This study aimed to evaluate the impact of AI-assisted non-contrast computed tomography (NCCT) interpretation on time to intervention (TTI) and diagnostic accuracy for stroke management in a rural hospital in India, where non-specialist physicians provide care. A total of 174 stroke patients (72 before AI deployment and 102 after) were analyzed for TTI, while 864 NCCT scans were assessed for diagnostic accuracy using teleradiologists’ reports as ground truth. AI implementation significantly reduced median TTI from 80 minutes (IQR: 56.8–139.5) to 58.5 minutes (IQR: 30.3–118.2) (Wilcoxon rank sum test—location shift: -21.0, 95% CI: -38.0, -7.0). AI demonstrated high accuracy in detecting intracranial hemorrhage (sensitivity: 0.89, specificity: 0.99, PPV: 0.96, NPV: 0.97) and infarcts (sensitivity: 0.84, specificity: 0.81, PPV: 0.58, NPV: 0.94). The findings suggest that AI-enabled NCCT interpretation can accelerate stroke diagnosis, leading to faster administration of lifesaving interventions in resource-limited settings.
Published 24 Jul 2024
Artificial Intelligence-based automated CT brain interpretation to accelerate treatment for acute stroke in rural India: An interrupted time series study
Author: Justy Antony Chiramal ,
Jacob Johnson,
Jemin Webster,
D. Rachel Nag,
Dennis Robert,
Tamaghna Ghosh,
Satish Golla,
Saniya Pawar,
Pranav Krishnan,
Paul K. Drain,
Stephen J. Mooney
