Blood Pump ML Estimation of Cardiac Output Without Catheters

The patent creates a virtual copy of the catheter measurement system through machine learning models. The model is trained on data from patients with catheters to learn the relationship between pump parameters and cardiac parameters. Once trained, the model can estimate cardiac parameters from pump parameters alone, creating a software-based copy that eliminates the need for physical catheters in routine monitoring while maintaining measurement precision.

The patent replaces the mechanical catheter-based measurement system with a computational model. Instead of physically inserting catheters to measure cardiac parameters, the system uses machine learning algorithms that process pump parameter data to estimate cardiac function. This substitution eliminates the mechanical intrusion while preserving the ability to obtain accurate cardiac parameter measurements.

The patent segments the patient population into different groups based on characteristics such as disease state, demographics, and clinical parameters. Multiple specialized models are trained for different patient segments, allowing the system to adapt to specific population characteristics while maintaining high accuracy for each group. The segmentation approach enables tailored modeling that preserves precision within each segment while achieving broad adaptability across all patient types.

The patent implements a dynamic model selection process where the system determines which patient population the current patient belongs to and selects the appropriate pre-trained model. This dynamic adaptation allows the system to maintain high accuracy by matching the right model to the right patient population, rather than using a single static model for all patients.