Analyte Concentration Calibration via Spectral Variance Isolation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for determining analyte concentrations in complex mixtures face challenges due to spectral overlap and over-fitting issues, particularly when dealing with insufficient data complexity and irrelevant variance sources, which impair predictive accuracy.
Innovation Solution
The proposed algorithm isolates the analyte's predominant contribution to a single independent variable, reduces colinearity, and creates adaptive regression using both training and test data sets, while projecting out the analyte spectrum and noise terms to mitigate over-fitting and extract variance from irrelevant sources.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional spectroscopic calibration methods are used to determine analyte concentration in complex mixtures, then the measurement can be performed, but spectral overlap between analyte and other compounds impairs prediction accuracy
Solution Approach 1:
The patent segments the spectral data into multiple components using multivariate curve resolution alternating least squares (MCR-ALS) algorithm. This separates the overlapping spectra of the analyte from interfering compounds, allowing independent analysis of each component's contribution to the total spectrum, thereby resolving the spectral overlap problem while maintaining measurement accuracy
Solution Approach 2:
The patent introduces an intermediary mathematical model (MCR-ALS algorithm) that acts as a mediator between the complex overlapping spectral data and the analyte concentration determination. This intermediary process decomposes the mixed spectrum into pure component spectra and quantifies each component's concentration, enabling accurate analyte measurement despite spectral interference from other compounds
2Device complexity
If calibration is created using insufficiently complex training data, then the calibration process is simpler, but the probability of over-fitting increases and predictive accuracy deteriorates
Solution Approach 1:
The patent implements dynamic calibration by allowing the calibration model to adapt its complexity based on the characteristics of the training data. The MCR-ALS algorithm dynamically determines the number of significant components and adjusts the model structure accordingly, preventing both over-fitting and under-fitting while maintaining predictive accuracy across varying data complexity levels
Solution Approach 2:
The patent incorporates feedback mechanisms through cross-validation and goodness-of-fit metrics that monitor calibration quality. The algorithm uses this feedback to iteratively refine the calibration model, adjusting parameters and component numbers to optimize predictive performance while preventing over-fitting, thus ensuring reliability without requiring excessively complex procedures
3Loss of information
If all spectral variance sources are included in calibration, then more information is available, but variance from irrelevant sources increases and impairs predictive accuracy
Solution Approach 1:
The patent extracts only the relevant spectral information corresponding to the analyte and significant interfering components using MCR-ALS. This extraction process separates meaningful variance (analyte concentration changes) from irrelevant variance (noise, instrumental drift, unrelated sample variations), retaining only the information necessary for accurate prediction while discarding harmful irrelevant variance
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
This approach enhances predictive accuracy by reducing the probability of over-fitting and effectively extracting variance associated with the analyte concentration, improving the calibration's robustness and accuracy in complex data sets.
Implementation Method 1
samples, which are illuminated with electromagnetic radiation, so as to produce a scattered spectrum
Data Source
AI summary
The concentration of analytes in a complex mixture can be ascertained by spectroscopic measurement, even if the spectra of substances other than the analyte overlap with that of the analyte. Both independently measured concentrations of the analyte in a training set and of the analyte spectrum are used. Variances in the spectral data attributable to the analyte are isolated from spectral variances from other causes, such as compositional changes associated with different samples that are independent of the analyte. For the special case of noninvasive glucose measurements on the skin of biological organisms, the volume averaged glucose in the sample is predicted from the blood glucose. A test for over-fitting of the data is also described.