Supplementary file 1_An interpretable machine learning model for predicting symptomatic pe

A dataset from 221 patients with cervical cancer who underwent pelvic lymphadenectomy at the Affiliated Hospital of North Sichuan Medical College between January 2023 and September 2024. It contains clinical characteristics and laboratory data used to develop and validate machine learning models for predicting symptomatic pelvic lymphocele, which occurred in 44 patients (19.9%). The dataset was published by Yiyue Wang on figshare.

Use Cases

• Training a predictive model for symptomatic pelvic lymphocele based on identified risk factors like diabetes and surgical approach.
• Analyzing feature importance for surgical outcomes using SHAP values on variables such as preoperative monocyte-to-lymphocyte ratio.
• Benchmarking model performance (e.g., KNN, XGBoost) for clinical prediction tasks using metrics like AUC and Brier Score.
• Investigating associations between preoperative fibrinogen levels or tumor size and post-operative complications.

Strengths

• Includes data from 221 patients, providing a basis for model development.
• Specific model performance metrics are reported, such as a KNN model AUC of 0.952 in the training set.
• Key predictive features are identified through SHAP analysis, including diabetes and preoperative lab ratios.

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 hospital in China, which may limit generalizability to other populations.

Provenance

Source

Affiliated Hospital of North Sichuan Medical College.

Collection Method

Retrospective analysis of clinical data.

Time Range

January 2023 to September 2024.

Freshness

Last updated 2026-04-15 05:44:16; freshness should be verified.

Geography

China (hospital location unspecified but likely Sichuan).