Systems and methods for measuring the concentration of analytes

Abstract

Techniques for acquiring and processing data in combination with a photonic sensor SoC (system-on-a-chip) (1) to provide real-time corrected concentration levels of an analyte (e.g., a constituent molecule within a biological material) are described. The raw signal (1300) to be analyzed is collected by the sensor chip (1) via diffuse reflectance or transmittance. Determination of the analyte concentration is based, in part, on Beer-Lambert principles and is facilitated by applying scattering correction (2240) to the raw signal prior to decomposition and analysis of the raw signal (1300).

Description

Technology Field

[0001] This application claims priority and interest to U.S. provisional patent application No. 62 / 944,644, filed on December 6, 2019, titled "Systems and Methods for Measuring Concentration of an Analyte," the entire contents of which are incorporated herein by reference.

[0002] Embodiments of the present invention relate to a method for acquiring data from a target biomaterial by optical communication between the target material and a III-V / IV semiconductor photonic sensor, and a data processing method for retrieving the absolute concentration level of a target molecule within the material. This is applicable, but not limited to, transdermal sensing and monitoring of blood glucose, urea, lactate, creatinine, ethanol, and other constituent molecules by tunable wavelength absorption spectroscopic sensing. The described technology is compatible with consumer electronics technology platforms in terms of manufacturing technology and size, weight, power, and cost requirements, and provides pivotal advantages in terms of utility for wearable healthcare device technology. This technology can be utilized by people impaired by chronic diseases, such as diabetes, for whom no non-invasive sensing solution currently exists. Furthermore, a novel approach is provided for the non-invasive and continuous monitoring of biophysiological markers in cases where only a point-of-care solution currently exists. Background Technology

[0003] Many techniques for spectroscopic non-invasive measurements of analytes, such as blood glucose measurements using near-infrared spectroscopy, utilize broadband light sources, such as halogen lamps. Electromagnetic radiation (EMR) emitted from these sources and received by the medium to be analyzed (e.g., diffusely reflected by the medium or transmitted through it) contains components of multiple wavelengths. Typically, components from the EMR received from the medium are separated using grating techniques to acquire a spectrum. Spectrometers with broadband sources and grating mechanisms are typically large and complex structures that can be cumbersome or impractical for field or home use.

[0004] Systems-on-a-chip (P-SoC) offers the ultimate potential for size reduction required for high-volume applications such as consumer electronics markets, automotive, and home medical devices. The P-SoC concept combines all or most of the functions of a typical photonic system, enabling these functions to be executed within a single-chip assembly. Typically, this can be realized as a monolithic photonic integrated circuit (PIC) based on III-V semiconductors or a combination of III-V and Group IV semiconductors. The first approach allows for a completely monolithic device by enabling all active and passive optical components to be realized on the same wafer. This is ideal because all light sources and detectors are essentially aligned with waveguides and require no assembly steps. However, along with complex techniques requiring multiple epitaxial growths, inherent III-V material properties—such as higher absorption and lower optical confinement in waveguides, and consequently larger waveguide bend radii to reduce bending losses—limit scalability to very large markets like consumer electronics, as these markets require very low costs per chip. As a compromise, hybrid III-V / IV P-SoCs provide a solution where photogeneration functions are realized within a III-V semiconductor chip, while optical routing, filtering, and other functions are realized within a Group IV semiconductor chip. Wavelength-dependent photodetection of EMR can be realized within either a III-V or Group IV semiconductor chip. The hybrid approach proves beneficial to high-volume markets because Group IV semiconductor manufacturing technologies, such as CMOS, offer unrivaled scalability. However, techniques for analyte measurement using P-SoCs are not generally known.

[0005] The hybrid integration of III-V semiconductor chips and Group IV semiconductor photonic integrated circuits provides the possibility of combining the best of the two worlds, as photodetection and photogeneration functions are realized within direct bandgap III-V semiconductors for optimal efficiency, performance, cost, and yield, while passive functions such as photofiltering, routing, locking, and feedback control are realized within photonic integrated circuits (PICs) in Group IV semiconductors, e.g., silicon-on-insulator, silicon-on-silicon nitride, silicon-on-silicon nitride, or silicon-on-insulator. In various embodiments, a swept wavelength laser-based photonic system-on-chip having integrated emission wavelength tuning (sweeping), wavelength shift tracking, and absolute wavelength correction functions is deployed for the remote acquisition of relevant data from a biological object, such as a living organism. Subsequently, in different embodiments, the acquired data is processed to provide biomolecular-specific absolute values, such as concentration levels, and / or concentration levels as a function of time (trend). The combination of technologies for processing acquired data on-chip and hybrid III-V / IV semiconductor platforms provides new opportunities for wearable device platforms, such as smartwatches for monitoring important physiological parameters in real time, for example.

[0006] Techniques for acquiring and processing data in combination with a photonic sensor system-on-a-chip are described to provide real-time calibrated concentration levels of an analyte (e.g., constituent molecules within a biological material). The biological material may be blood, interstitial fluid, tissue, or a combination of such materials. The photonic sensor system-on-a-chip (SoC) assembly comprises a hybrid III-V and IV semiconductor assembly, wherein the III-V semiconductor elements provide optical gain and detection functions, and optical feedback, optical routing, filtering, locking, and other passive functions are provided within the IV semiconductor photonic integrated circuit.

[0007] In use, the assembly communicates optically with the biological material, and the sensor can be located remotely from the material (in vivo scenario) or embedded within the material (implantation). The sensor interacts with the target material via optical communication, light from the sensor interacts with the material, and the optical signal is modulated due to photo-molecular interactions, where the interaction is molecule-specific. After the interaction, the signal is collected by the sensor chip based on diffuse reflectance or transmittance.

[0008] In practical scenarios where such photon sensors perform direct transdermal measurements with living organisms or are implanted within them, the raw signals collected by the sensors are highly complex due to the complex nature of typical biological materials, such as whole blood or tissues. In combination with hardware (e.g., Soc), the various data analysis techniques described herein can be used to retrieve corrected concentration level values ​​from the most complex biological materials. This is particularly important for transdermal / implantation monitoring of important metabolites such as glucose, lactate, urea, ethanol, serum albumin, creatinine, and others, both for subjects impaired by chronic diseases such as diabetes, renal or hepatic dysfunction, as well as acute clinical cases such as sepsis, fitness levels, and dietary monitoring for athletes and the general public.

[0009] Accordingly, in one aspect, a method for calibrating a sensor for measuring the concentration of an analyte in a medium is provided. The method comprises the step of collecting a plurality of raw spectra from an object having an analyte (e.g., a medium or a sample) using a hybrid III-V / IV semiconductor photonics SoC (system-on-a-chip). The method further comprises the step of dividing the raw spectra into sets of clusters according to the individual spectral features of the raw spectra, each cluster comprising a group of raw spectra. Within each cluster, the method further comprises: (i) applying an individual LSC (local scattering correction) to each raw spectra belonging to the cluster to obtain a group of locally corrected spectra; and (ii) deriving a cluster-specific optimization set of preprocessing parameters and a cluster-specific correction vector. The optimization set of preprocessing parameters and the correction vector are derived using gold standard analyte concentration values ​​corresponding to the locally corrected spectra and the group of raw spectra belonging to the cluster.

[0010] In some embodiments, the step of deriving a cluster-specific optimization set and a cluster-specific correction vector of preprocessing parameters for a specific cluster comprises: (i) evaluating each of a plurality of candidate sets of preprocessing parameters, wherein the evaluation of the specific candidate set comprises: (A) preprocessing each locally corrected spectrum belonging to the specific cluster using the specific candidate set; (B) deriving a candidate correction vector by applying multivariate regression correction to the preprocessed locally corrected spectra and using gold standard analyte concentration values ​​corresponding to a group of raw spectra belonging to the specific cluster; and (C) computing a corresponding accuracy measure for the candidate correction vector through cross-validation. Subsequently, the candidate set and the corresponding candidate correction vector associated with the maximum accuracy measure are designated as the cluster-specific optimization set and the cluster-specific correction vector of preprocessing parameters, respectively.

