The Application of Artificial Neural Networks and Logistic Regression in the Evaluation of Risk for Dry Eye after Vitrectomy

Abstract

Supervised machine-learning (ML) models were employed to predict the occurrence of dry eye disease (DED) after vitrectomy in this study. e clinical data of 217 patients receiving vitrectomy from April 2017 to July 2018 were used as training dataset; the clinical data of 33 patients receiving vitrectomy from August 2018 to September 2018 were collected as validating dataset. e input features for ML training were selected based on the Delphi method and univariate logistic regression (LR). LR and artificial neural network (ANN) models were trained and subsequently used to predict the occurrence of DED in patients who underwent vitrectomy for the first time during the period. e area under the receiver operating characteristic curve (AUC-ROC) was used to evaluate the predictive accuracy of the ML models. e AUCs with use of the LR and ANN models were 0.741 and 0.786, respectively, suggesting satisfactory performance in predicting the occurrence of DED. When the two models were compared in terms of predictive power, the fitting effect of the ANN model was slightly superior to that of the LR model. In conclusion, both LR and ANN models may be used to accurately predict the occurrence of DED after vitrectomy.

AI-Based Advanced Approaches and Dry Eye Disease Detection Based on Multi-Source Evidence: Cases, Applications, Issues, and Future Directions

“…The results demonstrate the discriminatory capacity of tear osmolarity in DED with the logistic classifier algorithm achieving a high accuracy of 85%. The research concludes that ML is an efficient approach for DED detection using tear osmolarity variables [58] . Furthermore, Yang et al [59] conducted research on the application of ML models for predicting DED in post-vitrectomy patients.…”

“…The researchers concluded that ML is an effective and resource-saving technique for DED detection that mitigates diagnostic biases. In another study, Cartes et al [58] employed ML algorithms, such as logistic classifier, naive Bayes, support vector machines, and random forests for classifying DED based on mean osmolarity records of 40 subjects (20 with DED and 20 normal). The results demonstrate the discriminatory capacity of tear osmolarity in DED with the logistic classifier algorithm achieving a high accuracy of 85%.…”

“…indicator for the detection of DED syndrome severity according to certain researchers [58,112] . Cartes et al [58] and Lemp et al [112] proposed an osmolarity cutoff of 311 mOsms/L, while Cartes et al [58] suggested 308 mOsm/L.…”

“…indicator for the detection of DED syndrome severity according to certain researchers [58,112] . Cartes et al [58] and Lemp et al [112] proposed an osmolarity cutoff of 311 mOsms/L, while Cartes et al [58] suggested 308 mOsm/L. Jiang et al [38] conducted an analysis of ultrasonic images and found a strong relationship between osmolarity and DED.…”

Developing an artificial neural network for detecting COVID-19 disease

“…The ANN model has robust error tolerance; thus, it can be extensively used in the fields of prediction and analysis. [ 72 ] Furthermore, leveraging the potential of an ANN-based CDSS would assist health-care providers to make better decisions concerning COVID-19 (diagnosis, classification, etc.). Despite standard statistical approaches (e.g., logistic regression) that need further modeling processes, ANNs do not necessitate distributional assumptions.…”