Device for mitigation of non-analyte derived signals - Patent Application 20070122999

Abstract

Disclosed herein are devices (120) and methods (200) for mitigating erroneous signals imparted by physical and / or chemical processes incident on an analyte-selective electrochemical sensor, the signals being non-analyte-related in origin. The devices include an analyte-selective sensor and an analyte-invariant sensor. The analyte-selective sensor includes a first electrode (40a) having a selective recognition element (41) and a membrane (42) configured to generate a product resulting from interaction of the selective recognition element (41) with the analyte. The analyte-invariant sensor includes a second electrode (40b) having a membrane (43).

Description

[Technical Field]

[0001] The present invention generally relates to an analyte-selective sensor and method for constructing same, as well as a microneedle applicator integrated within a wearable sensor body housing. [Background technology]

[0002] Continuous assessment of circulating blood glucose levels remains crucial for the management of diabetes mellitus, especially among those who require regular or continuous insulin infusion for the management of this chronic disease. 1 This challenge is addressed by continuous glucose monitors (CGMs), which are widely used by people with insulin-dependent diabetes mellitus. 1 CGM was developed throughout the 1990s and first commercialized in 1999 to provide more granularity for directing therapeutic insulin delivery rather than relying solely on infrequent finger-prick capillary blood sampling. 2 Despite the less-discussed clinical benefits of continuous glucose monitoring over fingerstick capillary blood sampling, today, adoption of CGM in intensively insulin-managed patients remains lacking, due in part to the limited reliability and accuracy of such systems. 3Indeed, accuracy is often compromised due to signal perturbations that are inherently intrinsic (sensor warm-up, in situ presence of reactive oxygen species, interstitial fluid convection) as well as extrinsic (pressure-induced signal attenuation, temperature fluctuations). In these scenarios, a single sensing element may often be compromised in its ability to faithfully track dynamic fluctuations in blood glucose or other physiologically relevant circulating analytes. However, with the goal of improving device accuracy, the integration of both analyte-selective and analyte-invariant sensing modalities as separable components within continuous analyte monitors (e.g., CGMs) is an active area of ​​development. As noted above, such sensing element integration presents its own unique set of challenges: integrating multiple sensing elements into a single transducer, minimizing excessive interactions between the sensing elements, and developing robust methods for the precise deposition of unique analyte-selective and analyte-invariant sensing chemistries within the transducer. In light of these challenges, much of the prior art has taught single-transducer designs configured solely for analyte detection. In such embodiments, the analyte sensing system is referred to as a plurality of electrodes made up of adjacent metal wires or adjacent metal conduits on a flexible substrate.

[0003] Prior art solutions have primarily concerned mitigating non-analyte-related signal perturbations through physical, chemical, algorithmic, and contextual methods. The most relevant and basic example involves requesting or otherwise prompting the user through a software interface to notify the system when engaging in specific activities, such as exercise, administering specific therapeutic agents (i.e., acetaminophen, insulin), or consuming carbohydrates (as relevant to CGM). Another example of a physical method involves reducing the profile of the body-worn sensing component to reduce susceptibility to pressure-induced perturbations of the analyte signal. Within the chemical realm, the synthesis of ever more selective receptor molecules and diffusive flux-limiting membranes aims to increase sensor selectivity in the face of the undue influence of complex arrays of endogenous (i.e., metabolites, hormones, neurotransmitters, small molecules) and exogenous (i.e., pharmaceuticals, supplements, drugs of abuse) analytes compatible with physiological fluids. Similarly, the implementation of advanced signal processing algorithms targeting outlier detection, fault compensation, and non-physiological rates of change is often employed to reduce the dominance of signal artifacts and the deleterious contributions of various physical and chemical processes to sensor performance. More recently, sensor fusion methods aimed at contextual assessment of the state of the user and their body-worn sensors have been investigated as a potential countermeasure for reducing the dominance of signal artifacts. In such solutions, data from orthogonal measurements (kinesthetic, electrophysiological, electrodermal, and optophysiological sensors, among others) are integrated into fusion algorithms to better understand possible non-analyte contributions to the signal transduction by an analyte-selective sensor.

[0004] Microneedle arrays (MNAs) require insertion within a specific range of speeds. Typically, MNAs are components of a mechanically rigid assembly that includes electronics, a housing, an adhesive, and the MNA attached to a sensor body. Inserting this entire sensor body presents several challenges, including accelerating the mass to a forceful velocity over a short distance, stretching the skin before accelerating the sensor body into the skin so that the skin is taut and does not displace away from the MNA upon impact, and then releasing the sensor body from the applicator, in addition to myriad other concerns such as the overall cost and size of the applicator and preventing unintended misuse of this complex, multi-step mechanism.

[0005] Stretching the skin causes discomfort to the user, and the impact of the sensor against the skin also causes discomfort to the user.

[0006] This same impact of the sensor body moving at a relatively high speed may also cause the body to react with an immunological response such as redness, swelling, erythema, and / or edema in the area where the sensor was placed, which increases the warm-up time of the sensor and increases discomfort and distress to the user.