[0011] A cluster-specific optimization set of preprocessing parameters may include a set of data processing parameters, such as a) the order of filtering, b) the alignment or type of filters used for smoothing, c) the order of derivatives used for baseline removal, etc. The optimization set of parameters may be stored in memory and may be subsequently used to preprocess data in detection mode.

[0012] The object may include tissue, and the analyte may include, among others, blood glucose, blood lactate, ethanol, creatinine, keratin, collagen, urea, serum albumin globulin, troponin, acetone, acetate, hydroxybutyrate, cholesterol, albumin, globulin, ketone-acetone, or water.

[0013] In some embodiments, the step of segmenting raw spectra according to individual spectral features of the raw spectra includes the step of applying global scattering correction (GSC) to each of the raw spectra to obtain some globally corrected spectra. The segmenting step may also include the step of clustering some globally corrected spectra according to (A) a specified number of clusters, (B) a specified maximum distance of the globally corrected spectra from the center of the clusters, or (C) both a specified number of clusters and a specified maximum distance of the globally corrected spectra from the center of the clusters. The segmenting step may further include the step of assigning to the cluster, within each cluster, individual raw spectra corresponding to the globally corrected spectra belonging to the cluster. Clustering may include k-means clustering, affinity propagation, or aggregation clustering.

[0014] In some embodiments, the method further comprises the step of storing a GSC reference spectrum generated as part of a global scattering correction in the SoC. The global scattering correction may be implemented as a global multiplicative scattering correction, a global standard normalized variable (SNV) correction, a global mean centering and normalization correction, a Kubelka-Munk (KM) correction, a Saunderson correction, or a combination thereof. Local and / or global scattering correction may incorporate particle-size difference correction and / or path length difference correction, and may utilize KM correction, Saunderson correction, multiplicative scattering correction, or a combination thereof. In some embodiments, the method comprises the step of storing, for each cluster, (i) a corresponding LSC reference spectrum and / or (ii) a corresponding correction vector, (iii) cluster centers, and / or (iv) an optimized set of preprocessing parameters for each cluster in the SoC. Local scattering correction can also be implemented as local multiplicative scattering correction, local standard normal variable (SNV) correction, local mean centering and normalization correction, KM correction, Saunderson correction, or a combination of the aforementioned correction techniques to achieve linearization effects. Global and local scattering corrections, when appropriately selected, allow for the consideration of particle-size difference effects on light scattering as well as correction for optical path difference in tissues to linearize raw spectra, for example, making linear regression techniques including linear Beer-Lambert absorption laws as well as multivariate partial least squares applicable.

[0015] In some embodiments, determining the individual spectral shapes of some raw spectra involves preprocessing the raw spectra by applying linear transformation and baseline correction to the raw spectra based on the reference spectrum of the selected analyte. The preprocessing may include Kubelka-Munk correction, Saunderson correction, multiplicative scattering correction, or a combination of any two or all three correction techniques.

[0016] In another aspect, a method for measuring the concentration of an analyte is provided, wherein the method comprises the steps of acquiring a raw spectrum from an object having an analyte (e.g., a medium or a sample) using a hybrid III-V / IV semiconductor photonics SoC (system-on-a-chip), and identifying a cluster to which the raw spectrum belongs from a plurality of clusters of spectra, wherein the cluster is identified based on the spectral shape of the raw spectrum. The method further comprises the steps of applying local scattering correction (LSC) to the raw spectrum to acquire a locally corrected spectrum, preprocessing the locally corrected spectrum using a cluster-specific optimization set of preprocessing parameters, and multiplying the preprocessed locally corrected spectrum by a cluster-specific correction vector to acquire a corresponding corrected concentration value for the analyte.

[0017] In some embodiments, the step of obtaining a raw spectrum comprises the steps of directing electromagnetic radiation (EMR) tunable to several different wavelengths from the SoC toward an object, measuring the intensities of the EMR received from the object at each of the different wavelengths using the SoC, and converting the intensities into absorbance values, wherein the raw spectrum comprises an absorbance spectrum. Several different wavelengths may be selected from the range of 1000 to 3500 nm or the range of 1900 to 2500 nm.

[0018] In some embodiments, a plurality of clusters of spectra correspond to spectra previously collected using an SoC, and each cluster may be represented by an individual LSC reference, an individual cluster center, and / or an individual correction vector, and the individual LSC reference, the individual cluster center, and the individual correction vector for each cluster may be stored on the SoC. The step of identifying the cluster to which the raw spectra belong from some clusters of spectra may include the step of deriving a globally corrected spectra using a Global Scattering Correction (GSC) reference. The identification of the cluster to which the raw spectra belong may also include, within each of some clusters, the step of comparing the globally corrected spectra with an individual LSC reference to obtain a distance corresponding to the cluster, and the step of selecting the cluster with the minimum corresponding distance.

[0019] Global scattering correction may be implemented as global multiplicative scattering correction, global standard normalized variable (SNV) correction, global mean centering and normalization correction, KM correction, Saunderson correction, or a combination thereof. Local and / or global scattering correction may incorporate particle-size difference correction and / or path length difference correction. Local scattering correction may be implemented as local multiplicative scattering correction, or local standard normalized variable (SNV) correction, or local mean centering and normalization correction, KM correction, Saunderson correction, or a combination thereof. LSC and GSC involve performing linearization transformations on raw spectra to account for tissue / object scattering and absorption to facilitate further data processing based on linear absorption techniques, such as those based on the Beer-Lambert law, where the spectra are decomposed into individual components and / or further processed using PLS linear regression or similar techniques.

[0020] In some embodiments, determining the spectral shape of the raw spectrum involves preprocessing the raw spectrum by applying a linear transformation and baseline correction to the raw spectrum based on the reference spectrum of the selected analyte. The preprocessing may include Kubelka-Munk correction, Saunderson correction, multiplicative scattering correction, or a combination of any two or all three correction techniques.

[0021] In another aspect, a system for measuring the concentration of an analyte comprises a hybrid III-V / IV semiconductor photonics SoC (system-on-a-chip) for acquiring a raw spectrum from an object having the analyte (e.g., a medium or a sample), and a processing unit comprising a processor and memory and configured to perform specific operations for measuring the analyte concentration, storing information, etc. Specifically, the processing unit is configured to use the hybrid III-V / IV semiconductor photonics SoC (system-on-a-chip) to acquire a raw spectrum from an object having the analyte and, based on the spectral shape of the raw spectrum, to identify the cluster to which the raw spectrum belongs from a plurality of clusters of spectra. The processing unit is also configured to apply cluster-specific local scattering correction (LSC) to the raw spectrum to acquire a locally corrected spectrum. The processing unit is further configured to preprocess the locally corrected spectrum using a cluster-specific optimization set of preprocessing parameters and to multiply the preprocessed locally corrected spectrum by a cluster-specific correction vector to obtain a corrected concentration value for the analyte.

[0022] In some embodiments, to obtain a raw spectrum, the SoC is configured to direct electromagnetic radiation (EMR) tunable to several wavelengths toward an object and to measure the intensities of the EMR received from the object at each of the wavelengths. A processing unit is programmed to convert the intensities into absorbance values ​​so that the raw spectrum includes or is represented as an absorbance spectrum. The SoC may be configured to emit EMR at wavelengths in the range of 1900 to 2500 nm or 1000 to 3500 nm.

[0023] Some clusters of spectra can correspond to previously collected spectra using the SoC. Each cluster can be represented by an individual LSC criterion and an individual correction vector. For each cluster, the SoC may include memory for storing the individual LSC criterion and the individual correction vector, as well as the global scattering correction criterion (also referred to as the global scattering correction vector). The memory of the SoC may also store a corresponding optimization set of preprocessing parameters for each cluster.

