The investment opportunities for generative AI in health care

Adobe

OpenAI’s ChatGPT has prompted a boom in interest in how generative AI technologies might change health care that has already resulted in high-profile investments.

A new report by venture capitalists from GSR Ventures and Maverick Ventures explores the potential for the novel tech across the health care ecosystem. Of interest, the authors chart uses on a matrix based on the complexity of the application and its level of adoption. Many of the simpler, more developed opportunities, like medical note taking and patient communication have already gotten a reasonable amount of attention. But applications in emerging fields like personalized medicine are difficult and still in their infancy.

The report also spells out the challenges startup founders face. Some, like long sales cycles, are established foibles of the health care system, but others are particular to this AI moment. Huge players like Google and OpenAI have made AI implementation as simple as an API call, the authors write, concluding that “as these models grow in popularity due to their efficiency and accuracy, new startups face the challenge of maintaining their technology defensibility in the industry.”

Mental health

Can AI provide a lifeline to 988?

Counselors who answer calls to crisis lines are carefully trained to manage sensitive situations and to complete important checks like screening for suicide risk. The quality of their work is reviewed at random, a process that’s inefficient and may not pick up problems. And the challenge is getting more difficult as crisis lines face a surge in demand following the launch of the 988 Lifeline last year.

Protocall Services, which operates crisis lines, and Lyssn, a startup that develops AI which automatically reviews recordings of behavioral health encounters, have teamed up to see if they can use tech to improve quality. The companies were recently awarded a $2 million grant from the National Institute of Mental Health to develop and test AI that checks whether counselors properly screen for suicide.

Read more in my story here. 

Startups

A company aims to parse AI options

Buying an AI tool in health care is a lot like buying a used car. It’s difficult to discriminate between clunkers and sturdy models, especially when staring into the eyes of charismatic salesperson. But newly-formed Dandelion Health says it can help customers take a deeper look under the hood. The New York-based company has compiled data from three very different health systems in Texas, California and South Dakota to independently test algorithms for their generalizability and vulnerability to bias. The firm’s first pilot will be free of charge and focus on algorithms that use electrocardiograms to predict heart problems, a hot area of AI development.

Read more from STAT's Casey Ross here.