Artificial Intelligence Powered Amharic Language Conversational Robotic Chatbot System for Selected Ethiopian Hospitals

Abstract

Artificial intelligence-powered chatbots are conversational agents designed to simulate intelligent conversations between intelligent robotic systems and users. The objective of this research is to provide healthcare services to all users in the Amharic language, without limitations, and to offer convenient and costeffective access to healthcare services from the comfort of their homes. This AI-powered interactive chatbot has been specifically developed for Nigist Eleni Comprehensive Hospital, the sole public hospital in Hossaena city catering to a significant population with services ranging from emergency care to referral cases. The study introduces an innovative AI-powered Amharic-based chatbot conversation that serves as a personal virtual doctor for each end user. This chatbot is trained extensively to provide relevant answers to health care related questions and interact with patients in a human-like manner. During the development of this intelligent system, a dataset of 12,127 pairs of questions and answers in Amharic was collected, encompassing greeting sentences, various  user queries, symptom descriptions, disease classifications, specialist details, and well wishes. These datasets were preprocessed using substring algorithms, and the structured datasets were accurately mapped to the artificial intelligence modeling language as AIML files. The Naïve Bayes algorithm was then employed to train the mapped datasets. As a result, the developed prototype offers an interactive Amharic chatbot conversation system for end users. This system can analyze user queries and provide corresponding answers with an accuracy rate of 95.7%.