AI in Healthcare: Experts Talk Data Privacy and Patient Trust | National News

Innovations in healthcare are happening at a breakneck speed that is increasingly supported by artificial intelligence.

These developments can help with many of the problems facing the healthcare industry today: rising costs, declining accessibility, data silos and cybersecurity. But they bring with them challenges, including earning and retaining clinician confidence and patient trust.

Here are key takeaways from a panel discussion Thursday moderated by U.S. News on “Scaling What Matters: Turning Medical Innovation into Measurable Impact” at the Edison Awards. U.S. News is a media partner for the awards.

“We really believe healthcare needs to be disrupted and transformed,” Dr. Matthew Callstrom of the Mayo Clinic said during the panel.

Callstrom said that when a patient comes in with a difficult problem, the vast majority of time they leave Mayo Clinic with a different diagnosis or a significant change in their treatment, underscoring how difficult diagnosing can be for most organizations. It’s an area where AI can help, he said.

For example, teams at Mayo Clinic are using voice as a biomarker for aortic stenosis, which is the narrowing of the aortic valve in the heart.

“When it calcifies, it narrows,” Callstrom said. “When it narrows, it’s hard to get blood through that valve, and the heart has to pump harder. Eventually, heart failure results if it’s undiagnosed.”

If it’s caught early enough, the patient can get a valve replacement and live a normal lifespan.

“Our team built a classifier on the voice signature of patients that had moderate to severe aortic stenosis,” Callstrom said. “Now we’re making a diagnosis for aortic stenosis with a phone rather than having to do more elaborate testing.”

“AI makes patient privacy really complicated,” Merage Ghane of the Coalition for Health AI, which works toward driving responsible AI adoption in the health sector, said on Thursday.

Even if an image is stripped of identifying information, certain factors can potentially predict an individual’s race, age and even where they live, according to Ghane.

“When you’re dealing with conditions like rare diseases or smaller geographic areas or smaller numbers, … I think you run into problems,” Ghane said.

She recommended gaining training and certifications and involving experts in cybersecurity to protect patient data.

Callstrom at the Mayo Clinic acknowledged the responsibility to the patient when it comes to securing private information, like the voice of the patient needed for the aortic stenosis analysis.

“It is characteristically you, and we know that it’s super important to protect,” he said.

Breakthrough technologies can reshape the future of healthcare, but they still have to earn the trust of patients.

“Every one of us will be a patient in the future, so we need to have a high standard,” Chun-Hao Huang, the co-founder of Algen Biotechnologies, a company with the goal of revolutionizing therapeutic drug discovery through advanced CRISPR gene-modulation and AI, said about data privacy.

The process of developing a new drug is long and costly. The failure rate for developing a new drug is about 90%, meaning only 10% see success and make it to patients.

“The whole process is just not very sustainable,” Huang said.

Huang’s company aims to speed up the process of finding next-generation therapies using CRISPR and AI to reverse engineer disease trajectory. And that involves a lot of data.

Patient privacy remains a top concern, but companies also have to be mindful of how representative their data sets are of the real world.

“If you discover a drug based on just a minority of the patient population, in the end, that drug won’t be that effective for all of us,” Huang said. “If we’re able to capture at an earlier stage the diversity of the patients, then in the end, that can ensure the drug that evolved is effective for every single patient.”