Serial entrepreneur Munjal Shah’s newest healthcare startup, Hippocratic AI, aims to leverage large language models (LLMs) to provide personalized support for patients managing chronic conditions. With generative AI chatbots like ChatGPT capturing public attention, Shah sees an opportunity for similar technology to make a real difference by improving outcomes for the 68 million Americans with multiple chronic illnesses.
“Why don’t we have 68 million nurses and what would happen to health care outcomes if we did?” asks Shah. Rather than replacing human nurses with AI, Hippocratic AI lo,oks to augment overstretched staff. Shah talks about “super staffing” to close the gap between patient needs and availability of the few hundred thousand chronic care nurses nationwide. With over 600,000 nurses projected to leave the field by two 027 due to factors like burnout, Shah believes AI can reduce nurse workloads while expanding access.
Unlike the classifier AI Shah worked with previously at e-commerce companies, LLMs like ChatGPT generate original responses adapted to specific inquiries. Their ability to compile medical knowledge and communicate it to patients makes them well-suited for non-diagnostic applications, SShah says. Diagnosis itself may remain unsafe for AI owing to the risk of passing along inaccurate information.
“In diagnosis, Shah concedes. “That’s why I said, ‘Let’s not do diagnoses.’ Focus on all the other applications in health care.”
Hippocratic AI instead looks at use cases like chronic care reminders, diet/lifestyle suggestions, appointment bookings, and explaining insurance bills. Patients may not need a nurse just to get negative test results, Shah points out. Untangling convoluted tasks also takes up scarce nurse time better spent on higher-value tasks.
What makes Hippocratic AI’s LLM uniquely qualified is its training on healthcare-specific data, including medical research and real-world insurance policies. This domain-specific content allows more accurate responses than a general chatbot like ChatGPT, according to Shah. But the human guide, according to Shah, remains essential: “You need the medical professionals who do that job today to say it’s rather than speculating on hypotheticals, the LLM relies on the empirical knowledge imparted by its training data. Shah sees communication as the biggest opportunity for immediate impact, allowing more patients to benefit from personal support. Suppose it can automate simple, inform most significant exchanges at scale. In that case, an LLM may help the overtaxed healthcare system offer the level of care Shah believes is possible with the right technology.
While AI cannot replace human judgment in sensitive medical decisions, Munjal Shah envisions a world where patients have an intelligent assistant empowering them to better manage their health. If generative chatbots reflect the knowledge given to them, Hippocratic AI hopes to imbue one with the ability to manage their health better and make knowledge more accessible to all. For Shah, focusing AI on patients’ daily health struggles could demonstrate the technology’s benevolent potential.