ISE Recalibration via Genetic Algorithm and Neural Network
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Ion Selective Electrodes (ISEs) used for assessing analyte ion concentration in liquids face degradation in sensitivity and selectivity over time, leading to inaccurate calibrations due to interference from other ions, and existing recalibration methods are impractical as they require multiple samples and recalibration of neural network models.
Innovation Solution
A method involving a Genetic Algorithm (GA) to recalibrate ISEs by simulating responses to calibration and interference ions using known ionic concentrations, allowing for automatic recalibration with fewer standard solutions and accounting for ion and electrode interference.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional calibration methods using multiple known ionic concentration samples are used to recalibrate ISEs, then measurement precision may be improved, but the complexity of the calibration process and time required increase significantly
Solution Approach 1:
The patent transforms the recalibration problem from a complex multi-parameter optimization task into a simpler problem by changing the approach: instead of adjusting multiple neural network parameters simultaneously, it uses a simulated response algorithm that calculates expected ISE responses based on known interference relationships. This parameter transformation allows accurate recalibration using minimal samples while avoiding the complexity of full neural network retraining.
Solution Approach 2:
The patent creates a simulated copy of the ISE response behavior through the simulated response algorithm. This virtual model replicates how ISEs respond to various ion combinations without requiring physical re-measurement of all training samples. The simulated responses are then used to update the neural network, effectively copying the essential calibration information from a simplified model.
2Measurement precision
If the neural network model is retrained using the complete training data set for each recalibration, then measurement precision is maintained, but the time and resources required for recalibration increase
Solution Approach 1:
Instead of performing the complete action of retraining the neural network with the entire training data set, the patent applies partial action by using only the simulated response data from a minimal set of recalibration samples. This partial approach captures the essential calibration information needed to maintain precision while dramatically reducing the time and computational resources required.
Solution Approach 2:
The patent performs preliminary action by pre-calculating the simulated ISE responses based on known interference relationships and sample compositions before updating the neural network. This preliminary simulation prepares the calibration data in advance, allowing the neural network to be updated efficiently without requiring time-consuming retraining on the full historical data set.
3Measurement precision
If ISEs are used to detect analyte ions in complex liquid samples, then the ability to assess concentration is achieved, but interference from other ions degrades selectivity and accuracy
Solution Approach 1:
The patent implements feedback by using the simulated response algorithm to calculate what the ISE responses should be for given sample compositions, then comparing these simulated responses with actual measurements. The neural network uses this feedback to learn and compensate for interference patterns, continuously improving its ability to distinguish target ions from interfering ions in complex samples.
Solution Approach 2:
The simulated response algorithm acts as an intermediary between the raw ISE measurements and the final concentration determination. It processes the complex interference relationships and transforms them into corrected calibration data that the neural network can use, effectively mediating the harmful effects of ion interference before they reach the final measurement output.
Data Source
AI summary
The present invention relates to a method of recalibrating a device for assessing concentration of at least one analyte ion in a liquid, the device having a plurality of ion selective electrodes (ISEs) generating a signal in response to sensing a selected ion in the liquid, and a data processing unit implementing a neural network algorithm that has been trained to calculate ion interference between the selected ion and other ions in the liquid sensed at one of the electrodes and/or electrode interference between ones of the electrodes sensing a same selected ion.