Abstract:
In a noninvasive system for detection/measurement of glucose and other analytes in a medium such as tissue, illumination is directed to the medium and corresponding radiation from the medium is collected. Spectral energy changes associated with fragment(s)/feature(s) obtained from the collected radiation are determined using collision computing. Such spectral energy changes generally represent analyte concentration. The collection of radiation and/or illumination is controlled either to target a particular volume of the medium or such that the spectral energy changes become directionally monotonic with respect to analyte concentration, or both. The collection parameters include: duration of collection, location and/or a size of a collection spot on the medium surface, and angle of a collector relative to the medium surface. The illuminated and/or collection spots may be treated to improve accuracy of analyte measurement.
Abstract:
A synthetic projection system determines analyte concentration, such as blood glucose concentration, from a spectral-energy change associated with an uncharacterized instance of a medium in which the analyte is likely present. The projection system is factory calibrated for different instances of the medium, without needing instance-specific training or calibration. The projection system includes a set of projector curves, each relating spectral-energy change values obtained by analyzing reference medium samples to analyte concentrations in those samples. Each projector curve also corresponds to a respective range of energy-change gradients, determined using a group of surrogate media characterized according to analyte concentrations measured using a reference system. A spectral-energy-change gradient for the uncharacterized medium may be computed to select one of the projectors curves. Analyte concentration in the uncharacterized medium can be reliably computed at a specified high level of accuracy using the spectral-energy change associated therewith and the selected curve.
Abstract:
A collision-computing system detects and amplifies the energy associated with a feature signal to determine occurrences or absence of events, such as ultrasonic and/or geophysical events, or to determine presence and/or concentrations of substances such as blood glucose, toxic chemicals, etc., in a noisy, high-clutter environment or sample. To this end, a conditioned feature, obtained by modulating a carrier kernel with a feature signal, is collided with a Zyoton—a waveform that without a collision can travel substantially unperturbed in a propagation medium over a specified distance. The conditioned feature and the Zyoton are particularly constructed to be co-dependent in terms of their respective dispersion velocities and the divergence of a waveform resulting from the collision. The collision operation can transfer at least a portion of the feature energy to the resulting waveform, and the transferred energy can be amplified in successive collisions for detecting/measuring events/substances.