Intelligent system for noninvasive blood analyte prediction

An intelligent system for measuring blood analytes noninvasively by operating on a near infrared (NIR) absorbance spectrum of in vivo skin tissue, said system comprising:

• a pattern classification engine for adapting a calibration model to the structural properties and physiological state of a subject as manifested in said NIR absorbance spectrum;
• and
• means for reducing spectral interference by applying calibration schemes specific to general categories of subjects that have been segregated into classes;
• wherein a priori information about primary sources of sample variability is used to establish said general categories of subjects.

An intelligent system for measuring blood analytes noninvasively by operating on a near infrared (NIR) absorbance spectrum of in vivo skin tissue, said system comprising:

• an execution layer that receives tissue absorbance spectra from an instrument and that performs rudimentary preprocessing;
• a coordination layer that performs feature extraction;
• a classification system that is used to classify a subject according to extracted features that represent the state and structure of a sample;
• wherein predictions from one or more existing calibration models are used to form an analyte estimate based on said classification.

A pattern recognition method for estimating a concentration of a target blood analyte, comprising the step of:

• classifying new spectral measurements into previously defined classes through structural and state similarities as observed in a tissue absorbance spectrum, according to a pattern classification method;
• wherein class membership is an indication of which calibration model is most likely to accurately estimate the concentration of the target blood analyte;
• said pattern classification method comprising the steps of:
• extracting features;
• and
• classifying said features according to a classification model and decision rule.