Munjal Shah, founder and CEO of health care startup Hippocratic AI, believes that large language models like ChatGPT have an obvious and immediate application: providing non-diagnostic patient care. His company aims to harness the power of generative AI to help address systemic issues in the U.S. healthcare system, including rising costs and widespread staff shortages.
With $50 million in early stage funding and the support of top VC firms, Hippocratic AI sees an opportunity to use AI for tasks like chronic care management, patient education, and appointment reminders. According to Munjal Shah, these conversational health services are well-suited to large language models, which can understand concepts and language much like humans do. And by focusing strictly on non-diagnostic use cases, his company aims to provide care safely, efficiently and at scale.
How Hippocratic AI is Training Its AI to Provide Patient Care
Unlike narrow AI models that rely on pattern recognition, large language models like Hippocratic AI learn conceptually, much more like humans do. As Munjal Shah explains, instead of analyzing millions of patient records, these models can simply read a few medical textbooks and derive an understanding of topics like immunology. This conceptual learning allows them to generate novel responses appropriate to the context, rather than just interpolate between data points.
To prepare its AI for delivering patient care services, Hippocratic AI takes a health care-centric training approach. This means ingesting materials like medical research papers, case studies, certification exams and textbooks. The AI model is then refined based on feedback from medical professionals who have real-world experience in areas like chronic care management and patient education.
So far, this training methodology seems promising – Hippocratic AI’s model has passed over 100 medical certification tests and even outperformed advanced models like GPT-4 in healthcare domains. By tailoring the training process to focus on care delivery, Munjal Shah believes his company’s AI can provide safe and high-quality care, upholding the key tenet to “first do no harm.”
Using AI to Help Address the Health Care Staffing Crisis
With over 68 million Americans suffering from multiple chronic conditions, there is tremendous unmet need for continuous, personalized care. But limitations around cost and availability of medical professionals make this level of care unattainable for most patients. AI could help change the math, according to Munjal Shah.
Rather than replace health workers entirely, Hippocratic AI envisions AI as a way to “super-staff” the healthcare system, vastly expanding capacity for chronic care management, patient education and other essential services that rely more on empathy and communication skills than diagnostic ability. As healthtech entrepreneur Munjal Shah argues, large language models can allow every patient to have access to the level of personalized care only the wealthiest receive today.
This vision is especially important given research predicting a global shortage of 10 million healthcare workers by 2030. Between cost constraints around hiring more staff and diminishing supplies of nurses and doctors, there are systemic barriers to meeting rising chronic care demands through traditional staff alone. AI offers a path to provide high-quality care at scale, helping address this growing crisis.
Steps Towards Responsible and Safe AI Integration in Healthcare
While optimistic about AI’s potential, Munjal Shah emphasizes responsible development and safe integration into clinical settings is imperative. Rather than build in isolation, Hippocratic AI is collaborating closely with health systems, clinicians and policy experts to ensure its AI care delivery model aligns to industry best practices.
Some priorities Munjal Shah highlights are ensuring AI has comprehensive training across medical disciplines, establishing external oversight protocols and pursuing various regional and institutional certifications before deployment. Without diligent governance and rigorous real-world validation, healthcare AI risks patient harm – an unacceptable outcome according to the Hippocratic oath.
In addition to safety considerations, truly universal access requires AI that can communicate across languages while maintaining care quality and empathy. As such, Hippocratic AI is investing heavily into multilingual models that can uphold continuity of care across patient populations.
Munjal Shah’s Vision for the Future Healthcare Workforce
Munjal Shah envisions a future dominated by AI “care specialists” that work alongside (not in place of) doctors, nurses and frontline healthcare staff. With stiff competition for skilled talent and ballooning costs, relying solely on human capital to address widening care gaps is likely impossible. Integrating responsive, empathetic and medically fluent AI into workflows allows health systems to extend their capabilities greatly without sacrificing quality, safety or jobs.
Rather than fear this future workforce, Munjal Shah sees profound opportunity for both patients and healthcare professionals. With the right governance and trust built between stakeholders, AI could help democratize chronic care management, reducing costs while providing every patient personalized support. After all, achieving comprehensive and compassionate healthcare requires a team effort between human ingenuity and technological progress. Hippocratic AI aims to walk this line responsibly, upholding duties to empower both patients and overburdened care providers alike through safe and ethical AI.