Infusion Pump Occlusion Detection Using SVM Force Compensation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional occlusion detection methods in infusion pumps are unreliable due to external factors influencing force sensor measurements, leading to false alarms and inaccurate occlusion detection.
Innovation Solution
A method and system using artificial intelligence, specifically a support vector machine, to precondition input data from infusion pump pressure samples, incorporating current and historical force measurements to differentiate between slowly changing forces and occlusions, enabling accurate real-time occlusion detection.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a force sensor is used to detect in-line IV tubing pressure, then occlusion detection capability is provided, but measurement precision deteriorates due to external factors influencing the force sensor measurements
Solution Approach 1:
The patent introduces an intermediary processing layer (algorithmic compensation) between the force sensor and the occlusion detection decision. The system uses reference measurements taken at different times (initial reference, intermediate reference) as mediators to compensate for external factors. By subtracting these reference measurements from current measurements, the system isolates the true pressure signal from confounding external influences, thereby improving measurement precision while maintaining occlusion detection capability.
Solution Approach 2:
The patent changes the parameter being measured by the force sensor through temporal reference comparisons. Instead of relying on absolute force sensor readings, the system measures the change in force relative to reference states taken at different times. This parameter transformation (from absolute force to force differential) eliminates the influence of external factors that remain relatively constant over time, thereby improving measurement precision for occlusion detection.
2Reliability
If force sensor measurements are used for occlusion detection, then occlusion detection is enabled, but false alarms increase due to external factors such as IV tubing relaxation and height variations
Solution Approach 1:
The patent implements feedback through continuous reference measurements taken at different stages of the infusion process. The system periodically updates reference measurements (initial reference at start, intermediate reference during infusion) and uses these feedback signals to adjust and compensate for changing external conditions. This feedback mechanism allows the system to distinguish between normal variations (like tubing relaxation) and actual occlusions, thereby reducing false alarms while maintaining reliable occlusion detection.
Solution Approach 2:
The patent takes preliminary action by establishing baseline reference measurements before and during the infusion process. By capturing reference data at the initial stage and intermediate stages, the system proactively accounts for external factors before they cause false alarms. This preliminary characterization of the system's baseline state enables subsequent measurements to be interpreted in context, reducing false positive occlusion detections.
3Reliability
If conventional force sensor methods are used, then occlusion detection is provided, but device complexity increases due to the need for baseline calculations and adjustments
Solution Approach 1:
The patent enables the system to self-adjust through automated algorithmic processing of reference measurements. Rather than requiring manual baseline setup and adjustment by operators, the system automatically captures reference measurements at different stages and computationally compensates for external factors. This self-service approach reduces operational complexity while maintaining reliable occlusion detection, as the compensation calculations are performed automatically by the infusion pump's control system.
Data Source
AI summary
Systems and methods for detecting an occlusion in an infusion device are disclosed. One method includes generating, during an infusion session at predetermined intervals, a feature vector including an initial measurement of a physical force caused by fluid motion through an infusion device at the start of an infusion session, a current measurement of the physical force, a long term change in the measurement of the physical force, and a short term change in the measurement of the physical force. The feature vector is inputted, during the predetermined intervals, into a support vector machine (“SVM”) to output an indication of a presence or an absence of an occlusion in the infusion device. The SVM may be trained using reference data from reference infusion sessions having known consequences regarding the presence or the absence of an occlusion at various times during the reference infusion sessions.