Development of hypertension models for lung cancer screening cohorts using clinical and thoracic aorta imaging...

Area of Science:

• Cardiology
• Radiology
• Oncology

Background:

• Hypertension is a significant comorbidity.
• Lung cancer screening identifies at-risk populations.
• Predictive tools for hypertension in this cohort are needed.

Purpose of the Study:

• Develop and validate nomogram models for hypertension risk prediction.
• Utilize clinical and thoracic aorta imaging factors.
• Assess model performance for clinical utility.

Main Methods:

• Included 804 patients for training (70%) and validation (30%) sets.
• Selected thoracic aorta imaging features using statistical methods and LASSO.
• Constructed five nomogram models (AIMeasure, BasicClinical, TotalClinical, AIBasicClinical, AITotalClinical).
• Evaluated models using ROC curves, calibration curves, and decision curve analysis (DCA).

Main Results:

• Models demonstrated good predictive capability in both training and validation sets (AUCs ranging from 0.73 to 0.84).
• Calibration curves and DCAs confirmed accuracy and clinical practicality.
• Nomogram models showed significant predictive performance for hypertension risk.

Conclusions:

• Nomogram models integrating clinical and thoracic aorta imaging factors are effective for hypertension risk assessment.
• These models offer clinical utility for early hypertension risk stratification.
• Facilitates timely non-pharmacological interventions to manage or prevent hypertension.