Artificial intelligence-assisted machine learning models for predicting lung cancer survival

Abstract

Objective

This study aimed to evaluate the feasibility of large language model-Advanced Data Analysis (ADA) in developing and implementing machine learning models to predict survival outcomes for lung cancer patients, with a focus on its implications for nursing practice.

Methods

A retrospective study design was employed using a dataset of lung cancer patients. Data included sociodemographic, clinical, treatment-specific, and comorbidity variables. Large language model-ADA was used to build and evaluate three machine learning models. Model performance was validated, and results were presented using calibration plots.

Results

Of 737 patients, the survival rate of this cohort was 73.3%, with a mean age of 59.32 years. Calibration plots indicated robust model reliability across all models. The Random Forest model demonstrated the highest predictive accuracy among the models. Most critical features identified were preoperative white blood cells (2.2%), preoperative lung function of Forced Expiratory Volume in one second (2.1%), preoperative arterial oxygen saturation (1.9%), preoperative partial pressure of oxygen (1.7%), preoperative albumin (1.6%), preoperative preparation time (1.5%), age at admission (1.5%), preoperative partial pressure of carbon dioxide (1.5%), preoperative hospital stay days (1.5%), and postoperative total days of thoracic tube drainage (1.4%).

Conclusions

Large language model-ADA effectively facilitates the development of machine learning models for lung cancer survival prediction, enabling non-technical health care professionals to harness the power of advanced analytics. The findings underscore the importance of preoperative factors in predicting outcomes, while also highlighting the need for external validation across diverse settings.