How Generative AI Is Transforming Healthcare
The world is aging. After centuries of relentless growth, many advanced economies are getting older, and even in poorer nations, the share of elderly people is rising. Larger and older populations are creating historic pressures on health care systems across the world. There are not enough doctors, nurses, and other health care workers, even as the number of patients increases. Health care workers are under constant pressure, with many suffering from burnout and others planning to leave the industry. According to the New England Journal of Medicine, in 2022, exhausted from the pressures they were under, 52% of nurses and 20% of doctors said they wanted to leave the industry. Data from the Health Resources&Services Administration (HRSA) shows that, where health workforce shortages have been identified, the United States needs more than 13,023 primary care practitioners, 9,926 dental health practitioners, and 6,140 mental health practitioners. The impact of this is a steep reduction in the quality of healthcare. Researchers believe that generative artificial intelligence (AI) could be used to improve the quality of care by making it possible to do more health care work with the health workforce that we have.
Source: HRSA
Over a decade ago, venture capitalist Marc Andreessen said that “software is eating the world”. Since the launch of ChatGPT in November 2022, generative AI has taken up such a space in the cultural consciousness and received so much investment, that we can now say that AI is eating the world, disrupting industry after industry, and offering novel use cases from editing code to writing essays. Health care has not been left behind. Although AI has been around for decades, so much so that researchers joke that AI is what can’t be done until it can be done, generative AI brings something very different. Scientist Stephen Wolfram explained that,
‘The first thing to explain is that what ChatGPT is always fundamentally trying to do is to produce a “reasonable continuation” of whatever text it’s got so far, where by “reasonable” we mean “what one might expect someone to write after seeing what people have written on billions of webpages, etc.”’
In other words, ChatGPT and generative AI tools in general, make probabilistic forecasts of what should appear, based on training data that today is really the entire internet. Generative AI produces content without understanding what that content is. Generative AI can be trained on very specific personal, or company or academic data, which reduces the computational needs and costs, while offering specialized technology. The average hospital produces 50 petabytes of data a year, or 137 terabytes a day, and, given the rise in the amount of technology is used and patient-generated health data is collected and the universal use of electronic health record (EHR) systems, the amount of data generated has been rising at a rate of 47% a year.
In a report on use cases of generative AI in healthcare, McKinsey gave the example of a ChatGPT-enabled technology which adds patient information to a mobile platform in real-time. A clinician records a patient’s visit, and the platform converses with the clinician to fill in any gaps, turning that conversational language into a structured note. After the visit, the clinician can edit the notes by voice or typing, before submitting the notes to the patient’s EHR. This reduces the friction of note-taking and speeds up administrative work, freeing the clinician for the actual practice of health care. That example reflects generative AI’s ability to take unstructured data sets, analyze them and turn them into structured content. That patient’s EHR, for instance, can quickly be attached to an insurance claim.
Broadly, companies such as Mercy Health, and Intermountain Healthcare have developed generative AI-enabled platforms to automate bookings, patient registration, refilling prescriptions, and other administrative tasks. Mercy Health, for example, partnered with Microsoft to develop a tool that allows lab results to be communicated, appointments to be scheduled, patients to be given recommendations, and, for its employees, provides HR services, and information on company policies and procedures. Mercy Health saved $30 million in 2023 thanks to these tools, savings which can be poured into improving health care quality.
Generative AI-enabled technologies can be used to guide patients through treatments for simple health problems, directing patients to the doctor for more serious or complex problems. This would ensure that doctors can focus on more serious cases, and patients could get rapid treatment for simple health problems. Sanofi, the French pharmaceutical company, has done this and also partnered with OpenAI, the not-for-profit that developed ChatGPT, to help with drug discovery. The company’s CEO, Paul Hudson, said that,
“This unique collaboration is the next significant step in our journey to becoming a pharmaceutical company substantially powered by AI. Next generation, first-of-its kind AI model customizations will be an important foundation in our efforts to shape the future of drug development for pharma and for the many patients waiting for innovative treatments.”
Not only is generative AI being used on its platform to help with treatments, they believe generative AI can be used to fine-tune its models, which makes sense given that generative AI is pretty good at coding and debugging code. Generative AI seems to shine at these kinds of jobs.
It is yet to be seen how far this revolution can go. A crucial element I think will be in getting better data. We have already seen a slow down in advances in generative AI, after they have basically trained themselves on all publicly available data. The next step in advances will require improving that data to deepen insights and open up novel use cases.