[0024] In some embodiments, to identify the cluster to which the raw spectrum belongs among several clusters of spectra, the processor is programmed to derive a globally corrected spectrum using a Global Scattering Correction (GSC) criterion stored in memory. The processor may also be programmed, within each cluster, to (i) compare the globally corrected spectrum with an individual LSC criterion to obtain a distance corresponding to the cluster, and (ii) select the cluster with the minimum corresponding distance. Global scattering correction may include global multiplicative scattering correction, or global standard normalized variable (SNV) correction, or global mean centering and normalization correction. Similarly, local scattering correction may include local multiplicative scattering correction, or local standard normalized variable (SNV) correction, or local mean centering and normalization correction. Local and / or global scattering correction may incorporate linearization transformations for particle-size difference correction and / or path length difference correction.

[0025] In some embodiments, the SoC includes a wavelength shift tracker for tracking a shift in the wavelength of radiation emitted by the SoC, and / or a wavelength tracker for tracking the absolute wavelength of radiation emitted by the SoC, and / or a temperature sensor for measuring the temperature of the chip, and / or an SoC output power monitor for monitoring or measuring the intensity of EMR emitted by the SoC during a wavelength sweep to obtain a power curve.

[0026] In some embodiments, to determine the individual spectral shapes of some raw spectra, the processor is configured to preprocess the raw spectra by applying linear transformation and baseline correction to the raw spectra based on the reference spectrum of the selected analyte. To perform the preprocessing, the processor may be configured to apply Kubelka-Munk correction, Saunderson correction, multiplicative scattering correction, or a combination of any two or three correction techniques. Brief explanation of the drawing

[0027] FIG. 1 is a schematic block diagram of a photon SoC arranged for remote sensing of an object according to one embodiment of the present invention. FIG. 2 is a simplified schematic diagram of a photon sensor system arranged for a detection experiment according to one embodiment of the present invention. FIG. 3 is a simplified schematic block diagram of an algorithm used in combination with the hardware illustrated in FIG. 1 and FIG. 2 to generate a calibration algorithm for a sensor according to one embodiment of the present invention. FIG. 4a is a graph of a large set of raw absorbance spectra accumulated from a piglet having a global MSC vector in bold according to one embodiment of the present invention, and FIG. 4b is a graph of spectra after baseline correction (MSC) and before the clustering procedure. Figure 5 is a graph of baseline-corrected spectra from Figure 4b clustered into six different clusters (by definition N = 6) using a k-means algorithm, with the maximum distance to the center within each cluster indicated in the graph. FIG. 6 is a block diagram illustrating schematic diagrams of algorithms for configuring individual correction models according to embodiments of the present invention. Figure 7a is a graph of raw spectra collected from an object (piglet) within a cluster before applying MSC baseline correction. The thick black spectrum is the calculated local MSC reference, and Figure 7b is a graph of spectra within the same cluster after baseline correction (MSC). Figure 8 is a graph of individual component concentration correction vectors obtained for glucose molecules using a transdermal diffuse reflectance sensing geometry with a piglet. FIG. 9 is a schematic block diagram of a sensing algorithm used in combination with a hybrid III-V / IV semiconductor photonics sensor SoC according to one embodiment of the present invention. FIG. 10 is a graph of the performance of a photon sensor system-on-chip transcutaneous blood glucose detection using the data processing method of one embodiment of the present invention in the case of a calmed Piglet. FIG. 11 is a graph of the performance of a photon sensor system-on-chip transdermal blood lactate detection using the data processing method of one embodiment of the present invention in the case of a calmed Piglet. FIG. 12 is a graph of the performance of a photon sensor system-on-chip transdermal blood ethanol detection using the data processing method of one embodiment of the present invention in the case of a calmed piglet. Figure 13 is a graph illustrating the effect of Kubelka-Munk pretreatment on the transdermal tissue spectrum during the analysis of ethanol. Figure 14a illustrates the decomposition of an observed transdermal signal using a Bitt-Lambert model without preprocessing the observed signal using Kubelka-Munk correction. Figure 14b illustrates the decomposition of the same observed dermal signal after preprocessing the observed signal using the same Bitt-Lambert model but with Kubelka-Munk correction. Specific details for implementing the invention

[0028] Optical remote sensing is a technology developed for a wide range of applications. Sensing can be performed in the form of distance measurement, namely by measuring distance using time-of-flight or frequency-modulated continuous wave (FMCW) technology, or by spectroscopic sensing to remotely detect, identify, and quantify the presence or absence of one or more molecules within an object.

[0029] The term spectroscopic sensing as used herein refers to the arrangement of a hybrid III-V / IV semiconductor photon P-SoC (photonic system-on-a-chip) that emits wavelength-tunable laser radiation and communicates with a remote target object. Wavelength shifts and absolute values ​​are monitored and considered within every sweep, and the SoC is automatically calibrated in terms of absolute wavelength, wavelength shift, and power spectrum.

[0030] Light strikes an object and penetrates to a specific depth defined by an optical length, which depends on the individual characteristics of the object, such as the scattering matrix and content. For example, when using laser radiation tunable in the 1900 to 2500 nm spectral region to perform transdermal sensing experiments on living organisms, the light penetrates up to approximately 1 mm below the skin surface, where it is scattered and partially absorbed by tissues, blood, and interstitial fluid. This absorption is molecule-specific, and each constituent molecule modifies the light spectrum with a unique spectral absorption signature. After interaction with the object, the transmitted, scattered, or reflected light is collected and detected by a photodetector.

[0031] A schematic block diagram illustrating an embodiment of the present invention is shown in FIG. 1. Here, the photon system-on-chip comprises a hybrid III-V / IV semiconductor chip (1) forming the hardware portion of a photon sensor chip and control and signal processing electronics (2). The photon sensor on the chip optically communicates with an object (3), and the object (3) may be a living organism, an isolated substance, etc. In the configuration, the photon system-on-chip is remote from the object.

[0032] In the illustrated embodiment, the hybrid III-V / IV semiconductor chip (1) includes a hybrid III-V / IV external cavity laser (100) that emits a sweeping wavelength laser radiation through an optical path (10). A portion of the beam is split through path (11) and supplied to a wavelength shift tracker (120) through path (11), an absolute wavelength reference (130) through an optical path (14), a laser power curve monitoring block (140) through an optical path (17), and an output section through an optical path (19). The chip (1) may also include a temperature sensor (110) for detecting the temperature of the chip, and the temperature of the chip may eventually be used for absolute wavelength reference calibration.

[0033] The wavelength shift tracker (120) may be any type of unbalanced interferometer, such as a Mach-Zender, Michelson, or Fabry-Perot. The unbalanced interferometer provides a bit signal at the output of 120 through the optical path (12), and the photodetector block (121) registers an oscillation signal, where the oscillation period depends on the optical path difference and wavelength within the interferometer. The optical path difference is defined by design and is a known parameter. Thus, if the absolute value of the wavelength at any given moment is known, the wavelength shift value can be extracted. This is provided by an absolute wavelength reference block (130) coupled to a monitoring photodetector (131) through the optical path (15). The absolute wavelength reference may be a distributed Bragg grating (DBR), a micro-ring resonator (MRR), a distributed feedback grating (DFB), or any other optical cavity structure having distinct characteristic transmission or reflectance features within the spectral region covered by the hybrid laser (100) sweep. In this way, the photodetector blocks (121 and 131) cooperate to provide information regarding the wavelength shift value and the absolute wavelength value at any given moment within the sweep.

[0034] Tracking wavelength shifts and absolute wavelength values ​​is often necessary to decouple system effects from object-related effects. For example, emission wavelengths can vary nonlinearly on the system side; therefore, without accurate knowledge of absolute wavelength shift and value information, it can be difficult to perform signal conversion from the time domain to the wavelength (or frequency domain). Another aspect is that collected spectra will be altered by changes on the object side, such as water displacement due to temperature or changes in other strong baseline contributors. Without always knowing the system output, it is impossible to decouple whether the spectrum collected from the object is shifted by changes in the system output or influenced by changes within the object. Therefore, tracking wavelength shifts and absolute wavelength information within every sweep enables the decoupler of system-specific modulations in the collected spectra from object-specific modulations, which are useful signals.

