Can AI reduce healthcare disparities? Debate over health officials

AI has been the talk of healthcare, with excitement building around its potential to improve patient care, but also the need to implement it in an ethical manner. But does the technology have the potential to reduce health disparities?

That was the question posed during a debate at the Reuters Total Health conference held in Chicago on Wednesday. The debate took place between Anil Saldanha, Chief Innovation Officer of Rush University System for Health, and Rebecca Kaul, PhD, senior vice president and head of digital innovation and transformation at Northwell Health.

Saldanha said he takes a negative stance on the topic, stating that AI is not yet there to reduce health disparities.

“In my opinion, it will continue to get stronger for a while,” Saldanha argued. “And I don’t have a time frame in which it will equalize or get better. The reason has nothing to do with AI. Overall, from an industry perspective, we are still struggling to bring health equity to our communities in everything we do in healthcare. So the job is not done yet, and now we are introducing this new paradigm called AI. For example, while AI can deliver improvements in efficiency, AI is not here to replace your doctor. So in no way do I feel like it’s there to help with health equity at this point.”

Kaul, meanwhile, argued that what matters is how people use AI, and she believes people will use it ethically and close the gaps.

“I think we are health care professionals at our core,” she said. “We approach it from a ‘do no harm, do good’ perspective. … It can close the gaps when it comes to providing better access to care, whether that’s by surfacing information for people who may not have in-person access to care. With the help of AI, they can gain more insight into their health status. It can provide doctors in rural communities with access to specialists.”

She added that the technology can make care more personalized and reduce language barriers. She claimed that AI has the ability to translate into more than 80 languages.

Saldanha pushed back on this, saying that data integrity is a major challenge in healthcare and that AI is “only as good as the data it is trained on.” If AI isn’t given the right data, the industry can’t expect to use it to solve inequalities, he argued.

Kaul responded that knowing that datasets often have biases “allows us to train the models, adjust the models to remove those biases, and then also find datasets to feed them to get the kind of diversity in the data introduce for training purposes. ”

In closing, Saldanha emphasized that he does not think AI will help eliminate healthcare disparities. He gave the example of a program used in Chicago called ShotSpotter, which uses sensors to detect gunshots. The program was recently discontinued, although there are efforts to keep it going, according to Block Club Chicago.

“The criticism of this program is that it is consistently deployed in communities of color and neighborhoods with the most disparities,” he said. “The champions say it can help with gun violence and prevention. The jury is still out.”

Kaul argued that the issue is not whether AI is ready to reduce disparities, but whether healthcare organizations are ready to use the technology to close gaps.

“All the signals I see, both from my own organization and from the market, would be a resounding yes,” she said. “There is a lot of discussion about the ethical use of AI. Many of the use cases put forward are about leveling out inequalities. We see governance models that ensure that unintended bad things don’t happen in the use cases that people put forward.”

The crowd seemed to largely agree with Kaul. In a poll shared at the end of the debate, 47% of the audience said they believe AI has some potential to reduce disparities, while 42% say it does so to a significant degree. Another 11% said “no, not really.”

Photo: Sylverarts, Getty Images