Serial entrepreneur Munjal Shah has launched a new health tech startup called Hippocratic AI that aims to leverage large language models (LLMs) to help patients with nondiagnostic needs. The goal is to use AI to “super-staff” the healthcare system and provide patients with more support and guidance, which nurses and other providers usually offer.
Shah points out the growing shortage of healthcare workers in the U.S. and globally. With over 68 million Americans suffering from multiple chronic conditions and only a few hundred thousand dedicated chronic care nurses nationwide, there is a massive gap between patient needs and available staff. Hippocratic AI hopes to use LLMs to help fill this gap.
LLMs like ChatGPT have shown an ability to communicate, synthesize knowledge, and provide personalized responses to unique questions and situations. While risky for diagnostics, Shah believes these models can safely assist with chronic care reminders, diet advice, specialist referrals, and administrative tasks. The AI could take over communicating simple test results or explaining complex billing details, freeing up human providers for more critical care.
Shah emphasizes that the goal is augmentation, not replacement, of healthcare workers. By handling more routine tasks, the AI could reduce provider burnout and improve health outcomes by multiplying the number of available ” staffers.” With the U.S. potentially losing over 600,000 nurses by 2027, AI assistance could be a crucial way to maintain care standards.
Training Rigor Needed for Healthcare LLMs
While optimistic about the potential, Munjal Shah acknowledges concerns over false or erroneous information from AI systems. He stresses that Hippocratic AI’s LLM will be trained on peer-reviewed medical literature and real-world insurance policies. This healthcare-specific training is essential.
The LLM will also need extensive feedback from doctors, nurses, and other human experts in the communication and services it aims to provide. Shah says medical professionals need to rigorously validate that the system is ready before interacting with actual patients. Ongoing monitoring and refinement of the AI based on clinical expert assessment will be critical.
Human judgment remains vital in health care, especially for diagnostic or treatment-related issues. But I like Hippocratic, as it could efficiently extend expert knowledge to more people for nonclinical assistance and communication.
The Vision: AI and Humans Working Together
Rather than AI versus human providers, Shah envisions collaborative systems where AI handles routine information sharing and simple service needs. This could work in chronic care management, diet support, administrative explanations, and more—meanwhile, precious clinical staff can focus on critical diagnosis and care decisions that require deep medical training.
The future Shah describes isn’t replacing nurses and doctors with AI bots. It’s about AI and human providers working together to improve health outcomes. With growing chronic disease and an aging population taxing the overstretched healthcare system even further, augmentative AI may provide part of the solution.
Of course, realizing this future depends on discipline and patience from AI developers. Responsibly training LLMs on quality medical data and validating their outputs will be challenging but essential work. Shah and the Hippocratic AI team understand that building safe, trustworthy AI for such a critical application area will take time. Still, augmentative systems could help provide patients with the level of care and communication currently out of reach for even the wealthiest nations.
Comments are closed.