AI System Detects H. pylori from Biopsy Reports

Researchers have developed the Nimblemind Multi-Agent System (nMAS) to automate the detection of Helicobacter pylori infection from gastric biopsy pathology reports. This system is designed to overcome the challenges of extracting information from heterogeneous coded and free-text fields, which often require complex contextual interpretation. In a pilot evaluation using 54 de-identified reports, nMAS demonstrated high accuracy, correctly classifying 98.61% of feature-case decisions across various fields related to H. pylori positivity and associated gastritis. While its predictive performance was comparable to other methods, nMAS's key contribution lies in its workflow integration and traceability, providing unified report-level outputs with supporting source sentences. This approach significantly reduces the time required for review, potentially saving substantial staff-hours and costs in healthcare settings.