Chronic Otitis Media Postoperative Complication Risk Prediction Model

Jinlei Wang developed and validated a machine learning prediction model for postoperative complications in patients with Chronic Otitis Media. The model was built using 21 Clinical Multi-omics indicators from 338 patients split into training (n=237) and validation (n=101) sets. The Random Forest model achieved an AUC of 0.885 in training and 0.853 in validation, with key predictors including history of diabetes, previous ear surgery, and cholesteatoma presence.

Use Cases

• Predicting individual patient risk for postoperative complications based on demographic, clinical, and laboratory indicators.
• Comparing performance of Random Forest, Logistic Regression, K-Nearest Neighbors, and Gradient Boosting Machine models for clinical prediction tasks.
• Interpreting model decisions for key risk factors like previous ear surgery and cholesteatoma presence using SHAP analysis.
• Assessing clinical net benefit of a predictive model across different risk probability thresholds via decision curve analysis.

Strengths

• Model performance is quantitatively validated with an AUC of 0.853 on a separate validation set of 101 patients.
• The study identifies seven independent risk factors through multivariate analysis, providing clinical interpretability.
• Model calibration and clinical utility are assessed using calibration curves and decision curve analysis.

Limitations

• Row count and column-level documentation are absent; field semantics must be inferred after download.
• The dataset is very small at 16.8 KB, indicating limited scope and likely containing only model results or summary statistics.
• The data may reflect bias inherent to the specific patient cohort used in the single study.

Provenance

Source

Jinlei Wang via figshare

Collection Method

Clinical data collection including demographic, clinical, laboratory, and imaging indicators from patients with Chronic Otitis Media.

Time Range

null

Freshness

Last updated 2026-04-13 04:34:51; freshness should be verified.

Geography

null