The added value of echocardiography in pulmonary arterial hypertension risk assessment: an artificial intelligence machine learning-derived analysis of the ULTRA RIGHT VALUE registry

Right ventricular morphological and functional echocardiographic parameters were associated with low-risk status: a one-unit decrease in any morphological parameter increased the odds of achieving/maintaining low-risk status by up to 69% and 92% for the ESC/ERS and REVEAL scores, respectively, while a one-unit increase in any functional parameter increased those odds by up to 1.28 and 1.78. Prior risk scores had not fully incorporated right ventricular assessment; this study quantified, in a 401-patient cohort, the strength of association between echocardiographic morphology and function and low-risk status. Based on a multicentre prospective cohort of 401 patients enrolled between January 2022 and December 2023, reporting odds ratios, i.e. observational association evidence.

Adding echocardiographic variables to the ESC/ERS and REVEAL 2.0 scores improved the c-index for morbi-mortality prediction from 0.73 (95% CI 0.69-0.76) to 0.78 (95% CI 0.75-0.81) and from 0.74 (95% CI 0.67-0.83) to 0.79 (95% CI 0.76-0.82), respectively, both with P<0.01. Relative to using the existing scores alone, adding echocardiographic variables produced a measurable gain in discrimination, consistent in direction across two independent scoring systems. Echocardiography and both scores were assessed in the same prospective cohort, with confidence intervals and P values reported, representing moderate-strength evidence of incremental predictive value.

Machine learning analysis showed that indices of RV-PA coupling had the highest influence on the c-index, making RV-PA coupling the most performant echocardiographic dimension. It localizes RV-PA coupling as the key dimension among many echocardiographic parameters, offering direction for future measurement and modelling. The conclusion derives from machine learning variable-importance analysis, an internal model inference; no external validation is reported in the text.

The results apply to adults with prevalent PAH in a European multicentre setting who underwent echocardiography and ESC/ERS and REVEAL 2.0 assessment, suggesting that adding right ventricular echocardiographic parameters—especially RV-PA coupling indices—to existing scores can improve risk discrimination; consistency for incident patients, other regions, and across different echocardiographic operators remains to be clarified.

A careful reader would still watch for the specific machine learning model type and validation approach, the exact measurement definitions of RV-PA coupling indices, the practical meaning of the c-index gain for clinical decisions, and whether the increment is stable across subgroups and follow-up durations; this summary is based on abstract-level text, so details in figures, tables, and supplementary material are not included.