Cardiometabolic Pattern of Workers Unfit for High-Altitude Occupational Exposure in Chile: A Large-Scale Retrospective Analysis Using a Random Forest Model ‡

Workers exposed to chronic intermittent hypobaric hypoxia at high altitude (HA) face substantial cardiometabolic demands that may compromise occupational safety. Fitness-for-work (FFW) assessments are routinely performed in Chile under nationally standardized guidelines, yet the internal structure of the unfit population and codependence among predictive biomarkers remain poorly characterized. This retrospective study analyzed 48,783 workers undergoing mandatory pre-employment FFW evaluations for HA exposure in Chile (2021–2024), of whom 8% were classified as unfit for work (UFW). A supervised Random Forest model, applied to more than 20 clinical, physiological, and laboratory variables, identified key predictors of FFW classification by capturing nonlinear interactions among variables that univariate approaches cannot detect. The model achieved precision 0.90 (95% CI: 0.89–0.92), sensitivity 0.93, specificity 0.88, and kappa 0.807. Body mass index was the most influential predictor, followed by fasting plasma glucose, triglycerides, and systolic blood pressure. Among unfit workers, severe obesity (BMI ≥ 35 kg/m2) was the most prevalent condition (34%), followed by altered fasting glucose (16%), elevated systolic blood pressure (10%), and severe hypertriglyceridemia (9%). The unfit population was predominantly male with age-related distribution. These findings reveal a cardiometabolic pattern as the dominant driver of HA occupational unfitness. Targeted strategies addressing obesity, glycemia, and dyslipidemia, alongside risk-based monitoring aligned with Chile’s national standards, may reduce exclusion rates and support more individualized FFW evaluations in HA settings. Aim: to characterize dominant cardiometabolic patterns in workers classified as unfit for HA occupational exposure in Chile and to quantify their relative weight using a Random Forest variable-importance framework. Methods: retrospective analysis of 48,783 pre-employment FFW evaluations (2021–2024) from a national occupational-health surveillance system; supervised Random Forest model with stratified 70/30 train-test split and repeated 10-fold cross-validation. Conclusion: targeted preventive strategies focused on obesity, glycemia, and dyslipidemia, combined with structured re-evaluation pathways, could reduce HA occupational exclusion and inform the future refinement of FFW criteria.