As the world is buzzing with the latest advancements in generative AI, a group of physicians, hospital administrators and tech experts banded together to build out large language models specifically for healthcare.
Hippocratic AI launched out of stealth armed with $50 million to build out what it refers to as the first LLM for healthcare with an initial focus on non-diagnostic, patient-facing applications.
This approach will allow the company to ensure patient safety while improving healthcare access and outcomes, executives said.
"I think the whole world and even healthcare and investors are very excited about this platform shift that has occurred with generative AI and how to utilize it," said Munjal Shah, a serial entrepreneur who co-founded Hippocratic AI earlier this year.
The startup is focused on using language models to massively increase healthcare access, reduce costs and close the healthcare skills gap left behind by the global pandemic.
"The thing I'm really excited about is what I call 'healthcare maximalism', which is this idea that right now the healthcare staffing is to a certain level of service. But if the cost of healthcare could be brought even lower, we could provide an even higher level of service. And what would that do to outcomes? What happens to health care outcomes if every person in the country has their own free dietitian? We just realized that there's some really big potential here, not just to close the gap, but to bring a whole different level of healthcare to the country," Shah said in an interview.
The healthcare industry is facing a critical workforce staffing crisis, said Meenesh Bhimani M.D., Hippocratic AI co-founder and chief medical officer.
"We now have this incredible opportunity to really augment our workforce, reduce fatigue, reduce burnout, and get healthcare workers back to the frontline and back to the bedside. Workers can now reach out to patients and connect with patients in ways that we've just never had the ability to do before," said Bhimani, who also serves as chief operating officer at El Camino Health.
Bhimani and Munjal teamed up with physicians, hospital administrators, Medicare professionals, and artificial intelligence researchers from El Camino Health, Johns Hopkins, Washington University in St. Louis, Stanford, UPenn, Google and Nvidia to form Hippocratic AI earlier this year.
Prior to this new venture, Munjal was co-founder and CEO of Health IQ which used AI to analyze the health records of seniors and help them select the best Medicare Advantage plans.
Hippocratic AI is backed by General Catalyst and Andreessen Horowitz.
Healthcare leaders are exploring how to use large language models (LLMs) and foundation models (FMs) like ChatGPT and GPT-4 in healthcare. While researchers have shown that these AI models can pass the U.S. Medical Licensing Exam (USMLE) no company has built a commercial model specifically tuned for healthcare applications.
"The healthcare industry needs its own AI platform, one that is focused on empowering the workforce, reducing burnout, and improving patient safety and experiences with the healthcare system. We joined forces with the Hippocratic AI team, our health assurance ecosystem, and the a16z team to build this platform. Our goal is to fundamentally increase the supply and scalability of healthcare professionals. This is the key to achieving the health assurance vision: a more proactive, more affordable, and equitable system of care for all," said Hemant Taneja, CEO and managing director at General Catalyst.
Shah and Bhimani see an opportunity to harness AI to bring deep healthcare expertise to every human. The startup is taking a safety-first approach to build out a large language model for healthcare with an initial focus on certification, bedside manner and incorporating feedback from healthcare professionals.
Other large language models have gotten attention for passing medical certification exams. But passing the USMLE is not enough to ensure a model is ready for the wide variety of healthcare roles that
exist in care and payer settings, according to startup executives.
Hippocratic AI focused on testing its model on a wide variety
of 114 healthcare certifications and roles with a goal of not just getting a passing score but outperforming existing language models such as GPT-4 and other commercially available models, executives said. The company was able to outperform GPT-4 on 105 of the 114 tests and certifications, outperform by 5% or more on 74 of the certifications and outperform by 10% or more on 43 of their certifications.
The company has also developed a novel benchmark measuring the bedside manner of large language models.
"In healthcare settings, it isn’t just important to answer the patient accurately. It is equally important that it is done with great bedside manner. Many studies have shown that bedside manner impacts emotional well-being and quality of outcomes. This isn’t just true for doctors but also true for everyone interacting with patients: billing agents, schedulers, and more,” Bhimani said.
Hippocratic AI plans to release the first of many bedside manner benchmarks for the entire industry to use. So far, the company says its language model scored 75% for showing empathy, 85% for showing care and compassion and 57% for making patients feel at ease.
Hippocratic AI also is using a technique called Reinforcement Learning with Human Feedback (RLHF) by using healthcare professionals to help guide and train the LLM by rating its responses. Many believe this technique is what led to the remarkable performance of ChatGPT compared to that of prior versions of OpenAI’s language models, Shah said.
Some of the roles and tasks the company is exploring include patient navigator, dietician, genetic counselor, enrollment specialist and medication reminders.
Healthcare professionals will determine when Hippocratic AI's LLM is ready for deployment in the healthcare system, Shah noted.
“RLHF with healthcare professionals isn’t just a feature but is really our commitment to partner deeply with the industry,” Shah said. “We aren’t just saying these professions will help us evaluate our system. We are saying we won’t launch each unique role for the LLM unless the professionals who do that exact task today agree the system is ready and safe.”
Julie Yoo, general partner at Andreessen Horowitz, said as a healthcare and tech operator Munjal understands what it takes to bring high-ROI products to market - especially at a time when existing industry players are in such dire need of better operating leverage and financial sustainability.
"We believe Hippocratic AI’s cross-disciplinary, safety-first approach is what the healthcare industry needs to be able to maintain trust in the power of responsible deployment of generative AI solutions," Yoo said.