A new artificial intelligence tool developed by University of Maryland researchers can automatically scan patient records, extract buried diagnostic data, and flag cases of pulmonary hypertension that clinicians might otherwise overlook, potentially connecting thousands of patients to specialized care and treatment. The system uses large language model technology to analyze electronic health records and predict missing measurements needed to diagnose this serious heart condition.

Why Is Pulmonary Hypertension So Hard to Diagnose?

Pulmonary hypertension is a condition where abnormally high blood pressure in the lung arteries forces the heart to work harder, increasing the risk of heart failure. Early diagnosis is crucial, but the disease can be difficult to recognize, especially for non-specialist clinicians. One major challenge is that critical diagnostic information often gets buried in dense medical records or goes missing entirely.

Diagnosing pulmonary hypertension requires measuring mean pulmonary artery pressure (mPAP) during a procedure called right heart catheterization. An mPAP reading above 20 millimeters of mercury indicates a patient has the condition. However, reports from these procedures are often inconsistent. In some large registries, the key mPAP measurement was missing in as many as 33 percent of cases. With thousands of patients undergoing these procedures each year, manually reviewing every report to find hidden data is simply not practical.

How Does the AI System Work?

Researchers at the University of Maryland Institute for Health Computing (UM-IHC) and the University of Maryland School of Medicine (UMSOM) developed an AI system based on large language model technology, which can quickly recognize, organize, and translate vast amounts of information. The system performs a series of steps to identify and extract diagnostic data from medical records.

• Data Identification: The AI first determines whether a right heart catheterization report contains a pulmonary artery pressure measurement.
• Information Extraction: Once identified, the system extracts that measurement and checks the result against the original medical record to confirm accuracy.
• Missing Data Prediction: When the specific value needed for diagnosis is absent from reports, the AI combines machine learning with the language model to predict those missing measurements using a formula based on other available pressure readings.

The researchers trained and tested their tool using 17,292 right heart catheterization reports from 11,029 patients treated within the University of Maryland Medical System between 2016 and 2024. The AI tool accurately pulled key pressure measurements more than 99 percent of the time when compared with reviews performed by experienced pulmonary hypertension clinicians.

What Real-World Impact Could This Have?

When the researchers applied their missing-data prediction method to 507 patient records where the key measurement was absent, the AI-assisted approach identified 382 patients whose estimated pressure met the criteria for pulmonary hypertension. These are patients who might have gone undiagnosed without the tool's help.

"Finding these patients lets us connect them with specialized care and treatment, and in some cases, clinical trials. This means better management of the condition and potential lives saved," said Bradley Maron, UM-IHC co-executive director and the Melvin Sharoky, MD Professor of Medicine at UMSOM.

Bradley Maron, UM-IHC co-executive director and Professor of Medicine at UMSOM

The research was published in the European Respiratory Journal on May 28, 2026, and was conducted by researchers including Seyed M. Shams, Mary E. Maldarelli, Yijun Yin, Steven Cassady, Gautam Ramani, Colleen M. Ennett, Bradley A. Maron, and Katarina Zeder.

Is This Tool Meant to Replace Doctors?

The researchers emphasize that the tool is not a replacement for physicians, but rather a way to support clinical decision-making by helping doctors find important data that may otherwise be overlooked. The AI acts as a second set of eyes, sifting through thousands of patient records to flag cases that warrant specialist attention.

"This tool showcases the power of where clinical medicine and academic research intersect in the treatment of complex diseases. It's a great example of how an IHC partnership lets us find and help thousands of at-risk patients using the de-identified clinical data available from across UMMS," noted Ian Brooks, senior director for research informatics at UMMS and the UM-IHC.

Ian Brooks, Senior Director for Research Informatics at UMMS and UM-IHC

What's Next for This Technology?

The next step is to test the pulmonary hypertension tool prospectively within the electronic health record system to see how it performs in real-time clinical care. The researchers hope the technology will eventually help thousands of patients receive earlier diagnoses and specialist referrals, leading to speedier treatments and better long-term health outcomes.

Because electronic health records contain vast sums of data in disparate formats, AI tools like this one could also be applied to identifying missed diagnoses of other complex diseases beyond pulmonary hypertension. The approach demonstrates how machine learning and large language models can work together to solve real clinical challenges that affect patient outcomes.