Briya AI Platform Uncovers Hidden Liver Disease Cases in Radiology Data

The research, which analyzed 28,795 abdominal ultrasound reports from 20,422 patients between 2020 and 2024, highlights a critical gap in current clinical data practices. While structured diagnostic codes only captured 22% of the affected population, the AIRE platform’s natural language processing successfully identified an additional 4,036 patients. This leap in detection allows for earlier clinical intervention before the condition progresses to irreversible stages.

Dr. Gadi Lalazar, head of the Liver Unit at Shaare Zedek Medical Center, noted that many patients were unaware of their condition because the findings were incidental to other medical procedures. By automating the extraction of these insights, the platform bypasses the need for manual review, which has historically hindered large-scale research. The study, led by Dr. Or Shaked, achieved 94% precision and 97% recall during validation. Beyond simple detection, the tool enabled researchers to calculate FIB-4 scores for 4,368 patients, effectively mapping those at the highest risk of disease progression. These findings will be presented at the AASLD's The Liver Meeting 2026 on November 6.