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

Bone is a common site of metastasis in non-small cell lung cancer (NSCLC), yet no validated prognostic model is currently available for patients presenting with bone metastases at diagnosis.

We retrospectively reviewed 1,299 NSCLC patients who underwent high-throughput sequencing between 2016 and 2023. Of these, 195 were diagnosed with bone metastases at presentation. Three machine learning algorithms were applied to identify prognostic variables. A nomogram constructed with Cox regression was used to predict overall survival (OS) and was internally validated with 1,000 bootstrap resamples.

Four independent prognostic factors were identified, including age, serum calcium, monocyte-to-albumin ratio, and prognostic nutritional index. The nomogram demonstrated strong predictive performance, with areas under the curve (AUCs) of 86.53%, 78.32%, and 77.85% for 6-month, 1-year, and 2-year OS, respectively. Calibration plots showed excellent agreement between predicted and observed survival outcomes.

This validated nomogram provides a practical and individualized tool for predicting survival in NSCLC patients with bone metastases at diagnosis, supporting risk stratification and clinical practice.