Author, venture capitalist and cardiologist Ron Razmi says AI might play a key role in health care’s future, but that shift will be slow in coming. In his new book, AI Doctor: The Rise of Artificial Intelligence in Healthcare, Razmi examines why that is. Cost is a big part of it. Health systems have limited cash for new technology, especially if insurers won’t reimburse them. Razmi talked with Ruth about how that and other problems are slowing AI’s rollout. The interview has been edited for length and clarity. What are the barriers to AI adoption in health care? The buyers of these technologies have to have a return on investment to invest in this kind of thing, and that’s not always clear. Medical centers, pharma companies, people who buy these products can’t be buying 50 apps or 50 new digital products every year. There are one or two use cases they can focus on. Are health systems ready to deploy AI? Digital products, especially health AI products, need data. Our health care information infrastructure was never built with these use cases in mind. Electronic health records were built for billing. A lot of the information is unstructured. There is no accepted template or standards for how you do your notes. Everybody uses their own abbreviations and acronyms, which makes these notes very difficult to work with in terms of extracting what you need to run your model. And information exists in a lot of different places. Without complete medical information all in one place, it becomes extremely risky to run AI models. How do you think we can ensure AI is safe? Insurance companies are saying they’re not going to pay for something unless you’ve shown that it improves patient outcomes and it’s safe. You have to be able to say, “I did a multicenter trial where this algorithm was taken into the real world, and this is how much it improved diagnosis.”
|
No comments:
Post a Comment