[0007] Existing needle-, trocar-, and cannula-based analyte sensors configured for the selective detection of a target analyte (i.e., blood glucose) are also sensitive to external mechanical, electrical, and chemical stimuli that impair the accurate determination of the target analyte or analytes. Specifically, these stimuli manifest primarily as undesirable perturbations in the signal(s) converted from the analyte-selective sensor, which serve to introduce errors into the measurement, thereby compromising the ultimate accuracy achievable with such devices. Indeed, these errors can often be exacerbated and potentially become life-threatening when the analyte determination provided by the analyte-selective sensor is utilized in a closed-loop system configured to manage a chronic disease. In such a scenario, a potentially lethal dose of a therapeutic agent can be delivered autonomously and without user intervention as a response to counter a perceived pathophysiological reading; a suitable example is in the realm of automated insulin delivery (AID) for people with intensive insulin-controlled diabetes. Analyte-selective sensors operate in a manner that enhances selectivity for target analytes through the implementation of receptor molecules (i.e., enzymes, antibodies, aptamers), capture probes (i.e., single-stranded DNA), or selective catalysts (i.e., noble metals, inorganic species, electrochemical mediators), but these devices often succumb to extrinsic influences, including, among other things, pressure-induced signal decay, nonspecific binding of foreign analytes to receptor molecules, changes in equilibrium conditions, and interactions with endogenous and exogenous chemical species that occupy physiological environments. This is primarily due to challenges associated with the inability to separate or otherwise measure contributions made by external stimuli that are primarily nondeterministic and stochastic in nature. To address this challenge, analyte sensing systems have been constructed featuring both analyte-selective and analyte-invariant sensing elements, each embodying a unique chemical element configuration, to ratiometrically scale the analyte-selective sensor response and mitigate external sources of excessive signal influence. 4However, in practice, significant difficulties arise when attempting to deposit different chemistries onto a sensor geometry where both analyte-selective and analyte-invariant sensors are co-located within a single integrated sensing element / transducer or intermingled in close proximity, which can result in undesirable effects such as crosstalk. More recent efforts have targeted algorithmic and contextual methods to denoise analyte signals without requiring the addition of a second sensing modality, but these approaches have met with very limited success. Summary of the Invention [Means for solving the problem]

[0008] The ability to distinguish signal contributions that are non-analyte in origin allows for the implementation of various mathematical methods to deconvolute or otherwise separate signals that are purely analyte in origin from signals resulting from external influences. The present invention teaches the implementation of at least two distinct sensing elements present within a microneedle array, whereby at least one unique sensing element embodies the ability to quantify the presence of a target analyte (an analyte-selective sensor) but is undesirably, but differently sensitive to external stimuli, and at least one unique sensing element is not selective for the presence of the target analyte (an analyte-invariant sensor) but is desirably, but differently sensitive to external stimuli.

[0009] The present invention teaches devices and methods for mitigating erroneous signals imparted by physical and / or chemical processes incident on an analyte-selective electrochemical sensor whose origin is non-analyte-related. These processes often act to corrupt the measurement signal provided by the analyte-selective sensor. The solution described herein relates to the implementation of analyte-invariant measurements that are differently sensitive to physical and chemical perturbations incident on the sensing system. This requires the construction of a sensing system featuring at least one analyte-selective sensor and at least one analyte-invariant sensor. In preferred embodiments, the analyte-invariant sensor exhibits the same structure and component configuration as the analyte-selective sensor, except for the addition of an active biorecognition element, affinity molecule, catalyst, or capture probe that is selective for the target analyte. In alternative embodiments, a deactivated biorecognition element that does not express residual biospecific activity can be included in the analyte-invariant sensor. In yet another embodiment, the active biorecognition element can be incorporated into the analyte-invariant sensor, but is subjected to a deactivation process during sensor fabrication. In this way, any non-analyte signal perturbations are incident on both the analyte-selective and analyte-invariant sensing elements and can be separated from the underlying analyte-derived signal through various mathematical transformations to maximize measurement accuracy and reliability. In this manner, mitigation of common-mode effects on the analyte-selective sensor that are also detected by the analyte-invariant sensor can be achieved, and therefore an overall improvement to analyte signal fidelity (e.g., signal-to-noise ratio or similar characteristics) can be expected.

[0010] Another objective is to eliminate the need to stretch the skin to insert the MNA.

[0011] Another objective is the ability to effectively insert a wide variety of needles, both blunt and sharp.

[0012] One aspect of the present invention is a device for mitigating non-analyte-derived signal perturbations incident on a body-worn microneedle array-based analyte sensor. The device includes a first electrode and a second electrode. The first electrode is positioned on a surface of a first microneedle of the microneedle array. A selected recognition element is disposed on the first electrode and configured to generate a product or a change in physical state resulting from interaction of the selected recognition element with an analyte. A membrane is disposed on the selected recognition element. A second electrode is positioned on a surface of a second microneedle of the microneedle array, and the membrane is disposed on the second electrode. The first electrode and the second electrode are positioned at spatially distinct locations within the viable epidermis or dermis of a user. A bias potential or current is applied to each of the first electrode and the second electrode. A subsequent electrical response from each of the first electrode and the second electrode is measured. A mathematical transformation is applied to the electrical response produced at the first electrode as a function of the electrical response produced at the second electrode to cause attenuation of the common mode signal.

