Health Life

New machine learning method allows hospitals to share patient data—privately

Credit: CC0 Public Domain

To answer medical questions that can be applied to a wide patient population, machine learning models rely on large, diverse datasets from a variety of institutions. However, health systems and hospitals are often resistant to sharing patient data, due to legal, privacy, and cultural challenges.

An emerging technique called federated learning is a solution to this dilemma, according to a study published Tuesday in the journal Scientific Reports, led by senior author Spyridon Bakas, Ph.D., an instructor of Radiology and Pathology & Laboratory Medicine in the Perelman School of Medicine at the University of Pennsylvania.

Federated learning—an approach first implemented by Google for keyboards’ autocorrect functionality—trains an algorithm across multiple decentralized devices or servers holding local data samples, without exchanging them. While the approach could potentially be used to answer many different medical questions, Penn Medicine researchers have shown that federated learning is successful specifically in the context of brain imaging, by being able to analyze magnetic resonance imaging (MRI) scans of brain tumor patients and distinguish healthy brain tissue from cancerous regions.

A model trained at Penn Medicine, for example, can be distributed to hospitals around the world. Doctors can then train on top of this shared model, by inputting their own patient brain scans. Their new model will then be transferred to a centralized server. The models will eventually be reconciled into a consensus model that has gained knowledge from each of the hospitals, and is therefore clinically useful.

“The more data the computational model sees, the better it learns the problem, and the better it can address the question that it was designed to answer,” Bakas said. “Traditionally, machine learning has used data from a single institution, and then it became apparent that those models do not perform or generalize well on

Health article

Why alcohol-use research is more important than ever

Alcohol use disorder (AUD) affects roughly 15 million people in the U.S. People with the condition may drink in ways that are compulsive and uncontrollable, leading to serious health issues.

“It’s the addiction that everyone knows about, but no one wants to talk about,” says George Koob, Ph.D., the director of the National Institute on Alcohol Abuse and Alcoholism (NIAAA).

As NIAAA celebrates an important milestone this year—its 50th anniversary—the institute’s research is more important than ever. Like NIAAA reported earlier this year, alcohol-related health complications and deaths as a result of short-term and long-term alcohol misuse are rising in the U.S.

“Alcohol-related harms are increasing at multiple levels—from emergency department visits and hospitalizations to deaths,” Dr. Koob says. He spoke about NIAAA efforts that are working to address this and how people can get help.

What has your own research focused on?

I started my career researching the science of emotion: how the brain processes things like reward and stress. Later, I translated this to alcohol and drug addiction and investigating why some people go from use to misuse to addiction, while others do not.

What are some major breakthroughs NIAAA has made in this area?

We now understand how alcohol affects the brain and why it causes symptoms of AUD. This has far-reaching implications for everything from prevention to treatment. We also understand today that AUD physically changes the brain. This has been critical in treating it as a mental disorder, like you would treat major depressive disorder.

Other breakthroughs have been made in screening and intervention, and in the medications available for treatment. All of this has led to a better understanding of how the body changes when one misuses alcohol and the proactive actions we can take to prevent alcohol misuse.

What is a misconception