UCLA BAIR Lab uses machine learning to analyze health care data, improve patient lives

The UCLA Biomedical Artificial Intelligence Research Lab is using machine learning to improve the lives of patients.

Machine learning is a field of AI that learns from existing data to make predictions relationships regarding the population of the data, according to UCLA Extension. Researchers at the lab use machine learning to find associations in health care data, Corey Arnold, the director of the BAIR Lab, said.

Arnold decided to combine the engineering and clinical aspects of health care, as he recognized AI’s potential to analyze existing health records with machine learning.

The lab applies machine learning to various forms of health care data to analyze potential health risks, according to Arnold. The data include imaging and textual data, as well as information on medical conditions, such as cancer and mental illness

The lab developed models that are trained on anonymized records with the goal of predicting future patient outcomes, Arnold added.

“We all generate very rich information in the health care record,” Arnold said. “There’s a question of, ‘Maybe that information could tell us something about our future risk.’”

One of the lab’s most notable projects was finding associations between its saved data – measured by a mobile activity tracker – and patients with known heart failure. The lab related the association to a specific predictor of a patient having a heart failure-related event, which was a significant correlation between daily steps and daily symptom severity.

Arnold added that the health care industry has delayed incorporating machine learning because of the potential consequences for AI-driven mistakes – such as incorrect diagnoses, complications with insurance and, in some instances, patient death.

Sixty percent of American adults reported that they would feel uncomfortable if their health care provider relied on AI, according to a 2022 survey conducted by the Pew Research Center.

Clinicians work with the BAIR Lab and provide direct feedback on its application to the real-world health care system, Arnold said. The clinicians give insight on how useful or applicable it is in a clinical setting, along with potential issues that it caused, according to Mara Pleasure, a doctoral student in bioinformatics in the BAIR Lab.

“We also really want to focus on the problems which will have an impact in this highly complex health care space,” Arnold said.

Pleasure said clinicians’ expert opinions and perspectives on models can help researchers finalize plans for the model.

Arnold added that the lab’s ultimate goal is to develop an algorithm that would warn patients of health risks based on mobile activity tracker data.

“I feel that my work is doing a difference when I talk to other people,” Pleasure said. “At conferences, I talk to other researchers who are also excited about what I did and can see the novelty, either in the method or even the application is interesting to them.”

The BAIR lab is also applying AI to digital pathology.

AI is able to analyze a given dataset faster and more effectively than humans, Arnold said. This technology can analyze radiology imaging to determine potential diagnoses, such as cancer, and predict outcomes.

“It’s a really good task for AI to think about things that humans might not be as good at,” Arnold said.

Arnold added that he wants to enhance the scale of which the BAIR Lab’s research is done.

“One area that I would really like to grow our lab is to build a larger learning laboratory, where we can really shorten the timeline of innovation; we can more quickly test something out in ‘the real world’ and know what its impact is going to be,” Arnold said.

Katya Redekop, a doctoral student in bioengineering in the BAIR Lab, said she has worked on various projects in the lab – such as developing AI tools and frameworks for electronic health records data and digital pathology.

“I personally get really excited by this novelty and being the first one to explore something, to deliver results, which are state of the art, and also its all application in health care,” Redekop said.

Federal funding cuts have not impacted the BAIR Lab, Arnold said. The Trump administration froze $584 million in federal research funding from UCLA in late July due to alleged “antisemitism and bias,” leaving many researchers and their labs without grants.

A federal judge filed temporary injunctions in August and September, which reinstated UCLA’s National Science Foundation and National Institutes of Health grants, constituting the bulk of UCLA’s previously frozen funding. The decisions, which came in response to a case brought by UC researchers, will hold while the case moves through the courts.

Arnold said he hopes the implementation of AI in health care will allow for improved patient-physician interaction.

“This is something that will be transformative for physicians, who are overworked and burned out and have less and less time to meet with patients,” Arnold said.

By Jiyoon Choi

SOURCE: Daily Bruin