Gen-AI Health Care Chatbot
In 2024, we started building generative AI based chatbot for healthcare, and our first product was successfully demoed at the LBSL international conference. LBSL stands for Leukoencephalopathy with Brainstem and Spinal Cord Involvement and Lactate Elevation, which is a rare genetic disorder severely impacting a patient’s life. With our chatbot, patients can easily research information about the disease and potential treatments, and medical professionals can discover new info efficiently to better treat the disease.
Recently, we’ve been actively working with a major hospital in TX to help treating gynecologic cancers. With limited resources to handle high volume of multi-lingual patient inquires, hospital staff are overwhelmed with surgeries, patient appointments, research, and reply to patient inquires. Our generative AI based chatbot offloads patient inquires workload, and frees up valuable time for hospital staff to do more research on available treatment and schedule more critical surgeries.
Our chatbot architecture is powerful and flexible. We can easily adapt it to any industry and use case, and demonstrate its values with rich data collected from user interactions. The following diagram illustrates how we architect the chatbot solution end to end.
As the diagram above illustrates, we ingest a variety of documents about the targeted diseases, and convert text in these documents into semantic vectors. When a patient question comes in from either a mobile device or desktop browser to the chatbot, the chatbot queries the vector database, and leverages generative AI to answer the question. In parallel, we collect rich information for our dashboard to monitor the chatbot performance.
This architecture can be easily customized to any industry, any use case, and run on any cloud environment. With data encryption and built-in access control, we protect the identity of the patients, provide easy-to-consume information on challenging diseases and potential treatments, and free up valuable time for hospital staff to focus on critical tasks. We will run a year-long pilot program throughout 2025, and we look forward to sharing our findings to empower hospital systems in the US to operate more efficiently while improving patient engagement and satisfaction.