Using an artificial intelligence (AI)-based platform in a large real-world analysis of ultrasound radiology reports led to a more than three-fold increase in identified metabolic dysfunction-associated steatotic liver disease (MASLD) patients – signalling its potential to facilitate earlier intervention before the disease becomes irreversible.
Health-tech company Briya used its AIRE AI and natural language processing (NLP) platform to analyse 28,795 abdominal ultrasound reports, comprising 20,422 unique patients, between 2020 and 2024.
Go deeper with GlobalData
Discover B2B Marketing That Performs
Combine business intelligence and editorial excellence to reach engaged professionals across 36 leading media platforms.
The analysis identified an additional 4,036 patients with a steatotic liver, expanding the relevant study population by 360%, from 1,122 to 5,158 patients.
“Briya AIRE fundamentally changed how we were able to identify patients for this study,” said Gadi Lalazar, study lead and head of the liver unit at Israel’s Shaare Zedek Medical Center.
“Many of these patients may not have known they had steatotic liver disease because the finding was buried in an ultrasound report performed for some other reason. Identifying them earlier creates an opportunity for physicians to initiate treatment before the disease progresses to more serious and potentially irreversible stages,” Zedek continued.
Meanwhile, Briya’s analysis also revealed “a major gap” when looking at structured records or diagnostic codes alone. Only 22% of additional steatotic liver patients identified in the analysis had a diagnosis documented within a structured electronic healthcare record (EHR), as per the company. Therefore, 78% of steatotic liver patients were identified exclusively through Briya AIRE’s NLP analysis of unstructured ultrasound report data, the company stated.
Unstructured reports refer to traditional medical imaging summaries, typically dictated by radiologists in a ‘stream-of-consciousness’ format, as opposed to structured reports that are filed in a more rigid summary template.
With a more complete study population, researchers then assessed the risk of disease progression using existing laboratory data and ultrasound reports. The analysis identified 4,368 patients with a Fibrosis-4 Index (FIB-4 score; a measure used to estimate the amount of scarring in the liver) that placed them above the conventional at-risk threshold.
Briya, which has offices in both New York and Tel Aviv, highlighted that these findings demonstrate that unlocking information from unstructured radiology reports gives researchers access to “larger, more representative patient populations” without requiring additional data collection. The company plans to share its findings at an upcoming medical meeting.
Dr Or Shaked, director of medical research solutions at Briya, commented: “AIRE’s specialised research agents were used to extract information from unstructured fields, harmonise it with existing structured data and build the study population. The goal is to take on more of the complex data preparation work so researchers can use more of the available information and focus on answering scientific questions that can ultimately influence care.”
AI is having a profound impact in the medical imaging space. According to GlobalData analysis, the AI market across healthcare is growing at a compound annual growth rate (CAGR) around 37% and is expected to reach a $57.4bn valuation in 2029.
