Artificial Intelligence as a tool for diagnostic support in chronic venous disease

This retrospective study aimed to assess the diagnostic accuracy of an AI-powered clinical decision support platform in diagnosing chronic venous disease (CVD), using specialist diagnosis as the reference. A total of 130 patients who presented to a vascular surgery consultation in January 2024 with both venous and non-venous pathologies were included. Clinical data, including patient histories, physical examination findings, and complementary test results, were manually entered into the AI platform. The AI-generated diagnoses were then compared to those made by vascular specialists using a standardized classification system. The level of agreement between the AI and the specialists was evaluated using Fleiss' kappa coefficient. The AI platform demonstrated a diagnostic accuracy of 97% for CVD, with a very high concordance (kappa = 0.925, 95% CI 0.81-1.03, p < 0.001). While the platform demostrated excellent diagnostic accuracy for CVD, the limited representation of other pathologies in the sample limited the generalizability of the findings. The results suggest that while the AI platform shows promise as a tool to support clinical decision-making in CVD, further research is necessary to determine its effectiveness in broader diagnostic contexts. Accurate and detailed data input remains essential to maximize the platform's diagnostic capabilities.