[0035] In practical cases, target molecules, such as glucose, lactate, and ethanol, have very small concentrations compared to major baseline contributors—namely, major proteins (collagen, albumin, keratin) and water—in the case of transdermal sensing. These major contributors provide signals more than 10,000 times stronger than the target molecules; therefore, small changes in water displacement due to temperature effects can lead to baseline changes, which, if unnoticed, can eliminate any useful signal that could be contributed to by glucose. Thus, the ability to track wavelength shifts and absolute values ​​within every sweep allows access to tracking baseline changes within every sweep.

[0036] While wavelength shifts can be monitored as beat signals during a sweep, absolute values ​​are measured once per sweep, and information from both wavelength shifts and absolute wavelengths is used to correct recorded information immediately after the sweep is completed. The accuracy for determining wavelength shifts ultimately depends on system design, such as optical path differences within the wavelength shift tracer that provides the beat signals. In practical scenarios, this depends on the finesse of the absorption characteristics of the target molecular species within the object. When the object is a biological material and the molecules exhibit a liquid phase (characterized by very broad spectral signatures), wavelength shift tracers can have an accuracy of 0.1 nm to several nm, with 3 to 5 nm being typical values.

[0037] In the case of gas detection, if the absorption line width of interest can be in the range of 100 MHz, wavelength shift tracking needs to be designed to have better resolution, and the absolute wavelength reference needs to be designed to provide an absolute wavelength with sufficiently high resolution. In practical cases, this can be achieved with very good accuracy. For example, conventional group IV semiconductor manufacturing techniques rely on node sizes as low as 160 nm or even down to 7 nm, which is 1,000 times larger than the typical emission wavelength. The time duration for a single sweep is defined by the system architecture and lasts from minutes when the tuning mechanism is performed by the mechanical motion of the tuning element, to several microseconds when tuning is electronic. In practical cases for hybrid III-V / IV sensor chips, the sweep rate can range from tens of Hz to a maximum of MHz, depending on actual practical system design and application requirements.

[0038] Depending on the requirements for sensor design and spectral bandwidth coverage, a single sweep may include tens to hundreds of discrete wavelengths. Typical practical cases for transdermal glucose sensing require about 100 or more discrete wavelengths to perform accurate predictions. Based on existing, state-of-the-art, extensively tunable (swept-wavelength) laser concepts, the sweep may be nearly continuous when a Vernier filter is operated in combination with phase control. In some embodiments, the absolute value of the emission wavelength is tuned within a specified range, e.g., 1000 to 3000 nm, 1900 nm to 2500 nm, etc. Thus, the tuned value of the emission wavelength at a specific time may be 1898 nm, 1905 nm, etc. The corresponding wavelength shift may be 1 nm, 2 nm, 10 nm, etc.

[0039] EMR received from the medium of interest is converted into electrical signals from the optical domain within photodetectors (121, 131), and the electrical signals from the photodetectors are routed to an electrical path (30) through electrical paths (13, 16), and the electrical path (30) is connected to a driving and control electronic device block (2) and an analog-to-digital converter (ADC) and amplifier block (210) within it. Here, an analog signal from the photonic chip is amplified and digitized. The digitized signal is supplied to a CPU (220), which performs signal filtering, averaging, and other processing. The CPU (220) includes a memory block having a calibration model. This calibration model is applied to the collected data to retrieve a calibrated concentration level value, and then the calibrated concentration level value is supplied to an output port, e.g., a display (240), through an electrical path (39). Another function of the CPU (220) is to provide control signals to the driver and digital-to-analog converter (DAC) block (230) via path (38), which ultimately provides control and driving signals to the SoC via path (40). The entire sensor system is powered by the power supply unit (200) via electric buses (31, 32, 33, 34, 35, 36).

[0040] A simplified version of the sensor system of FIG. 1 is provided in FIG. 2, having several internal blocks such as an analog-to-digital converter (ADC) block (210), a wavelength shift tracker photodetector block (121), an absolute wavelength reference photodetector (131), a laser power curve photodetector (140), a signal photodetector (150), and a CPU (220), which are highlighted separately for clarity.

[0041] When deployed in the field, the photon sensor on the chip (1) transmits a wavelength-tunable signal to a remote object (3) via an optical path (20). The signal strength I can be expressed as any function of frequency ω (or wavelength) and time t:

[0042] (1)

[0043] Light interacts with the object (3) and undergoes multiple scattering and absorption events within the object. A portion of the scattered and diffusely reflected light is collected by a signal photodetector (150) via an optical path (21). This light signal intensity can be expressed as a function of frequency and time I':

[0044] (2)

[0045] This signal is modulated due to interaction with the object and conveys object-specific information, such as the concentration levels of constituent elements. The latter are individual absorbances A i It can be evaluated as absorption A, which can be expressed as a linear superposition of:

[0046] (3)

[0047] Here, ε(ω) i is the frequency-dependent individual molar absorption rate of component i, and c i is the individual molar concentration of component i, and l is the effective optical length within the object.

[0048] In practical cases where the object is a living organism, individual absorption contributions can be expressed as contributions by different constituent elements, such as 1-keratin, 2-glucose, 3-lactate, 4-urea, 5-collagen, etc. This provides a path for the elemental decomposition of the complex matrix and thus provides the possibility for detection. A procedure for collecting and processing data and deriving corrected concentration values ​​is illustrated in the form of block diagrams in FIGS. 3, 6, and 9.

[0049] The basic operation method for performing detection first includes the step of using a calibration algorithm in combination with hardware to create a calibration model and store it in the CPU's memory. This model can be considered universal and can be deployed with all sensors in the field without the need to modify the sensors during use. Subsequently, the next step is to use the detection algorithm according to FIG. 9 in combination with the calibration model and hardware stored in system memory or the CPU.

[0050] According to one embodiment of the present invention, when arranged in a sensing configuration, the photon system-on-chip provides several output channels containing information regarding the state of the photon chip, such as a wavelength shift value through a photodetector (121), an absolute wavelength reference value through a photodetector (131), a laser intensity curve through a laser power curve monitoring block (140), and / or a reflected signal containing object-specific information through a signal photodetector (150). These electrical signals are routed to a control and signal processing electronics block (2). Here, the signals are supplied to an analog-to-digital converter and amplifier block (210).

[0051] System calibration for analyte measurement

[0052] An algorithm for processing acquired analog signals received from the photonic SoC (1) begins by amplifying and digitizing the signals received from the ADC and amplifier block (210). In this phase, the signals are still processed as time-domain signals. Subsequently, these amplified and digitized signals are fed to a central processing unit (CPU) (220), where the object-specific signal (22) is processed and converted from the time domain to the frequency domain using information on wavelength shift received through the electrical path (13) and absolute wavelength correction received through the electrical path (16), and normalized with respect to the laser power curve received through the electrical path (18). This procedure first enables having signals in the frequency domain to further process the signals that primarily carry object-specific data, as indicated by step (2210) in FIG. 3, and also resolves system-related nonlinearities.

[0053] To reduce noise, multiple spectra are collected, averaged, and filtered. For example, in FIG. 4a, each individual curve represents an averaged spectrum. Subsequently, the corrected intensity is converted into absorbance according to Equation (3) in step (2220), and a large number of raw absorbance spectra are accumulated as indicated in step (2230). The raw spectra typically have highly diverse spectral shapes due to different tissue physiologies (e.g., from different tissue samples having different scattering particle sizes, etc.) and intensities due to differences in optical path length and / or particle size. To correct for scattering effects, Multiplicative Scattering Correction (MSC) is applied to the raw absorbance data in step 2240, and in step (2250), a global MSC reference file (or average spectrum) (indicated by a thick line roughly in the center of the graph in FIG. 4a) is extracted and then stored in system memory. This global MSC spectrum is subsequently used to assign raw data to the right cluster based on exactly the same baseline correction procedure. Then, all baseline-corrected data within step (2240) (see FIG. 4b) (e.g., after MSC) is grouped into clusters based on spectral shape similarity in step (2260). Other types of scattering correction techniques, such as standard normal variable (SNV) correction, Kubelka-Munk correction, Saunderson correction, or mean centering and normalization correction, may be applied as alternatives to multiplicative scattering correction.

