcare industry, conversational chatbots have emerged as crucial tool that offers several benefits, including personalized healthcare advice, assisting in early diagnosis, and ensuring timely intervention. In order to provide patient-centric healthcare services, it is necessary to understand and capture both intents and slots accurately during the conversation. In recent years, various researchers have utilized Sequence-to-Sequence (Seq2Seq) learning models with Recurrent Neural Networks (RNNs) for intent classification and slot (entity) extraction during conversation in chatbots. However, these models tend to overlook crucial information within the context of patients’ inputs and exhibit limitation to capture the dependencies from long-term complex input sequences. To overcome these limitations, this research work proposes a multi-attention mechanism that allows the learning model to understand and capture different aspects of users’ utterances. The study aims to develop an end-to-end conversational chatbot that focuses on intent classification, slot extraction, managing dialogue, and response generation during patient’ health-related conversation. The proposed model leverages transformer-based multi-head attention mechanisms to effectively handle the complexities of natural language understanding in healthcare contexts. Our experimental and numerical results corroborate the superior performance of proposed model as compare to baseline techniques. Additionally, the dialogue management and response generation capabilities of chatbot are evaluated through patient-chatbot interactions, confirming its effectiveness in managing dialogues and providing appropriate responses.