Abstract Body: Introduction Pupillary reactivity serves as an essential metric for rapid triaging and assessment of potential neurological decline in stroke patients. Traditional methods, which employ a medical penlight, depend on subjective evaluations of pupillary reactivity.
Materials and Methods In the case study involving five subjects, a pharmacological agent (tropicamide) was administered to induce a state of pupillary unresponsiveness controllably simulating a condition potentially encountered in an Intensive Care Unit (ICU) setting. This methodology enabled the real-time monitoring of changes in pupillary reactivity over a short duration (~6 hours), governed by the pharmacokinetics of the administered drug. The software application (AI Pupillometry System, Solvemed Inc.), utilizing artificial intelligence (AI) models analyzing videos, yielded ten distinct physiological parameters, including pupil diameter and constriction velocity among others. These parameters were subsequently employed to construct a computational score, the Pupil Reactivity (PuRe) score, aimed at the quantifiable assessment of pupillary reactivity. This score was validated against a dataset comprising measurements obtained from a bioethically approved study acquired without tropicamide administration.
Results The validation process demonstrated differentiation between non-reactive and reactive pupils. Specifically, the peak effect of tropicamide was compared for N=5 subjects against measurements under normal conditions for N=1,800 measurements, achieving 100% discrimination accuracy. Furthermore, the performance of the software application was benchmarked against other commonly used diagnostic devices yielding comparable or superior reliability of physiological readouts.
Conclusion Considering the brief duration of each measurement (approximately 5 seconds) and the portability of the mobile solution relative to a traditional medical penlight, it is reasonable to suggest that the smartphone-based software could serve as a reliable data source for the Neurological Observation Chart. This, in turn, has the potential to enhance the monitoring of neurological health in ICU patients.