Postoperative Feeding Intolerance Risk in Oral Cancer Patients

752 patients undergoing oral cancer free flap reconstruction between 2017 and 2025 were analyzed to predict enteral feeding intolerance. The dataset includes 35 perioperative variables, with a random forest model achieving an AUC of 0.889 for predicting the 36.04% incidence of FI. This interpretable model was developed by Baolin Jia and shared under a CC-BY-4.0 license.

Use Cases

• Predicting feeding intolerance risk based on perioperative metabolic and inflammatory markers.
• Analyzing the nonlinear threshold effects of variables like fasting glucose and operative time.
• Developing interpretable clinical decision support tools using SHAP analysis.
• Comparing the performance of seven machine learning algorithms for a specific surgical outcome.

Strengths

• Includes data from 752 patients, a substantial cohort for a single-center study.
• Model performance was rigorously validated, achieving an AUC of 0.889 with good calibration.
• The study evaluated 35 perioperative variables and identified nine key predictors using LASSO regression.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• Data is from a single-center retrospective study, which may limit generalizability.

Provenance

Source

figshare

Collection Method

Single-center retrospective study.

Time Range

2017 to 2025

Freshness

Last updated 2026-06-04 04:27:20; freshness should be verified.