Predicting exercise pulmonary hypertension: the right-net machine learning model a pilot study.

Area of Science:

• Cardiology
• Pulmonary Hypertension Research
• Medical Machine Learning

Background:

• Exercise transthoracic Doppler echocardiography (Ex-TTE) assesses mean pulmonary arterial pressure (mPAP)/cardiac output (CO) slope for diagnosing cardiorespiratory diseases.
• Clinical application of Ex-TTE for mPAP/CO slope determination requires validation.
• Pulmonary hypertension (PH) risk assessment can be improved with reliable noninvasive methods.

Purpose of the Study:

• To develop and validate a machine learning (ML) model for predicting abnormal exercise TTE-derived mPAP/CO slope (>3 mmHg/L·min).
• To identify individuals at risk of exercise PH using only clinical and resting TTE parameters.
• To establish a noninvasive tool for early identification of exercise PH.

Main Methods:

• Trained three ML models (Elastic Net, Classification and Regression Tree, LogitBoost) on resting clinical and TTE data from 417 participants (221 healthy, 196 connective tissue disease).
• Used data split into training/test sets to evaluate model performance in predicting mPAP/CO slope >3 mmHg/L·min.
• Selected the best-performing model based on the highest area under the curve (AUC) on the test set.

Main Results:

• The Elastic Net ML model demonstrated high performance with an AUC of 0.92.
• Key predictors for an abnormal mPAP/CO slope included lower tricuspid annular plane systolic excursion/systolic PAP ratio, female sex, and smaller left ventricular outflow tract diameter.
• These findings highlight specific clinical and echocardiographic features indicative of exercise PH risk.

Conclusions:

• A machine learning algorithm utilizing resting clinical and TTE parameters effectively predicts abnormal exercise TTE-derived mPAP/CO slope.
• This ML approach supports the use of noninvasive resting parameters to identify individuals at risk of exercise PH.
• The study validates ML as a valuable tool for early detection and risk stratification in potential exercise PH cases.