ECG-Based Analyte Prediction Using GPR Data Filtering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing non-invasive analyte monitoring techniques face challenges in accurately determining electrolyte concentrations due to errors in training data, particularly when the distribution of analyte measurements is non-uniform, leading to reduced accuracy in machine learning models.
Innovation Solution
The use of filtering rules and Gaussian Process Regression (GPR) analysis to identify and filter potentially inaccurate measurements, adjusting labeling data to improve the accuracy of machine learning models by considering confidence intervals and thresholds for analyte levels in electrocardiogram data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If machine learning models are trained on raw analyte measurement data, then the model can be developed quickly, but the prediction accuracy decreases due to errors and non-uniform distribution in training data
Solution Approach 1:
The patent applies preliminary action by filtering and cleaning training data before model training. The system identifies and removes potentially inaccurate measurements using statistical analysis and confidence interval calculations, ensuring that only high-quality data is used to train the machine learning model, thereby improving prediction accuracy without significantly delaying development
Solution Approach 2:
The patent introduces an intermediary data processing layer between raw analyte measurements and model training. This intermediary system applies Gaussian Process Regression and other statistical methods to assess data quality, generate confidence intervals, and filter measurements, acting as a mediator that transforms raw data into reliable training samples
2Measurement precision
If filtering rules and GPR analysis are applied to training data, then prediction accuracy improves, but system complexity and processing time increase
Solution Approach 1:
The patent applies parameter changes by adjusting confidence interval thresholds and filtering criteria based on the specific analyte being measured. The system dynamically modifies processing parameters such as confidence levels, error margins, and filtering strictness to optimize the balance between accuracy improvement and computational complexity for different measurement scenarios
3Reliability
If confidence intervals and thresholds are calculated for all measurements, then data accuracy improves, but processing time and computational resources increase
Solution Approach 1:
The patent applies partial action by calculating confidence intervals and applying filtering rules selectively rather than uniformly to all measurements. The system identifies measurements that require rigorous analysis based on initial quality indicators, applying comprehensive confidence interval calculations only to borderline or suspicious cases, thereby maintaining reliability while reducing overall processing time
Data Source
AI summary
Disclosed are techniques for training a machine learning model. A training data set comprising a plurality of electrocardiogram (ECG) measurements and corresponding analyte measurements for each of the plurality of ECG measurements is provided. For each ECG measurement of the training data set, an estimated analyte level at a time of the ECG measurement is determined based on the corresponding set of analyte measurements. The estimated analyte level may be determined using statistical estimation techniques. If it is determined that the estimated analyte level at the time of the ECG measurement meets a certainty threshold, the ECG measurement is labeled based on the estimated analyte level at the time of the ECG measurement. A machine learning is trained to predict a level of an analyte based on ECG data, where the training is done using each of the plurality of ECG measurements that are labeled.