Association of artificial intelligence derived cardiothoracic ratio assessment on non-cardiac chest CT with heart failure and all-cause mortality: A retrospective single center study.

BACKGROUND

The cardiothoracic ratio (CTR) is estimated by dividing cardiac width by thoracic width. The Area Deprivation Index (ADI) is a metric to quantify socioeconomic conditions. This study investigates the use of artificial intelligence (AI) to identify elevated CTR as a predictor for heart failure (HF) and mortality and assessed ADI's influence on these risks.

METHODS

This retrospective cohort study included 9,693 consecutive patients with non-cardiac chest CTs. An AI algorithm automated measurement of cardiac and thoracic diameters to calculate the cardiothoracic ratio (CTR). Patients were categorized into CTR tertiles: <0.5 (normal), 0.5-0.55 (borderline cardiomegaly), and ≥0.56 (cardiomegaly). Socioeconomic metrics were derived from electronic health records and stratified by ADI quartiles. Cardiovascular outcomes were extracted from ICD-10 codes over a six-year follow-up. Associations between CTR and cardiac outcomes were assessed using multivariate logistic regression and Cox proportional hazard models adjusted for age, sex, race, and ADI.

RESULTS

Elevated CTR, notably in the cardiomegaly group, was associated with increased risk of prevalent HF (OR = 6.17, 95% CI = 5.27-7.23) and all-cause mortality (OR = 1.66, 95% CI = 1.47-1.87). Higher ADI scores were linked to increased risk of HF and mortality, although there was inconsistent interaction between elevated CTR and ADI regarding mortality.

CONCLUSIONS

AI-derived CTR on non-cardiac chest CT may provide a cost-effective and efficient method for identifying patients at increased risk for HF and mortality. The utility of this technology proves to be efficacious amongst patients with high socioeconomic burden.