Development and validation of prognostic models for bone metastasis in Non-Small cell lung cancer based on Machine...

Area of Science:

• Oncology
• Medical Informatics

Background:

• Bone metastasis is frequent in non-small cell lung cancer (NSCLC).
• No validated prognostic model exists for NSCLC patients with bone metastases at diagnosis.

Purpose of the Study:

• To develop and validate a prognostic model for overall survival (OS) in NSCLC patients with bone metastases at diagnosis.

Main Methods:

• Retrospective review of 1,299 NSCLC patients (2016-2023).
• Application of three machine learning algorithms to identify prognostic variables.
• Construction and internal validation of a Cox regression nomogram for OS prediction.

Main Results:

• Identified four independent prognostic factors: age, serum calcium, monocyte-to-albumin ratio, and prognostic nutritional index.
• Nomogram showed strong predictive performance (AUCs: 86.53% for 6-month, 78.32% for 1-year, 77.85% for 2-year OS).
• Calibration plots confirmed excellent agreement between predicted and observed survival.

Conclusions:

• A validated nomogram offers a practical tool for predicting survival in NSCLC patients with bone metastases.
• Supports individualized risk stratification and clinical decision-making for this patient group.