Why We Need Patients to Trust AI in Medicine

In this video, Leo Anthony Celi, MD, MPH, principal research scientist at the MIT Laboratory for Computational Physiology (LCP) and an intensive care unit physician at Beth Israel Deaconess Medical Center in Boston, explains the importance of gaining patient trust in order to generate accurate artificial intelligence (AI) models. Celi also calls for all medical professionals to learn data science, so that they can continue to be “at the table.”

The following is a transcript of his remarks:

The ethical concerns around AI revolve around who is using the data and for what reasons.

That brings up the very important point that there has to be complete transparency and accountability of what is being done with the data and who is benefiting from the use of the data. That is the only way that we can gain societal trust.

If the algorithm development is being done behind closed doors within companies that we know are only after increasing revenue, then there will be this mistrust, there will be hesitation in sharing the data. What is then going to happen would be that the algorithms that will be developed and validated and deployed will not be representative of the people who mistrust the researchers, mistrust the system. And they’re the ones who are most likely to benefit from artificial intelligence.

So, we need to create an environment and ecosystem where trust is at the center and, as I mentioned, there is complete openness and transparency and accountability of exactly how we’re using the data and how algorithms are being deployed in healthcare.

That’s where the ethicists would come in, because they have come up with frameworks on how to share data and on how to perform research, and that’s why I think their input is crucial in terms of designing what the system is going to look like.

So, I’m calling all the doctors out there and other healthcare professionals, it’s time to start learning data science. Everyone should be a data scientist. The reason why we have the problems that we have now is that, in the past, we have been passive recipients of medical knowledge. I think that we have to be more active in terms of contributing to and governing that knowledge system, because we have learned from past experience that there could be conflicts of interest from professional societies who are generating treatment guidelines.

To me, this calls for us being active contributors and active wardens of the knowledge system. That means that we need to brush up on statistics, we need to brush up on data science, so that we are able to contribute. Otherwise, if we rely on other people to build the algorithms for us, chances are we’re not going to have products that will lead to equitable outcomes.

So I think the onus is on us to educate ourselves so that we understand data science more, because we are not going back to the previous century. Data is going to increasingly be part of the way we care for patients. And it’s not just the regular amount of data that we are used to; this will be a tsunami of data.

If we’re not able to work with computers on how to sift through that data, how to develop the algorithms, then you’re not at the table and you’re probably on the menu — someone is going to be taking advantage of you.

So, to me, the next generation of clinicians all have to be data scientists. This is also a call to medical educators, that they start looking into incorporating data science into their curriculum.

Please enable JavaScript to view the comments powered by Disqus.