A multilingual chatbot application in India has been developed for cancer patients to answer their common queries in eight Indian languages.
Users can ask cancer-related questions by text or voice in eight Indian languages and receive spoken and written responses in their preferred language.
It is based on two layers of AI. A large language model uses retrieval-augmented generation to answer questions, tapping into a clinician-curated medical knowledge base.
A second AI layer checks whether each response is grounded in the approved knowledge base and flags answers that cannot be verified.
Bhashaverse translates non-English queries into English for BandhuCare's LLM and translates the answer back into the patient's chosen language.
BandhuCare also gathers information about patients' symptoms and experiences between hospital visits.
Rather than using numerical questionnaires, it prompts patients to describe their experiences in a conversational manner.
It then converts their responses into standard patient-reported outcome measure scores and combines them with the users' journal entries to generate concise summaries for clinicians.
【MORE】