External Validation of a Machine-Learning derived risk prediction model for Post Operative Pulmonary Complications

Postoperative pulmonary complications (POPC) are a significant cause of morbidity and mortality following surgery, contributing to prolonged hospital stays and increased healthcare costs.

Machine learning (ML) algorithms have shown promise in predicting POPC risk based on preoperative variables.

However, external validation of these models is essential to evaluate their generalizability and clinical utility.

A recent review found that only a few of the developed scores have been externally validated viz. ARISCAT Score

 Recently, such a model was developed in the department.

The machine learning model performed reasonably well in the internal validation cohort, but the investigators were unable to validate in a large external cohort at that time.

This study will aim to externally validate the developed model, thereby increasing the generalizability of the model and pave the way for further multicentric validations and development of a tool for the routine clinical use adapted to the Indian population.