Validation of a Universal Cataract Intelligence Platform

This study established and validated a universal artificial intelligence (AI) platform for collaborative management of cataracts involving multi-level clinical scenarios and explored an AI-based medical referral pattern to improve collaborative efficiency and resource coverage.The datasets were labeled using a three-step strategy: (1) categorize slit lamp photographs into four separate capture modes; (2) diagnose each photograph as a normal lens, cataract or a postoperative eye; and (3) based on etiology and severity, further classify each diagnosed photograph for a management strategy of referral or follow-up. A deep residual convolutional neural network (CS-ResCNN) was used for the image classification task. Moreover, we integrated the cataract AI agent with a real-world multi-level referral pattern involving self-monitoring at home, primary healthcare, and specialized hospital services.

A universal diagnostic system. An artificial intelligence to make comprehensive evaluation and treatment decision of cataract.

An artificial intelligence to make comprehensive evaluation and treatment decision of different types of cataracts.

AUC: area under the receiver operating curve; accuracy (ACC) = (TP + TN) / (TP + TN + FP + FN); sensitivity (SEN) = TP / (TP + FN); specificity (SPE) = TN / (TN + FP); TP = true positive; TN = true negative; FP = false positive; FN = false negative.

Inclusion Criteria: Patients who underwent ophthalmic examination of the eye and recorded their ocular information in the primary healthcare center. Exclusion Criteria: The patients who cannot cooperate with the examinations.