[0054] Referring to Fig. 5, baseline-corrected spectra from Fig. 4b are clustered into six different clusters (by definition, N=6) using a k-means algorithm. The maximum distance to the center within each cluster is shown in the graph.

[0055] As illustrated, the global MSC-corrected data is used solely to assign raw spectra to each cluster. Thus, the assigned clusters contain raw or unprocessed data. Clustering can be performed in a number of ways. Two possible paths are illustrated in FIG. 3. In the first path, i.e., step (2270), spectra are grouped into a fixed and defined number (N) of clusters based on spectral shape similarities. A disadvantage of this approach is that the cluster centers σ assigned from the errors or spectra k The distance to (where σ is an array of cluster centers and k is a cluster number) can vary significantly between different clusters (see FIG. 5). In the potentially better path illustrated as step (2275), clustering is performed by defining the maximum distance from any spectra to the cluster center, resulting in an arbitrary number of clusters that may be large in practical cases. Thus, an intermediate route indicated by 2276 may be used, where both the number of clusters N and the maximum allowable distance to the cluster centers are defined. In such cases, spectra that satisfy the distance to the center criteria within the defined number of clusters are considered outliers and are not used in data analysis. The predefined number of clusters can be arbitrarily large, but in practical cases, it may be between 10 and 50 depending on the analyte and detection geometry. A set of cluster centers is stored in CPU memory, which will subsequently be used by the detection function to assign GS-corrected spectra.

[0056] Once clustering is complete, individual calibration models within each cluster are generated in step (2280). The individual calibration models assign calibrated concentration level values ​​to all spectra within each cluster as measured by the gold standard as indicated. Subsequently, this set of calibration models is stored in CPU memory after the MSC reference vector in step (2300).

[0057] An algorithm for constructing individual correction models (2280) is illustrated in FIG. 6. Referring also to FIG. 7a and FIG. 7b, according to one embodiment of the present invention, the step (2280) of constructing individual correction models within each cluster begins by applying scattering correction (e.g., MSC) to the raw spectra within the cluster. This calculates individual local MSC criteria for each cluster, which are stored in CPU memory (see bold line in FIG. 7a). These local criteria are used to process raw data acquired in detection mode. Other types of scattering correction techniques, such as standard normal variable (SNV) correction, Kubelka-Munk correction, Saunderson correction, or mean centering and normalization correction, may be applied as alternatives to multiplicative scattering correction.

[0058] Next, the local criteria from 2281 are used to construct a partial-least-square (PLS) model within each cluster and to obtain optimal model parameters, such as noise filtering parameters, derivative order, and the number of PLS ​​latent vectors, using the cross-validation method in step (2282). This operation yields an optimal set of data preprocessing parameters (2283), which are then applied to all clusters containing raw spectra to construct individual calibration models (2284). In other words, within each cluster, the raw spectra are modified using local scattering correction criteria. This ensures that all data are processed in the same way using the same set of parameters. Subsequently, the calibration model assigns the calibrated concentration level(s) of the analyte(s) of interest, measured by a selected reference technique (also referred to as the gold standard), to each locally calibrated spectrum. The calibration model maps the absorbance represented by the spectrum at a specific wavelength to the analyte concentration level. Referring to Fig. 8, the acquired individual correction vectors are subsequently stored in CPU memory. The correction vectors are the outputs of the multivariate regression correction. After training the model using all spectra within the cluster, the model determines weights for the locally corrected and preprocessed absorbance spectrum values ​​at each wavelength. In the prediction, every i-th wavelength value of the preprocessed absorbance is multiplied by the corresponding weight and then summed across all wavelengths to obtain the following predicted concentrations:

[0059] , where n is the number of wavelengths in the spectrum. In some cases, when the sample is associated with a relatively simple scattering matrix and contains fewer components, reasonable concentration predictions can be simply obtained by preprocessing the spectral data acquired from the sample to correct for nonlinear effects of scattering using Kubelka-Munk correction, MSC, Saunderson correction, or a combination thereof, and then removing the baseline to obtain the spectrum of the component of interest. For higher accuracy, and especially for more complex samples such as biological tissues, scattering correction (or linearization transformation) can be used in combination with multivariate linear regression such as PLS.

[0060] Generally, during calibration, the EMR is directed toward a sample (also referred to as the medium), where it is swept through a range of wavelengths. In response, the EMR is received from the sample, where the received EMR is diffusely reflected by the sample or transmitted through the sample. The received EMR, having components of different wavelengths, is converted into a raw absorption spectrum (also referred to as the raw spectrum). This process can be repeated multiple times to obtain multiple raw spectra, which are then averaged to obtain an averaged raw spectrum. For simplification, the term "averaged" is omitted in the following discussion. These raw spectra are It may be represented as, where index i represents individual averaged raw samples and may be in the range of 1 to M, where M may be any number such as 50; 100; 2000; 10,000 or greater. The aforementioned process is repeated at different times, where the analyte concentration in the sample may differ at different times using different zones of samples or different samples, where the analyte concentration may differ in different zones of the same sample or in different samples.

[0061] Next, Global reference and globally corrected spectra indicated by To obtain, scattering corrections (MSC, Gubelka-Munk correction, Saunderson correction, etc.) are applied to the raw spectra Applies to. Global criteria is stored in memory. Subsequently, globally corrected spectra to identify N clusters Clustering is performed using . The number N (e.g., 4, 5, 6, 10, etc.) can be specified for the clustering behavior, or alternatively, the clustering itself can determine the best N. Each Regarding, the corresponding cluster This is identified, and then, the corresponding raw spectrum is designated as the same cluster. After clustering, the optimal number of clusters, cluster centers, and maximum allowable distance to cluster centers are stored in memory to be used for the detection function.

[0062] When all raw spectra are assigned to their individual clusters, the aforementioned process is repeated within each cluster. Specifically, stored in memory, To obtain the local standard denoted by, raw spectra within a specific cluster k Scattering correction is applied to. Raw spectra within cluster k By locally applying scattering correction to, locally corrected spectra for cluster k is generated. This process is repeated for all clusters, and individual local criteria and each Locally corrected spectra for Acquires.

[0063] Different raw spectrum It is noted that it can correspond to different levels of analyte concentrations. These concentration levels, indicated by , are obtained from the samples using the selected gold standard technique. Finally, a correction vector for each cluster k is obtained through multivariate linear regression correction. is generated. The correction vector used to generate the correction vector. , local standards , and the data preprocessing set can be stored in the SoC's memory module for each cluster. The data preprocessing set defines whether the calibration vector was acquired using absorbance, absorbance processed by the n-th derivative, filtering order, Kubelka-Munk correction, Saunderson correction, multiplicative scattering correction, etc. This is essential to ensure that all raw data is processed in exactly the same way when the sensor is deployed for detection. Global criterion It can also be stored in the memory module of the SoC.

[0064] One exemplary process for obtaining an optimal data preprocessing set is as follows:

[0065] 1. Within the cluster, signal smoothing (noise filtering) is applied to the locally corrected spectra using an iteratively selected filter and its degree (e.g., Savitzky-Golay, Fourier transform filter, percentile, moving average). Additionally, first or second derivative baseline removal may also be applied.

[0066] 2. Locally corrected and pre-processed spectra and corresponding concentrations are randomly divided into training and test sets.

[0067] 3. A multivariate regression correction algorithm is applied to the training set, and after the model is trained, the concentration is predicted using the test set, and the prediction accuracy is evaluated.

