AI Patient Monitoring With Adaptive Thresholds and Caregiver Feedback

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current patient monitoring systems face challenges with inaccurate alerts due to pre-defined static thresholds and limited learning capabilities, leading to both missed events (false negatives) and unnecessary alerts (false positives), reducing caregiver trust and increasing workload.

Innovation Solution

An AI-powered patient monitoring system that integrates real-time data analysis, adaptive alert thresholds, multimodal caregiver feedback, and machine learning-driven data labeling to refine alert generation models, enhancing accuracy and reliability.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Ease of operation

If pre-defined static thresholds are used for alert generation, then the system is simple to operate and implement, but the alert accuracy deteriorates leading to false positives and false negatives

Engineering Contradiction:
Improveease of operationVSAvoidalert accuracy
Core Design Contradiction:
Ease of operationVSMeasurement precision

Solution Approach 1:

The patent transforms static alert thresholds into dynamic, adaptive thresholds that automatically adjust based on learned patient patterns and historical data. The system continuously refines threshold values through machine learning algorithms, enabling the monitoring system to adapt to individual patient variations while maintaining operational simplicity.

Inventive Principle:
Principle #15Dynamics

Solution Approach 2:

The system implements feedback loops where alert outcomes and caregiver responses are continuously fed back into the machine learning model. This feedback mechanism allows the system to learn from false positives and false negatives, progressively improving alert accuracy while maintaining the simplicity of the user interface and operation.

Inventive Principle:
Principle #23Feedback

2Loss of time

If machine learning models are trained on limited data, then the training process is faster and requires less computational resources, but the model generalization capability deteriorates

Engineering Contradiction:
Improvetraining timeVSAvoidmodel generalization
Core Design Contradiction:
Loss of timeVSReliability

Solution Approach 1:

The system performs preliminary actions by continuously pre-processing and storing patient data in structured formats ready for training. Historical data is maintained and pre-labeled with outcomes, so when training is needed, the system can quickly utilize this prepared dataset without requiring extensive data collection and processing time.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The machine learning system performs self-service by automatically identifying data quality issues, selecting relevant features, and performing hyperparameter tuning without extensive manual intervention. This automation reduces the time required for model preparation and training while improving generalization through consistent, reproducible processing pipelines.

Inventive Principle:
Principle #25Self-service

3Measurement precision

If complex AI models are deployed for patient monitoring, then alert accuracy improves, but the device complexity and computational requirements increase

Engineering Contradiction:
Improvealert accuracyVSAvoiddevice complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent segments the AI processing into multiple components: edge computing on the patient monitor for real-time feature extraction, cloud-based processing for model training and updates, and mobile device processing for caregiver interface. This segmentation allows complex algorithms to be distributed across multiple devices, reducing the computational burden on any single device while maintaining high alert accuracy.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The system introduces an intermediary layer that translates complex AI model outputs into simple, actionable alerts for caregivers. This intermediary processing layer simplifies the interface between complex computational models and the user interface, maintaining high accuracy while reducing perceived device complexity for end users.

Inventive Principle:
Principle #24Intermediary (Mediator)

4Reliability

If real-time data processing is implemented continuously, then patient safety monitoring is improved, but the energy consumption and computational load increase

Engineering Contradiction:
Improvepatient safety monitoringVSAvoidenergy consumption
Core Design Contradiction:
ReliabilityVSUse of energy by moving object

Solution Approach 1:

The system implements periodic action by processing data at varying frequencies based on patient risk levels and data stability. During stable periods, processing occurs at lower frequencies to conserve energy, while during periods of change or high risk, the system increases processing frequency to maintain patient safety. This adaptive periodic processing maintains reliability while reducing overall energy consumption.

Inventive Principle:
Principle #19Periodic action

Data Source

PatentEP4645335A1Systems and methods for ai-powered patient monitoring
Publication Date: 2025.11.05 GE PRECISION HEALTHCARE LLC
  • EP4645335A1 patent drawingFigure 1
  • EP4645335A1 patent drawingFigure 2
  • EP4645335A1 patent drawingFigure 3

AI summary

Methods and systems are proposed that integrate real-time data analysis, adaptive alert thresholds, multimodal caregiver feedback, and data labeling to continuously refine alert generation models (500, 600, 650) relied on by patient monitoring systems (220). The proposed approach enhances the effectiveness of AI-powered patient monitoring systems (220) by addressing the challenges of inaccurate data labeling and high false alert rates. To minimize false alerts, caregivers are provided with an easy-to-use feedback tool (700) to confirm receipt of alerts, categorize alerts as true or false positives, and provide contextual information. This caregiver-provided information is then analyzed, and criteria may be extracted from the information that may be used to retrain or refine the alert generation models (500, 600, 650).