Data Sheet 1_Machine learning-based risk prediction of overt hepatic encephalopathy after

A retrospective cohort of 297 patients with cirrhosis who underwent a transjugular intrahepatic portosystemic shunt (TIPS) procedure at the First Hospital of Shanxi Medical University from 2019 to 2024. The dataset, authored by Lixin Song, was used to develop and validate machine learning models for predicting postoperative overt hepatic encephalopathy (OHE).

Use Cases

• Developing risk prediction models for post-TIPS overt hepatic encephalopathy based on preoperative clinical characteristics.
• Comparing the performance of machine learning algorithms like XGBoost and logistic regression on clinical outcome data.
• Identifying key predictive factors such as BUN, GGT, and portal vein puncture site using feature selection methods.
• Interpreting model predictions for clinical decision support using techniques like Shapley additive explanations.

Strengths

• Includes data from 297 patients, with 89 labeled cases of the target outcome (postoperative OHE).
• Model development used five distinct algorithms, with the XGBoost model achieving an AUC of 0.792.
• Feature selection was performed using established methods (LASSO and random forest).

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 medical center, which may limit generalizability.

Provenance

Source

First Hospital of Shanxi Medical University

Collection Method

Retrospective cohort study of patient records.

Time Range

2019 to 2024

Freshness

Last updated 2026-05-29 04:48:25; freshness should be verified.

Geography

China (Shanxi)