[0068] 4. Steps 2 and 3 are repeated multiple times in a process called cross-validation to obtain the average prediction accuracy for the current data preprocessing set (e.g., repeated n times).

[0069] Steps 1-4 can be repeated with different sets of parameters selected in Step 1. The optimal set of parameters is the set that resulted in the best average prediction accuracy.

[0070] Multivariate regression algorithms model the relationship between predictors and response variables. Therefore, the correction spectrum matrix can be considered as a predictor, where d is the number of wavelengths and the analyte concentration vector is considered as a response. Each i-th row of the spectrum matrix corresponds to a locally corrected and preprocessed spectrum (e.g., a Savitzky-Golay filter and second derivative applied to a locally corrected absorbance spectrum), and each i-th row of the response vector corresponds to an analyte concentration measured by a gold standard. Once the relationship between the predictor and the response is determined, an unknown value of the analyte concentration can be predicted based on the new locally corrected and preprocessed spectrum. Multivariate regression may include partial least squares regression and modifications thereof, multiple linear regression, support vector regression, artificial neural networks, and / or principal component regression.

[0071] Detection or analysis measurement

[0072] Referring to FIG. 9, individual calibration vectors, a global MSC reference, and local MSC vectors stored in memory enable the sensing algorithm to perform the sensing function of the hybrid photonic SoC. In particular, when deployed in the field, the hybrid III-V / IV photonic SoC collects diffuse reflectance signals, which are then amplified and digitized within the ADC+amp section (210) along with absolute wavelength references, wavelength shift values, and laser power curve signals. Subsequently, in step (2210), the time-domain signals are converted to the frequency domain and averaged and calibrated in terms of absolute wavelength, wavelength shift, chip temperature, and laser power curve. Next, the reflected intensity is converted to absorption in step (2220).

[0073] Next, the collected absorbance spectra undergo baseline correction using a global scattering correction GSC reference taken from CPU memory in step (2221) to initiate a clustering procedure. To cluster the collected spectra, cluster centers and maximum allowable distances to the cluster centers are provided from CPU memory, and the data is classified accordingly in step (2223). If the provided distance to the cluster centers exceeds the maximum allowable distance, in step (2224), the CPU initiates an error message to indicate to the user to adjust the sensor position and restart data collection until the error is no longer greater than the maximum allowable distance. After baseline correction in step (2225), if the collected data has a distance to the cluster centers within an allowable range, in step (2226), the corresponding collected raw spectrum is assigned to a cluster having the minimum distance to the center.

[0074] Next, in step (2227), the raw spectrum within the newly assigned cluster undergoes baseline correction using a local scattering correction criterion from CPU memory, and the data is preprocessed using a data processing set from CPU memory in step (2228) to qualify for the data prediction step (2229), wherein individual correction vectors from CPU memory obtained by multivariate regression correction are It is multiplied by. Multiplying the row vector of the spectrum by the column vector of regression weights yields a single value for the analyte concentration. Each different analyte will have a different calibration vector and corresponding weights, that is, different wavelength specificity for a specific analyte. For example, 2100 nm may be relevant to both lactate and glucose, but the weights will be different. The analyte concentration is is. Here, w n is the correction weight at the nth wavelength, and A n is the locally corrected and pre-treated absorbance at the nth wavelength. Subsequently, the output is the corrected concentration level of the analyte of interest.

[0075] Generally, the detection process begins in a manner similar to the calibration process. Specifically, the EMR is directed toward a sample (also referred to as the medium) in which the analyte concentration is to be determined. The EMR is swept across a range of wavelengths. In response, the EMR is received from the sample, where the received EMR is either diffusely reflected by the sample or transmitted through the sample. The received EMR, having components of different wavelengths, is converted into a raw absorption spectrum (also referred to as the raw spectrum). This process can be repeated multiple times to acquire multiple raw spectra, and subsequently, these It is averaged to obtain the averaged raw spectrum indicated by . Here again, for the sake of simplification in the discussion below, the term "averaged" is omitted.

[0076] Next, the globally corrected spectrum To obtain (generated during the calibration process) Raw spectrum using the global standard indicated by Scattering correction is applied to it. Subsequently, clustering is performed using the cluster center value σ k and is performed using the distance from memory to the center value. The cluster is It can be displayed as, where And, the number N was specified for the clustering operation, or alternatively, determined while performing clustering as part of the calibration process. Subsequently, the corresponding raw spectrum is the same cluster It is designated as.

[0077] After that, Selected clusters using the corresponding local criteria indicated by . Internal raw spectrum Scattering correction is reapplied to the cluster. Internal raw spectrum By locally applying the data preprocessing parameters set and scattering correction, the locally corrected and preprocessed spectrum is created. Selected cluster Spectrum for and correction vectors Using the absorbance values, the concentration level of the analyte of interest is estimated. This entire process can be repeated multiple times to obtain several estimates of the analyte concentration to provide an averaged estimated analyte concentration.

[0078] Examples of the performance of Piglet's transdermal sensor for three different analytes, namely blood glucose, blood lactate, and blood ethanol, according to one embodiment of the present invention, are provided in FIGS. 10 to 12. Here, for all experiments, a sow weighing approximately 40 kg was sedated for a duration of 8 hours, and a buffered analyte solution, a glucose solution, was injected intravenously to raise the pig's blood analyte levels. In the case of glucose in FIG. 10, blood glucose levels were raised by injecting the buffered glucose solution, and insulin was administered to lower blood glucose levels. In the case of lactate in FIG. 11, blood lactate levels were raised by intravenous injection and lowered by terminating the administration of the buffered lactate, allowing the pig to naturally eliminate the lactate levels. In the case of ethanol, again, blood ethanol levels were raised by injecting the buffered solution intravenously and lowered by terminating and allowing the body to naturally eliminate the ethanol. In all cases, the III-V / IV sensor was in contact with the skin of the pig on the abdomen of the sedated pig. The sensor was sampling the pig at a frequency of 40 Hz (40 sweeps / sec or 40 spectra / sec). Blood samples were drawn from the pig's artery every 6 minutes and analyzed using a clinical analyzer used as a gold standard. In the described embodiment, as clinical gold standards, two Abaxis Piccolo Xpress analyzers for blood glucose calibration, an EKF Biosen C-line analyzer for lactate calibration, and an Agilent 8860 gas chromatographer for blood ethanol calibration were used. Subsequently, the collected spectra were assigned to the calibrated glucose concentration level values ​​as measured by the gold standard, and the data were processed according to the procedure described in the embodiment of the present invention.

[0079] In FIG. 10, data points (1002) represent data points used to form a calibration model, and red data points (1004) represent multiple predictions using that model for a specific pig under study. In this case, the model and validation use data obtained from the same pig. During a 1-day period, the pig's blood glucose level was ramped up and down, and calibration data was measured every 6 minutes using a gold standard. Optical spectra collected between two calibration points were interpolated, and absolute glucose concentration values ​​were assigned.

[0080] Representative results demonstrate excellent sensor performance with a wide dynamic glucose concentration level range of 75 mg / dl (4.16 mmol / l) to 400 mg / dl (22.2 mmol / l), a coefficient of determination of 97.2%, a root mean square error of prediction (RMSEP) of 14.7 mg / dl (or 0.8 mmol / l), and a mean absolute relative difference of 6.7% across the entire range.

[0081] In FIG. 11, the green data points (1006) represent the data points used to form the calibration model, and the red data points (1008) represent multiple predictions using that model for a specific pig under study. In this case, the model and validation use data from the same pig. Representative results demonstrate transcutaneous blood lactate detection in the concentration level range of 1 mmol / l to 15 mmol / l with a coefficient of determination of 92.4% and an RMSEP of 0.954 mmol / l.

[0082] In FIG. 12, the green data points (1010) represent the data points used to form the calibration model, and the red data points (1012) represent multiple predictions using that model for a specific pig under study. In this case, the model and validation use data from the same pig. Representative results demonstrate transcutaneous blood ethanol detection in the concentration level range of 0.2 ‰ - 4.2 ‰ with a coefficient of determination of 96.4% and an RMSEP of 0.217 ‰.

