Artificial Intelligence-Derived Risk Prediction: A Novel Risk Calculator Using Office and Ambulatory Blood Pressure.

Area of Science:

• Cardiovascular disease risk prediction
• Machine learning in healthcare
• Hypertension management

Background:

• Accurate cardiovascular risk quantification is crucial for personalized hypertension treatment.
• Existing risk scores may not fully capture individual patient risk.
• Novel predictive models are needed to improve cardiovascular event prediction.

Purpose of the Study:

• To develop and validate a machine learning model for predicting cardiovascular mortality risk.
• To compare the novel model's performance against established risk scores.
• To assess the utility of clinical variables in predicting ambulatory blood pressure phenotypes.

Main Methods:

• Utilized a large dataset (59,124 patients) from the Spanish Ambulatory Blood Pressure Monitoring registry.
• Applied machine learning techniques including logistic regression, gradient-boosted decision trees, and deep neural networks.
• Employed stepwise forward feature selection to identify key predictive variables.

Main Results:

• Deep neural networks achieved the highest performance in predicting cardiovascular mortality.
• The novel model, using both office blood pressure (OBP) and ambulatory blood pressure (ABP), outperformed existing risk scores (Framingham, SCORE2, ASCVD).
• ABP-based prediction significantly increased the area under the curve (AUC) compared to OBP (0.870 vs. 0.865), enhancing accuracy and specificity.

Conclusions:

• Machine learning models, particularly deep neural networks, show significant potential for improving cardiovascular risk prediction.
• Utilizing ambulatory blood pressure data substantially enhances the accuracy of cardiovascular mortality prediction over office blood pressure.
• The developed model offers a promising advancement for individualizing hypertension treatment and cardiovascular risk assessment.