Myopia Progression Prediction Model Using Ocular Biometric Parameters

An XGBoost model achieved an R² of 0.913 on training data for predicting myopia progression. Peng Zhou developed this machine learning model using retrospective patient data from Shanghai Parkway Health and Peking University People's Hospital. The dataset, last updated in April 2026, likely contains the model code or related documentation.

Use Cases

• Building predictive models for myopia progression risk based on ocular biometric parameters.
• Analyzing the importance of features like pupil size and corneal thickness for axial elongation.
• Comparing performance of different ML algorithms (e.g., XGBoost, random forest) on clinical prediction tasks.
• Developing quantitative tools for early clinical intervention in myopia management.

Strengths

• Model performance metrics are provided, including R² values of 0.913 (training) and 0.766 (test).
• Key predictive factors are identified with specific importance scores (pupil size score 100, corneal thickness 40.88).
• The dataset is openly licensed under CC-BY-4.0.

Limitations

• The dataset is only 6.6 KB, suggesting limited scope, likely containing code or a small document rather than raw patient data.
• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.

Provenance

Source

Peng Zhou

Collection Method

Retrospective analysis of patient data from Shanghai Parkway Health Ophthalmology Department and Peking University People's Hospital Optometry Center.

Freshness

Last updated 2026-04-24 17:43:43; freshness should be verified.

Geography

China (Shanghai and Beijing)