[0083] In FIGS. 13 and 14, the effect of data preprocessing / correction is highlighted. In FIG. 13, a typical experimental raw absorption spectrum (1300) collected from a perfused pig ear based on diffuse reflectance is shown. The spectrum includes signals from tissue, skin, its components (collagen, water, etc.), and the perfusion solution (in this particular case, a 2% aqueous ethanol solution). In this experiment, ethanol is the analyte of interest. The solution is injected into the artery of the ear and collected back through the vein. A sensor is attached to the surface of the skin of the ear and collects the diffuse reflectance of the tissue as well as the perfusion solution.

[0084] Due to the nonlinear nature of diffuse reflection, one of the critical steps in data preprocessing is the linearization and scattering correction of the collected spectrum. When applied correctly, this enables further data processing, such as Beer-Lambert-based analysis, where the linearized and corrected spectrum is decomposed into individual components. This subsequent analysis can be performed in combination with other linear regression techniques to obtain corrected values ​​for the concentration levels of the components / analytes of interest.

[0085] In FIG. 13, Kubelka-Munk linearization is performed when decomposing the raw spectrum (1300), and ethanol can be isolated / decomposed from the observed transdermal spectrum (1500) by using the pure ethanol absorption spectrum (1400) obtained from calibrated transmission measurements, which is also referred to as the reference spectrum of the selected analyte. Since no additional processing was performed while noise was present, the isolated spectrum (1500) shows three ethanol-specific peaks.

[0086] As illustrated in FIGS. 14a and 14b, further processing of the isolated spectrum may be performed. Here, 24-hour perfusion cycles of different ethanol concentrations ranging from 0.1% to 2% were performed. Control flow cuvettes for arterial input and venous output were used to monitor the concentrations of the perfused solution and their stability, and are illustrated as reference cuvette signals (1600). Detection was performed transdermally based on diffuse reflectance geometry.

[0087] In FIG. 14a, the acquired raw spectrum (1300) was processed directly by applying -In(x) to fit a Beer-Lambert model without any linearization transformation / correction. The components used for fitting included water, skin, ethanol, fat, slope, path lengths, and offset; thus, the spectrum was decomposed into water, skin, fat, ethanol, slope, path lengths, and offset. The resulting fitting was compared with a control cuvette measurement, i.e., a reference cuvette signal (1600). It can be seen that while there is some correlation between the ethanol trace (1700a) and the reference trend (1600), this is mostly indecisive and does not provide reliable readings for sensing applications.

[0088] In FIG. 14b, the same diffuse reflection spectrum was processed for linearization and scattering correction using a Beer-Lambert approximation (decomposition and fitting of individual components) following the Kubelka-Munk correction. In this case, the extracted transdermal ethanol trace (1700b) matches the reference cuvette signal (1600) well over the entire 0.1% - 2% range, including profiles that increase / decrease sharply.

[0089] The described embodiments of the present invention are intended merely as exemplary, and numerous variations and modifications are intended to be within the scope of the invention as defined in the appended claims.

