Back to 2026 Abstracts
Impact Of Ai Enabled Decision Support On Lung Nodule Clinic Performance Across Three Health Systems
K. Adam Lee1, Amit Borah2, garrett Fiscus3, Lisa Petricca3, sarah dean2, Annika Rings4, Cantarina Santos5, Lyndsey Pickupp5, Omar Ibrahm3.
1Jupiter Medical Center, Jupiter, FL, USA, 2Atlanticare Regional Medical Center, Atlantic City, NJ, USA, 3University of Connecticut School of Medicine, Farmington, CT, USA, 4Optellum, Oxford, United Kingdom, 5Optellum, oxford, United Kingdom.
BACKGROUND:Health systems and payers increasingly require robust evidence that artificial intelligence (AI) tools meaningfully improve care delivery at scale. In lung nodule management, AI-enabled clinical decision support may streamline workflows, enhance patient identification and triage, and facilitate earlier lung cancer diagnosis. Yet real-world data demonstrating clinical utility remain limited. This study assessed the impact of deploying a commercially available AI platform across multidisciplinary lung nodule clinics in three diverse U.S. health systems.
METHODS:We conducted a multi-site before-and-after analysis using matched 12.5-month baseline and post-implementation periods. Primary outcomes included new patient volume and procedure counts (percutaneous or bronchoscopic biopsy and surgical resection). Secondary outcomes were total lung cancers diagnosed, early-stage detections (stage 0-II), and time from baseline CT to histologic diagnosis. Statistical comparisons utilized Wilcoxon signed-rank and Mann-Whitney U tests, with results summarized as medians and interquartile ranges (IQRs).
RESULTS:New patient volume rose from 138 to 232 (11.0 to 18.6 per month; +68%; p = 0.0002). Median monthly patients increased from 9 (IQR 7-14) to 16 (IQR 13-21). Monthly procedures increased from 5.0 to 7.8 (p = 0.0093). Monthly lung cancer diagnoses increased from 3.0 to 3.8 (p = 0.3242), and early-stage cases from 1.4 to 2.4 (p = 0.0996). Median time from baseline CT to diagnosis remained stable (58 days [IQR 33-153] pre-implementation vs. 57 days [IQR 27-85] post-implementation; p = 0.28), with a substantial narrowing of the IQR from 120 to 58 days.
CONCLUSIONS:Across three health systems, integrating an AI-enabled decision support platform into lung nodule clinics was associated with marked increases in patient throughput and procedural activity, while preserving consistent diagnostic timelines. These findings support the potential for AI-driven case identification and triage to expand timely access to specialty evaluation without compromising diagnostic efficiency.
Back to 2026 Abstracts
