Optimizing point-of-care ultrasound video acquisition for probabilistic multi-task heart failure detection.

Authors

Abstract

Echocardiography with point-of-care ultrasound (POCUS) must support clinical decision-making under tight bedside time and operator-effort constraints. We introduce a personalized, budget-constrained data acquisition strategy in which a reinforcement learning agent, given a partially observed multi-view study, selects the next view to acquire or terminates acquisition to support heart failure (HF) assessment. Upon termination, a diagnostic model jointly predicts aortic stenosis (AS) severity and left ventricular ejection fraction (LVEF), two key HF biomarkers, and outputs calibrated uncertainty, enabling an explicit trade-off between diagnostic performance and acquisition cost. We model bedside POCUS as a sequential, cost-constrained acquisition problem in which an RL agent retrospectively selects among five predefined standard views, each represented by one clip with equal acquisition cost, or terminates acquisition. Upon termination, a shared multi-view transformer performs multi-task inference with two heads, ordinal AS classification and LVEF regression, and outputs Gaussian predictive distributions yielding ordinal probabilities over AS classes and EF thresholds. These probabilities drive a reward that balances expected diagnostic benefit against acquisition cost (e.g., number of acquired views), producing patient-specific acquisition pathways rather than a fixed protocol. The video selector is trained with online RL, updating on-policy in a partial-observation simulator built from complete multi-view POCUS studies. The dataset comprises 12,180 patient-level studies, split into training/validation/test sets (70/15/15). On the 1820 test studies, our method matches full-study performance while using <math xmlns="http://www.w3.org/1998/Math/MathML"><mrow><mn>32</mn> <mo>%</mo></mrow> </math> fewer videos, achieving <math xmlns="http://www.w3.org/1998/Math/MathML"><mrow><mn>77.2</mn> <mo>%</mo></mrow> </math> mean balanced accuracy across AS severity classification and LVEF estimation, demonstrating robust multi-task performance under acquisition budgets. In this constrained retrospective setting, patient-tailored view selection preserves decision quality while reducing the number of acquired views, supporting adaptive acquisition policies. The framework is extensible to additional cardiac endpoints and warrants prospective evaluation in live bedside POCUS. The code is available at https://github.com/Armin-Saadat/Double-Precise.

Tags

Topics