Jefferson Investigates August 2026

Patients with breast and lung cancer are at increased risk of cardiotoxicity, or heart-related damage caused by cancer treatments, because of the proximity of the heart, lungs and breasts. Cardiotoxicity increases the risk of heart attacks, heart failure, and other cardiac conditions. Looking for evidence of cardiac disease in these patients requires reading through hundreds of patients’ health records, which is often too time-consuming to be practical.

Thomas Jefferson University researchers have created an algorithm for large language models (LLMs) that can successfully extract cardiac event data from patients’ electronic health records (EHRs) in a fraction of the time that it takes humans to comb through EHRs. The paper was published in the International Journal of Radiation Oncology, Biology, Physics.

“It took physicians, residents and students a long time to collect patients’ data manually,” says Wenchao Cao, PhD, the study’s first author. “We hope our study demonstrates that this process can be taken over by the large language model, to save time.”

Researchers searched for evidence of cardiotoxicity in patient EHRs using open-source LLMs, an advanced type of artificial intelligence. The prompting framework that they created discovered cardiac event data in patient EHRs 71% to 85.5% of the time. Prompts identified keywords and false positives, so “ruled out a heart attack” wasn’t flagged as a cardiac event.

“You would be surprised to see how many different ways there are to describe a negative situation, like ‘patient denied’ a certain disease,” Dr. Cao says.

Humans and LLMs scoured the EHRs of 411 breast and lung cancer patients. Humans took about two hours to review each EHR, while the LLMs took 20 to 42 seconds.

Further refinement of the LLM framework may advance cardiotoxicity research.

“Hopefully we can identify patients who are more likely to develop cardiotoxicity,” Dr. Cao says. “This could inform a more individualized radiation therapy plan to improve patient  outcomes and fewer side effects.”

Jefferson oncology resident Nilanjan Halder, MD; medical students in Jefferson’s summer oncology program – Isis Lloyd, Moorin Khan, Michael Dichmann and Patrick Faherty – and Jefferson undergraduate Femi Adejolu contributed to the research.

By Lisa Fields

Limb apraxia is a neurological disorder that can occur after damage to the brain’s left hemisphere and is characterized by difficulty in using tools despite normal motor control and cognition. It affects roughly a quarter of stroke survivors and is observed in other conditions, including Alzheimer’s disease.

A popular theory proposes that understanding how to use tools, like a hammer, and understanding how objects physically interact are one and the same. New research from Thomas Jefferson University challenges this and could reshape how scientists understand limb apraxia and eventually inform treatment.

In a study published in Cortex, researchers led by senior author Aaron Wong, PhD, an institute scientist at Jefferson Moss Rehabilitation Research Institute, tested 11 patients who had a left-hemisphere stroke. Participants viewed photos of familiar tools and demonstrated how they would use them. Participants’ physical reasoning was assessed by predicting how objects would interact based on factors like mass and velocity.

The researchers found interesting dissociations. A patient without apraxia demonstrated proper tool use but performed poorly on physical reasoning tests, such as identifying which of two balls was heavier after they collided. Another patient with apraxia showed the opposite pattern. Even though these opposite behavioral patterns were only found in two patients, it showed that the behaviors are independent of one another.

“There’s an assumption that if you’re impaired in one area, you should be impaired in the other,” says Dr. Wong. “And it turns out that’s not true.”

In ongoing research, the researchers developed tool-free tasks modeled after a Rube Goldberg machine. Participants predicted how a ball would move through systems involving ramps and levers. They also completed separate tool-use tasks with unfamiliar tools.

“Again, we could see that there are people who tend to be better at one than the other,” says Dr. Wong.

“If being able to use tools does not rely on physical reasoning, there are other possibilities,” says Laurel Buxbaum, PsyD, co-investigator and institute scientist at Jefferson Moss Rehabilitation Research Institute. “It may rely, at least in part, on mental simulation of body movements. Or tool use may be its own special ability.”

Understanding these puzzling mechanisms could ultimately help researchers develop more targeted therapies for limb apraxia.

By Deborah Balthazar

In a 2021 study, Eric Stulberg, MD, MPH, assistant professor of neurology at Thomas Jefferson University, found an association between lower neighborhood socioeconomic status (nSES) and worse stroke recovery outcomes. “Even after accounting for individual economic status, as measured by their education level and insurance status, those in lower socioeconomic neighborhoods tend to have worse stroke recovery outcomes than those who are in the higher socioeconomic neighborhoods,” Dr. Stulberg says.

While these results have since been replicated by other researchers, it’s still unclear what factors are behind this link. However, stroke is a leading cause of disability in the U.S., and identifying these factors could be essential for improving patient outcomes, particularly in those living in lower socioeconomic neighborhoods. “If we found something that was modifiable, it would basically give us a target to intervene upon, which hasn't been done yet,” says Dr. Stulberg.

In a new study published in JAMA Network Open, Dr. Stulberg’s team hypothesized that healthcare access and neighborhood factors like receipt of emergency treatment following a stroke, neighborhood density of home health services, density of rehabilitation clinics, density of recreation centers, transportation access, and walkability could mediate the link between nSES and stroke recovery. It is one of the first studies to rigorously test the impact of modifiable factors on this link.

To test this, they analyzed data on stroke recovery outcomes, including functional status, quality of life, and depressive symptoms, in a cohort of 2,203 stroke patients in Corpus Christi, Texas, for 90 days post-stroke. Patient nSES was defined based on the census tracts they lived in at the time of their stroke.

While results showed some indication that density of home health services and rehabilitation clinics may mediate some of the association between nSES and stroke recovery, these findings were not statistically significant. No other factors were identified as potential mediators of this link.

Because of this, Dr. Stulberg says, future research should look at other potential mediators like the type of care setting patients are discharged to, time to emergency treatment, and the total amount of therapy patients receive. “I hope that other people will adopt our methodological framework going forward to better understand this link,” says Dr. Stulberg.

By Zoe Cunniffe