A Web-Based Calculator to Predict Early Death Among Patients With Bone Metastasis Using Machine Learning Techniques:...

Area of Science:

• Oncology
• Medical Informatics
• Machine Learning

Background:

• Bone metastasis significantly limits patient survival, complicating treatment decisions.
• Accurate survival prediction is crucial for guiding clinical management in these patients.

Purpose of the Study:

• To develop a machine learning-based web calculator for predicting early death in bone metastasis patients.
• To provide an accurate assessment of mortality risk to aid clinical decision-making.

Main Methods:

• Analysis of a large cohort (118,227 patients) with bone metastasis from a national cancer database (2010-2019).
• Implementation and evaluation of six machine learning models: logistic regression, extreme gradient boosting machine, decision tree, random forest, neural network, and gradient boosting machine.
• External validation using a separate cohort (332 patients) to confirm model robustness.

Main Results:

• The gradient boosting machine model demonstrated superior prediction performance (54 points) and discrimination (AUC 0.858).
• The model effectively stratified patients into high-risk (71.96% early death) and low-risk (15.62% early death) groups.
• External validation confirmed the model's robust performance (AUC 0.847).

Conclusions:

• A machine learning-based calculator was successfully developed to predict early death in patients with bone metastasis.
• The calculator can significantly aid clinical decision-making by identifying high-risk patients, potentially improving care and outcomes.