Validation of AI-based software for objectification of conjunctival provocation test

BackgroundProvocation tests are widely used in allergology to objectively reveal patients’ sensitivity specific allergens. The objective quantification of an allergic reaction is a crucial characteristic these tests. Because the absence quantitative measurements, conjunctival provocation test (CPT) less frequently method despite its and simplicity.ObjectiveWe developed new artificial intelligence (AI)-based method, called AllergoEye, for evaluation reactions validated it clinical study.MethodsAllergoEye was implemented as 2-component system. first component based on Android smartphone camera screening imaging patient’s eye, second personal computer–based image analysis quantification. For validation open-label, prospective, monocentric study carried out 41 patients. Standardized CPT performed with sequential titration grass allergens 4 dilutions, evaluated by subjective/qualitative symptom scores AllergoEye scores.ResultsAllergoEye demonstrated high (98%) specificity (90%) compared human estimation reaction. Tuning cutoff thresholds allowed us increase 97%, at which point correlation between detected allergen IgE carrier–polymer system class becomes obvious. Strikingly, such not found basis subjective qualitative scores.ConclusionThe that sensitive efficient instrument measurement studies well routine therapy control. Provocation simplicity. We study. scores.