Claims

Claim 1 A method for calibrating a sensor for measuring the concentration of an analyte, comprising: collecting a plurality of raw spectra from an object having said analyte using a hybrid III-V / IV semiconductor photonics SoC (system-on-a-chip); dividing said raw spectra into a set of clusters according to individual spectral shapes of said raw spectra, wherein each cluster comprises a group of raw spectra; and within each cluster: applying an individual LSC (local scattering correction) to each raw spectra belonging to said cluster to obtain a group of locally corrected spectra; The method comprises the step of deriving a cluster-specific optimized set and a cluster-specific calibration vector of preprocessing parameters using gold standard analyte concentration values ​​corresponding to the locally corrected spectra and the group of raw spectra belonging to the cluster, wherein the step of deriving the cluster-specific optimized set and the cluster-specific calibration vector of preprocessing parameters for a specific cluster comprises the step of evaluating each of a plurality of candidate sets of preprocessing parameters, wherein the evaluation of the specific candidate set comprises: the step of preprocessing each locally corrected spectra belonging to the specific cluster using the specific candidate set; the step of deriving a candidate calibration vector by applying multivariate regression correction to the preprocessed locally corrected spectra and using gold standard analyte concentration values ​​corresponding to the group of raw spectra belonging to the specific cluster; and the step of computing a corresponding accuracy measure for the candidate calibration vector through cross-validation.A method comprising the step of designating the corresponding candidate correction vector associated with the candidate set and the maximum accuracy measure as the cluster-specific optimization set and cluster-specific correction vector of the preprocessing parameters, respectively. Claim 2 A method according to claim 1, wherein the object comprises a tissue; and the analyte comprises at least one of blood glucose, blood lactate, ethanol, urea, creatinine, troponin, cholesterol, albumin, globulin, ketones—acetone, acetate, hydroxybutyrate, collagen, keratin, and water. Claim 3 A method for calibrating a sensor for measuring the concentration of an analyte, comprising: collecting a plurality of raw spectra from an object having said analyte using a hybrid III-V / IV semiconductor photonics SoC (system-on-a-chip); dividing said raw spectra into a set of clusters according to individual spectral shapes of said raw spectra, wherein each cluster comprises a group of raw spectra; and within each cluster: applying an individual LSC (local scattering correction) to each raw spectra belonging to said cluster to obtain a group of locally corrected spectra; The method comprises the step of deriving a cluster-specific optimized set and a cluster-specific calibration vector of preprocessing parameters using gold standard analyte concentration values ​​corresponding to the locally corrected spectra and the group of raw spectra belonging to the cluster, and the step of dividing the plurality of raw spectra according to the individual spectral shapes of the plurality of raw spectra includes: applying a global scattering correction (GSC) to each of the plurality of raw spectra to obtain a plurality of globally corrected spectra; and clustering the plurality of globally corrected spectra according to (A) a specified number of clusters, or (B) a specified maximum distance from the center of the cluster to the globally corrected spectra, or (C) both a specified number of clusters and a specified maximum distance from the center of the cluster to the globally corrected spectra.A method comprising the step of, within each cluster, assigning to the cluster an individual raw spectrum corresponding to a globally corrected spectrum belonging to the cluster. Claim 4 A method according to claim 3, wherein the clustering step comprises at least one of k-means clustering, affinity propagation, and agglomerative clustering. Claim 5 A method for calibrating a sensor for measuring the concentration of an analyte, comprising: collecting a plurality of raw spectra from an object having said analyte using a hybrid III-V / IV semiconductor photonics SoC (system-on-a-chip); dividing said raw spectra into a set of clusters according to individual spectral shapes of said raw spectra, wherein each cluster comprises a group of raw spectra; and within each cluster: applying an individual LSC (local scattering correction) to each raw spectra belonging to said cluster to obtain a group of locally corrected spectra; A method comprising the step of deriving a cluster-specific optimized set and a cluster-specific calibration vector of preprocessing parameters using gold standard analyte concentration values ​​corresponding to the locally corrected spectra and the group of raw spectra belonging to the cluster, and further comprising the step of storing a GSC reference spectrum in the SoC after the step of dividing the plurality of raw spectra into sets of clusters according to the individual spectral shapes of the plurality of raw spectra. Claim 6 A method according to claim 3, wherein the global scattering correction comprises global multiplicative scattering correction, global standard normal variate (SNV) correction, Kubelka-Munk correction, Saunderson correction, or global mean centering and normalization correction. Claim 7 A method according to paragraph 3, wherein the local or global scattering correction comprises a particle-size difference correction or a path length difference correction, and said particle-size difference correction or said path length correction comprises a Kubelka-Munk correction, a Saunderson correction, a multiplicative scattering correction, or a combination thereof. Claim 8 A method according to claim 1, further comprising the step of storing, for each cluster, (i) a corresponding LSC reference spectrum, (ii) a corresponding correction vector, and (iii) a cluster center in the SoC. Claim 9 A method according to claim 8, further comprising, for each cluster, (iv) storing a cluster-specific optimization set of the preprocessing parameters in the SoC. Claim 10 A method according to claim 1, further comprising the step of storing an optimized set of preprocessing parameters for each cluster in the SoC. Claim 11 A method according to claim 1, wherein the local scattering correction comprises local multiplicative scattering correction, local standard normal variable (SNV) correction, Kubelka-Munk correction, Saunderson correction, or local mean centering and normalization correction. Claim 12 A method according to claim 1, wherein determining the individual spectral shapes of the plurality of raw spectra comprises preprocessing the plurality of raw spectra by applying linear transformation and baseline correction to the plurality of raw spectra based on a reference spectrum of a selected analyte. Claim 13 A method according to claim 12, wherein the preprocessing comprises Kubelka-Munk correction, Saunderson correction, multiplicative scattering correction, or a combination thereof. Claim 14 A method for measuring the concentration of an analyte, comprising: acquiring a raw spectrum from an object having said analyte using a hybrid III-V / IV semiconductor photonics SoC (system-on-a-chip); identifying a cluster to which said raw spectrum belongs from a plurality of clusters of spectra based on the spectral shape of said raw spectrum; applying local scattering correction (LSC) to said raw spectrum to acquire a locally corrected spectrum; preprocessing said locally corrected spectrum using a cluster-specific optimization set of preprocessing parameters; and multiplying said preprocessed locally corrected spectrum by a cluster-specific correction vector to acquire a corrected concentration value for said analyte. Claim 15 In claim 14, the step of obtaining the raw spectrum comprises: directing electromagnetic radiation (EMR) tunable to a plurality of wavelengths from the SoC toward the object; measuring the intensities of the EMR received from the object at each of the plurality of wavelengths using the SoC; and converting the intensities into absorbance values, wherein the raw spectrum comprises an absorbance spectrum. Claim 16 A method according to claim 15, wherein the plurality of wavelengths are selected from the range of 1000 nm to 3500 nm or the range of 1900 nm to 2500 nm. Claim 17 A method according to any one of claims 14 to 16, wherein a plurality of clusters of the spectra correspond to spectra previously collected using the SoC; and each of the plurality of clusters is represented by an individual LSC reference, a cluster center, and an individual correction vector, and the individual LSC reference, the individual cluster center, and the individual correction vector for each cluster are stored on the SoC. Claim 18 A method according to any one of claims 14 to 16, wherein the step of identifying the cluster to which the raw spectrum belongs from a plurality of clusters of the spectra comprises: deriving a globally corrected spectrum using a global scattering correction (GSC) criterion; within each cluster of the plurality of clusters: comparing the globally corrected spectrum with an individual LSC criterion to obtain a distance corresponding to the cluster; and selecting the cluster having the minimum corresponding distance. Claim 19 A method according to claim 18, wherein the global scattering correction comprises global multiplicative scattering correction, global standard normal variable (SNV) correction, Kubelka-Munk correction, Saunderson correction, global mean centering and normalization correction, or a combination thereof. Claim 20 In paragraph 18, the method wherein the local or global scattering correction includes particle-size difference correction or path length difference correction. Claim 21 A method according to claim 20, wherein the particle-size difference correction or the path length correction comprises a Kubelka-Munk correction, a Saunderson correction, a multiplicative scattering correction, or a combination thereof. Claim 22 A method according to any one of claims 14 to 16, wherein the local scattering correction comprises a local multiplicative scattering correction, a local standard normal variable (SNV) correction, or a local mean centering and normalization correction, a Kubelka-Munk correction, a Saunderson correction, or a combination thereof. Claim 23 A method according to any one of claims 14 to 16, wherein the step of determining the spectral shape of the raw spectrum comprises the step of preprocessing the raw spectrum by applying linear transformation and baseline correction to the raw spectrum based on the reference spectrum of the selected analyte. Claim 24 A method according to claim 23, wherein the preprocessing step comprises Kubelka-Munk correction, Saunderson correction, multiplicative scattering correction, or a combination thereof. Claim 25 A system for measuring the concentration of an analyte, comprising: a hybrid III-V / IV semiconductor photonics SoC (system-on-a-chip) for obtaining a raw spectrum from an object having said analyte; and a processing unit including a processor and memory, wherein the processing unit uses said hybrid III-V / IV semiconductor photonics SoC (system-on-a-chip) to obtain a raw spectrum from an object having said analyte; identifies a cluster to which said raw spectrum belongs from a plurality of clusters of spectra based on the spectral shape of said raw spectrum; applies local scattering correction (LSC) to said raw spectrum to obtain a locally corrected spectrum; preprocesses said locally corrected spectrum using a cluster-specific optimization set of preprocessing parameters; and multiplies said preprocessed locally corrected spectrum by a cluster-specific correction vector to obtain a corrected concentration value for said analyte. Claim 26 In paragraph 25, to obtain the raw spectrum, the SoC is configured to direct electromagnetic radiation (EMR) tunable to a plurality of wavelengths toward the object; to measure the intensities of the EMR received from the object at each of the plurality of wavelengths; and the processor is programmed to convert the intensities into absorbance values, wherein the raw spectrum comprises an absorbance spectrum. Claim 27 A system according to claim 26, wherein the plurality of wavelengths include a range of 1000 nm to 3500 nm or a range of 1900 nm to 2500 nm. Claim 28 A system according to any one of claims 25 to 27, wherein a plurality of clusters of the spectra correspond to spectra previously collected using the SoC; each of the plurality of clusters is represented by an individual LSC reference, an individual cluster center, and an individual correction vector; and the SoC comprises a memory for storing, for each cluster, the individual LSC reference, the individual cluster center, and the individual correction vector. Claim 29 A system according to any one of claims 25 to 27, wherein the SoC comprises a memory for storing an optimized set of preprocessing parameters for each cluster. Claim 30 A system, wherein, in any one of claims 25 to 27, to identify the cluster to which the raw spectrum belongs from a plurality of clusters of the spectra, the processor derives a globally corrected spectrum using a global scattering correction (GSC) criterion; and within each cluster of the plurality of clusters: compares the globally corrected spectrum with an individual LSC criterion to obtain a distance corresponding to the cluster; and is programmed to select the cluster with the minimum corresponding distance. Claim 31 A system according to paragraph 30, wherein the global scattering correction comprises global multiplicative scattering correction, global standard normal variable (SNV) correction, Kubelka-Munk correction, Saunderson correction, or global mean centering and normalization correction. Claim 32 A system according to paragraph 30, wherein the local or global scattering correction comprises a particle-size difference correction or a path length difference correction, and said particle-size difference correction or said path length difference correction comprises a Kubelka-Munk correction, a Saunderson correction, a multiplicative scattering correction, or a combination thereof. Claim 33 A system according to any one of claims 25 through 27, wherein the local scattering correction comprises local multiplicative scattering correction, local standard normal variable (SNV) correction, Kubelka-Munk correction, Saunderson correction, or local mean centering and normalization correction or a combination thereof. Claim 34 A system according to any one of claims 25 to 27, wherein the SoC comprises: a wavelength shift tracker for tracking a shift in the wavelength of radiation emitted by the SoC; a wavelength tracker for tracking the absolute wavelength of the radiation emitted by the SoC; a temperature sensor for measuring the temperature of the SoC; and an SoC output power monitor for monitoring the intensity of the EMR emitted by the SoC during a wavelength sweep. Claim 35 A system according to any one of claims 25 to 27, wherein, in order to determine the individual spectral shapes of the plurality of raw spectra, the processing unit is configured to preprocess the plurality of raw spectra by applying linear transformation and baseline correction to the plurality of raw spectra based on the reference spectrum of the selected analyte. Claim 36 A system according to claim 35, wherein, while performing the above preprocessing, the processing unit is configured to apply Kubelka-Munk correction, Saunderson correction, multiplicative scattering correction, or a combination thereof.

Images