Modern Healthcare reporters delve deeply into industry leaders who stand out and make a difference in their organization or field. We heard from Dr. Tiffany Kung, researcher at the virtual rehabilitation center AnsibleHealth, about how the ChatGPT model—using natural language processing to generate text responses based on available data available—may eventually be used in the healthcare industry.
What are some ways that health systems can use artificial intelligence technology like ChatGPT to enhance care and providers?
At AnsibleHealth, we used ChatGPT on a daily basis. We’ve incorporated it into our electronic health records so our providers can use ChatGPT to better communicate with patients, and we’re using it to talk to providers. issue your insurance—to do things like rewrite your appeal if [payers have] deny a claim. All of our providers have gone through training to ensure that everything is identified, so it’s HIPAA compliant.
ChatGPT is currently the most commonly used to communicate with insurance [companies] and do a lot of administrative work, as doctors now spend so much time dealing with things that aren’t directly taking care of patients: paperwork and bills.
Regarding the potential shortcomings of chatbots, where can vendors run into issues with ChatGPT? In what ways is this technology ill-equipped for use in the healthcare sector?
ChatGPT and most other existing AIs are not currently HIPAA compliant. That means it cannot process any sensitive or confidential patient data. That’s really one of its major shortcomings. In order for us to incorporate ChatGPT and other AI more into our daily use, we had to do a lot of rigorous testing. Just like any new drug or any new technology, we need to test its safety, usability, and effectiveness.
You recently led a study in which researchers asked ChatGPT to take the US Medical Licensing Exam. How is the chatbot’s performance in that exam an indicator of its possible effectiveness in medical education?
We are really excited to see that ChatGPT has the ability to pass the US Medical Licensing Exam. It [scored] about 60%, which is the threshold to pass. That’s just 1 to 2 percent performance on this exam.
So it doesn’t mean that ChatGPT has the ability to be your doctor or be a good doctor right now. There is much work to be done. It’s still very early days, but we’re really excited about the potential.
How much do you think some of that potential could be?
There are many different applications. It’s still very early. At AnsibleHealth, we care for critically ill patients: They have respiratory conditions like chronic obstructive pulmonary disease, and they also have other comorbidities like heart disease and kidney disease. Much of what we do is coordinate care between the many doctors and specialists these patients need. We help improve communication between patients, cardiologists, nephrologists and pulmonologists. That’s what AI can do: improve care coordination.
Healthcare leaders have some concerns about chatbot inaccuracies, which could adversely affect patient care. What is your impression of the healthcare industry’s perception of this tool?
In general, health care has a very high demand for anything used to care for a patient. Our standards are very high because we are dealing with patients’ lives. So whatever we use should be as safe as possible.
Also, a lot of doctors are cautious when working with new technology or new drugs. Every day in the hospital we communicate with pagers: the technology is pretty outdated, but it shows that sometimes the medical industry is wary of new technologies and we like the things we feel comfortable with. roof.
