Chronic Pancreatitis, Machine Learning, Pancreatic Neoplasm
Conditions
Keywords
Pancreatic neoplasm, chronic pancreatitis, machine learning, SHAP
Brief summary
This study aims to develop XGBoost machine learning model to predict pancreatic neoplasms in CP patients with focal pancreatic lesions.
Detailed description
Pancreatic neoplasms include various types, with pancreatic cancer being the most common and having a poor prognosis. Chronic pancreatitis (CP) can progress to pancreatic cancer, and detecting neoplasms in CP patients is challenging due to similar imaging and clinical presentations. Current diagnostic methods like CT and tumor markers have limitations, and endoscopic ultrasound-guided tissue acquisition has moderate sensitivity. Machine learning (ML) shows promise in medical fields, but its black box nature limits its application. SHapley additive exPlanations (SHAP) can provide intuitive explanations for ML models. This study aims to develop an ML model to predict pancreatic neoplasms in CP patients with focal pancreatic lesions and use SHAP to explain the model, aiding future research.
Interventions
DIAGNOSTIC_TESTXGBoost machine learning
XGBoost is a powerful machine learning algorithm known for its efficiency and performance. It is an optimized gradient boosting library designed to be highly efficient, flexible, and portable. XGBoost works by combining multiple weak prediction models, typically decision trees, to produce a strong predictive model. It supports various objective functions and evaluation metrics, making it suitable for a wide range of tasks, including classification and regression. XGBoost also includes features like regularization to prevent overfitting and can handle missing data effectively.
Sponsors
Changhai Hospital
Study design
Observational model
COHORT
Time perspective
RETROSPECTIVE
Eligibility
Sex/Gender
ALL
Healthy volunteers
No
Inclusion criteria
* Diagnosis of chronic pancreatitis * Patients has indeterminate focal pancreatic lesions discovered through contrast-enhanced CT scans
Exclusion criteria
* Patients had incomplete clinical data * Patients had no surgical pathology results for the focal pancreatic lesions and loss to follow-up, indicating that a final diagnosis of the focal pancreatic lesions could not been established
Design outcomes
Primary
| Measure | Time frame | Description |
|---|---|---|
| Diagnostic yield | 10 years | The diagnostic yield of XGBoost machine learning, including AUC、Sensitivity、Specificity |
Countries
China
Outcome results
None listed