Multicentre prospective cohort study to develop and validate a machine learning-based model for predicting 6-month all-cause mortality in elderly patients with advanced chronic obstructive pulmonary disease in China: study protocol

Clinical Snapshot

Key Findings

• P Value: Not yet available - protocol paper

• Effect Size: Not yet available - protocol paper

• Primary Outcome: 6-month all-cause mortality

• Nnt Or Sensitivity: Model discrimination and calibration metrics to be reported

• Confidence Interval: Not yet available - protocol paper

Clinical Application

Requires implementation of comprehensive geriatric assessment and standardised data collection protocols Limited direct applicability due to population-specific development in Chinese cohort; would require validation in Australian elderly COPD populations before clinical implementation Elderly patients with advanced COPD in Chinese healthcare settings

Abstract

INTRODUCTION: Chronic obstructive pulmonary disease (COPD) has an unpredictable clinical course, causing difficulties in short-term mortality prediction, overtreatment and delayed palliative care. Existing prediction models are limited and lack applicability to Chinese elderly patients with advanced COPD. Given the heavy disease burden and limited palliative care in China, we designed this multicentre cohort study to develop a 6-month mortality prediction model for elderly patients with advanced COPD to aid risk stratification, timely palliative care and efficient healthcare resource allocation. METHODS AND ANALYSIS: Patient recruitment has been ongoing since May 2024 and will be completed by December 2026, with a 12-month follow-up to be completed by December 2027. Eligible patients are being enrolled, and multidimensional baseline data including demographic characteristics, clinical indicators, laboratory results, comprehensive geriatric assessment and COPD-specific prognostic factors are being systematically collected. All participants will receive 12 months of standardised follow-up (monthly for the first 6 months and quarterly thereafter) to monitor 6-month all-cause mortality (primary outcome), as well as survival duration, end-of-life healthcare utilisation and do-not-resuscitate status (secondary outcomes). After completion of data collection, we will employ multiple machine learning algorithms to develop and internally validate a 6-month mortality prediction model with pre-specified centres reserved for external validation. Model performance will be evaluated by discrimination and calibration and head-to-head comparisons with the Body Mass Index, Airflow Obstruction, Dyspnoea and Exercise Capacity (BODE) and Age, Dyspnoea and Airflow Obstruction (ADO) indices will be conducted to verify its clinical value. The findings will provide a China-specific prediction tool for elderly patients with advanced COPD to guide clinical intervention, palliative care referral and healthcare resource allocation. ETHICS AND DISSEMINATION: This study was approved by the Biomedical Ethics Review Committee of West China Hospital, Sichuan University (No. 2024-2662) and registered at ChiCTR2500100351. Informed consent is being obtained from all participants. Results will be published in peer-reviewed journals and presented at academic conferences. TRIAL REGISTRATION NUMBER: ChiCTR2500100351.

References

1. 1.Pu, H., Liu, L., Chang, Y., Su, L., Zeng, X., Cheng, W., Jiang, Y., He, J., & Mo, L. (2026). Multicentre prospective cohort study to develop and validate a machine learning-based model for predicting 6-month all-cause mortality in elderly patients with advanced chronic obstructive pulmonary disease in China: study protocol. BMJ Open. https://doi.org/10.1136/bmjopen-2025-110648