CatBoost Model 1H: A Predictive Model for 1-Year Mortality in Ontario, Canada

A CatBoost model trained on data from 12,080,801 Ontarians to predict 1-year mortality with an AUROC of 0.933. The model, created by Steven Habbous and shared on figshare in April 2026, uses administrative health data and outperformed logistic regression and other ensemble methods. It incorporates features like age, sex, comorbidities, and healthcare utilization measures.

Use Cases

• Benchmarking mortality prediction models based on the reported AUROC and PR-AUC performance metrics.
• Studying feature importance for mortality risk based on the described use of age, long-term care residence, and palliative care.
• Improving calibration in high-risk populations based on the sensitivity analysis results mentioned in the description.
• Comparing tree-based ensemble methods like CatBoost against logistic regression for epidemiological risk adjustment.

Strengths

• Model performance is quantitatively validated with multiple metrics, including an AUROC of 0.933 and an ICI of 0.0003.
• Training data covers a large, population-scale cohort of over 12 million individuals from Ontario.
• The model comparison is comprehensive, including logistic regression and seven tree-based ensemble methods.
• Feature importance is assessed using multiple techniques, including permutation importance and marginal effects.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count for the underlying dataset is unknown, which may limit suitability assessment.
• The data is specific to Ontario, Canada, which may limit generalizability to other populations.

Provenance

Source

Administrative healthcare databases for Ontario, Canada.

Collection Method

Data captured from administrative databases using a window of up to 3 years prior to January 1, 2022.

Time Range

Data likely spans up to 3 years prior to 2022, with mortality follow-up through 2023.

Freshness

Last updated 2026-04-23 17:25:38; freshness should be verified.

Geography

Ontario, Canada.