Related Experiment Video
Updated: Apr 28, 2026
Comparison of Agreement and Accuracy using Binocular Wavefront Optometer with Autorefractor and Phoropter
Published on: September 16, 2025
A TabPFN-based prediction system for refractive error and dry eye comorbidity: a retrospective study using
Danyi Qin1, Wenying Guan1, Shinan Wu1
1Xiamen University affiliated Xiamen Eye Center, Fujian Provincial Key Laboratory of Ophthalmology and Visual Science, Fujian Engineering and Research Center of Eye Regenerative Medicine, School of Medicine, Eye Institute of Xiamen University, Xiamen University, Xiamen, Fujian, China.
Machine learning accurately predicts dry eye disease risk in refractive error patients. This tool aids early detection and personalized management, improving patient outcomes and reducing healthcare burdens.
Area of Science:
- Ophthalmology
- Artificial Intelligence
- Data Science
Background:
- Refractive error and dry eye disease are prevalent conditions impacting quality of life.
- Evidence suggests a correlation between refractive error and dry eye disease.
- Early prediction of dry eye comorbidities in refractive error patients is crucial.
Purpose of the Study:
- To develop and validate a machine learning (ML) model for predicting concurrent dry eye comorbidities in patients with refractive error.
- To identify key risk factors associated with dry eye disease in this population.
- To create an accessible tool for clinical decision support.
Main Methods:
- Analysis of a large-scale outpatient database (n = 114,579) from Xiamen Eye Center.
- Utilized Random Forest for feature selection and evaluated eight ML algorithms.
- Selected the Tabular Prior-Data Fitted Network (TabPFN) model based on ROC, PR, and decision curve analysis.
Main Results:
- The TabPFN model achieved high screening efficacy with specificity and accuracy of 0.945.
- Identified longer duration of refractive error as a risk factor, especially in older females.
- Developed an online web calculator for the predictive model.
Conclusions:
- A high-performance, interpretable ML system was developed for early dry eye risk prediction in refractive error patients.
- The system offers significant potential as a clinical decision aid for timely, personalized management.
- This predictive tool holds substantial clinical value and promising application prospects.
