NCT05718999

XGBoost for Predict Incisional Hernia

Development and Internal Validation of a Machine Learning Model to Predict the Occurrence of Incisional Hernia After a Midline Laparotomy

• Status: Unknown
• Phase: —
• Study type: Observational
• Enrollment: 1,000 (estimated)
• Sponsor: Hospital Regional de Alta Especialidad del Bajio · Academic / Other
• Sex: All
• Age: 18 Years – 80 Years
• Healthy volunteers: —

Summary

The objective of this study is to develop a predictive model of IH based on machine learning with the use of the XGBoost technique, this will help surgeons in charge of abdominal wall closure to have objective support to determine high-risk patients and in them modify the closure technique or use a mesh according to their choice or the degree of contamination of the abdominal cavity.

Detailed description

Retrospective and observational study. The predictions will make using machine learning models. The programs use the scikit-learn, xgboost and catboost Python packages for modeling. The evaluation of models will be using fourfold cross-validation, the receiver operating characteristic (ROC) curve, the area under the ROC curve (AUC), and accuracy metrics calculated on the union of the test sets of the cross-validation. The most critical factors and their contribution to the prediction will identify using a modern tool of explainable artificial intelligence called SHapley Additive exPlanations (SHAP).

Conditions

• Incisional Hernia

Interventions

Type	Name	Description
DIAGNOSTIC_TEST	Not intervention	Not having intervention is an observational study

Timeline

• Start date: 2023-01-30
• Primary completion: 2023-02-22
• Completion: 2023-11-30
• First posted: 2023-02-08
• Last updated: 2023-02-27

Locations

1 site across 1 country: Mexico