The Need

Clinicians managing pediatric patients supported by a ventricular assist device must routinely assess pump performance by visually estimating membrane fill and ejection during pump cycles. These evaluations are critical for patient management but are inherently subjective, time-consuming, and prone to inter-observer variability. Because membrane motion occurs rapidly, clinicians may rely on slow-motion video review to make judgments. A standardized, objective, and reproducible method for quantifying pump status could improve consistency of care, reduce monitoring burden, and support more informed clinical decision-making.

The Technology

OSU and Nationwide Children's Hospital researchers have developed a machine learning and computer vision technology to automatically analyze video recordings of external heart pump membrane motion and generate quantitative measurements of pump fill and ejection status. The platform is trained using expert clinician assessments to produce standardized outputs that align with clinical interpretation. Rather than relying solely on subjective visual review, the system converts membrane motion into objective performance metrics. The software is designed to operate using routine clinical video data and can ultimately be integrated into a user-friendly workflow for bedside monitoring, decision support, and training applications.

Benefits/Advantages

• Quantitative outputs: Provides actionable numerical information rather than subjective qualitative observations.
• Standardized assessment: Reduces variability between clinicians by providing objective, reproducible measurements.
• Improved workflow efficiency: Eliminates the need for extensive manual slow-motion video review.
• Compatible with existing practices: Designed to leverage routine clinical video acquisition workflows without requiring specialized hardware.