AI Stroke Volume Prediction From Arterial Blood Pressure
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for calculating stroke volume, such as arterial pressure-based cardiac output (APCO) are less accurate and invasive, while thermodilution-based cardiac output (TDCO) is more accurate but invasive and poses complications.
Innovation Solution
A device and method using AI to calculate stroke volume by pre-training and transfer learning a stroke volume calculation model with filtered arterial blood pressure data, excluding outliers and performing data smoothing and delay processing to enhance accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If APCO equipment is used to calculate stroke volume, then the method is less invasive, but the accuracy is reduced
Solution Approach 1:
The patent transforms the mathematical calculation parameters from using average or standard deviation of pulse pressures to using a machine learning model that processes multiple arterial blood pressure parameters (including systolic pressure, diastolic pressure, mean pressure, and pulse pressure) to predict stroke volume. This parameter transformation enables the non-invasive APCO method to achieve accuracy comparable to invasive TDCO methods.
Solution Approach 2:
The patent replaces the traditional mechanical/mathematical calculation system with a machine learning-based prediction system. By training a model on large datasets containing both arterial blood pressure measurements and corresponding stroke volume values, the system substitutes complex mathematical derivations with pattern recognition, significantly improving prediction accuracy while maintaining non-invasiveness.
2Measurement precision
If TDCO equipment is used to calculate stroke volume, then the accuracy is improved, but the invasiveness increases and complications occur
Solution Approach 1:
The patent creates a virtual copy of the invasive TDCO measurement capability through machine learning. By training the model on datasets that include both APCO measurements and TDCO reference measurements, the system learns to predict TDCO-equivalent stroke volume values using only non-invasive arterial blood pressure data, effectively copying the accuracy of invasive methods without the associated risks.
Solution Approach 2:
The patent introduces a machine learning model as an intermediary between the non-invasive APCO measurements and the invasive TDCO reference standard. This intermediary translates arterial blood pressure parameters into accurate stroke volume predictions by learning the complex relationship between these measurements from training data, eliminating the need for direct catheter insertion while maintaining accuracy.
3Device complexity
If APCO equipment uses average or standard deviation of pulse pressures, then the calculation is simpler, but the prediction accuracy decreases
Solution Approach 1:
The patent transitions from static mathematical formulas (using fixed relationships between pulse pressure and stroke volume) to a dynamic machine learning model that adapts to individual patient characteristics. The model processes multiple varying parameters including systolic pressure, diastolic pressure, mean pressure, and pulse pressure, adjusting predictions based on the specific pattern of each patient's blood pressure waveform rather than applying a uniform calculation method.
Data Source
Figure 1~2B
Figure 2C~3
Figure 4
AI summary
The present invention provides a device for calculating stroke volume using AI, which includes a filtering unit configured to filter an arterial blood pressure value and a stroke volume which are in a preset range, among first data and second data including a plurality of arterial blood pressure values and stroke volumes corresponding to the arterial blood pressure values; a pre-training unit configured to pre-train a first stroke volume calculation model which calculates a stroke volume based on the arterial blood pressure value, by using third data filtered from the first data; a transfer learning unit configured to transfer learn the first stroke volume calculation model which calculates a stroke volume based on the arterial blood pressure value, by using fourth data filtered from the second data, thus to generate a second stroke volume calculation model; and a stroke volume calculation unit configured to calculate a stroke volume corresponding to the input arterial blood pressure of a specific patient by using the second stroke volume calculation model.