Introduction: Rapid management of critically ill patients on vasopressors demands immediate, precise data, yet ICU workflows face latency and fidelity issues from manual entry or complex middleware. This study evaluated the longitudinal impact of a novel Optical Character Recognition (OCR) tool for direct bedside vital sign acquisition on pressor initiation times and patient mortality. Methods: We conducted a multi-centre longitudinal study across 85 ICUs in 18 Indian states, analyzing 21,123 patient admissions (3,051,032 patient hours on pressors) from December 2019 to June 2025 with a median age of 55 (IQR 41, 66), 57.51% male, and 58.27% with sepsis. from June 2023, data were captured directly from monitors via a proprietary OCR model, a multi-stage neural network (object detection, segmentation, classifiers). Designed for real-time efficiency and bypassing middleware, its human-in-the-loop validated accuracy reached 98%. Outcomes were time to pressor initiation (hours from first vital to first pressor) and overall hospital mortality. Longitudinal trends were analyzed via linear and multivariate regression; segmented regression assessed changes post-OCR deployment (June 2023). Results: Analysis showed significant shortening of pressor initiation time over the study period. Median time decreased by 0.15 hours/month (p< 0.001), a net 10.19 hour reduction (from ~11.67 hours in Dec 2019 to ~1.48 hours in June 2025). This improvement coincided with the OCR tool’s progressive integration and accuracy. Hospital mortality among patients on pressors also significantly declined. The rate decreased by 0.34 percentage points/month (p< 0.001), a net 22.57 percentage point reduction (from ~44.97% to ~22.39%), irrespective of age, length of stay, APACHE II score or city tier. This reduction occurred alongside advancements in critical care, including OCR. Conclusions: Our study demonstrates the OCR tool’s integration was associated with significantly reduced time to pressor onset. This technology, alongside broader critical care improvements, coincided with a substantial decline in mortality among pressor patients. These findings underscore the critical role of streamlined, real-time data acquisition in enhancing timely interventions and improving patient outcomes in high-volume settings.
Unnikrishnan et al. (Sun,) studied this question.