[0013] Another aspect of the present invention is a device for mitigating non-analyte-derived signal perturbations incident on a body-worn analyte sensor. The device includes a first electrode and a second electrode. A selected recognition element is disposed on the first electrode and configured to generate a product resulting from interaction of the selected recognition element with an analyte. A membrane is disposed on the selected recognition element. The membrane is disposed on the second electrode. The first electrode and the second electrode are positioned at spatially distinct locations within a user's viable epidermis or dermis. A bias potential or current is applied to each of the first electrode and the second electrode. Subsequent electrical responses from each of the first electrode and the second electrode are measured. A mathematical transformation is applied to the electrical response generated at the first electrode as a function of the electrical response generated at the second electrode to cause attenuation of common-mode signals.

[0014] Yet another aspect of the invention is a device having an analyte-selective sensor and an analyte-invariant sensor for mitigating non-analyte-derived signal perturbations incident on a body-worn analyte sensor. The device comprises the analyte-selective sensor and the analyte-invariant sensor. The analyte-selective sensor comprises a first electrode and a selection recognition element disposed on the first electrode, the selection recognition element configured to generate a product or a change in physical state resulting from interaction of the selection recognition element with an analyte. A membrane is disposed on the selection recognition element. The analyte-invariant sensor comprises a second electrode and a membrane disposed on the electrode. The analyte-selective sensor and the analyte-invariant sensor are positioned at spatially distinct locations within the viable epidermis or dermis of a user. A bias potential or current is applied to each of the analyte-selective sensor and the analyte-invariant sensor. A subsequent electrical response is measured from each of the analyte-selective sensor and the analyte-invariant sensor. A mathematical transformation is applied to the electrical response produced at the analyte-selective sensor as a function of the electrical response produced at the analyte-invariant sensor to cause attenuation of the common-mode signal.

[0015] Yet another aspect of the present invention is a method for mitigating non-analyte-derived signal perturbations incident on a body-wearable microneedle array-based analyte sensor. The method includes positioning a first microneedle and a second microneedle of a microneedle array at spatially distinct locations within the living epidermis or dermis of a user, the first microneedle featuring a first electrode and a selected recognition element disposed on the first electrode, the selected recognition element configured to generate a product or physical state change resulting from interaction of the selected recognition element with an analyte, a membrane disposed on the selected recognition element, and the second microneedle featuring a second electrode and a membrane disposed on the second electrode. The method also includes applying a bias potential or current to each of the first electrode and the second electrode. The method also includes measuring a subsequent electrical response from each of the first electrode and the second electrode. The method also includes applying a mathematical transformation to the electrical response generated at the first electrode as a function of the electrical response generated at the second electrode to cause attenuation of the common mode signal.

[0016] Yet another aspect of the present invention is a method for mitigating non-analyte-derived signal perturbations incident on a body-worn analyte sensor. The method includes positioning a first electrode and a second electrode of the analyte sensor at spatially distinct locations within a user's living epidermis or dermis, the first electrode comprising a selected recognition element disposed thereon, the selected recognition element configured to generate a product or physical state change resulting from interaction of the selected recognition element with an analyte, a membrane disposed thereon, and the second electrode comprising a membrane disposed thereon. The method also includes applying a bias potential or current to each of the first electrode and the second electrode. The method also includes measuring subsequent electrical responses from each of the first electrode and the second electrode. The method also includes applying a mathematical transformation to the electrical response generated at the first electrode as a function of the electrical response generated at the second electrode to cause attenuation of common-mode signals.

[0017] Yet another aspect of the invention is a method for mitigating non-analyte-derived signal perturbations incident on a body-worn analyte sensor system. The method includes positioning an analyte-selective sensor and an analyte-invariant sensor of the analyte sensor system at spatially distinct locations within a user's living epidermis or dermis, where the analyte-selective sensor features a first electrode, a selected recognition element disposed on the first electrode, the selected recognition element configured to generate a product or physical state change resulting from interaction of the selected recognition element with an analyte, and a membrane disposed on the selected recognition element, and the analyte-invariant sensor includes a second electrode and a membrane disposed on the second electrode. The method also includes applying a bias potential or current to each of the analyte-selective sensor and the analyte-invariant sensor. The method also includes measuring subsequent electrical responses from each of the analyte-selective sensor and the analyte-invariant sensor. The method also includes applying a mathematical transform to the electrical response generated at the analyte-selective sensor as a function of the electrical response generated at the analyte-invariant sensor to cause attenuation of common-mode signals.

[0018] The analyte preferably comprises at least one of a biomarker, a chemical, a biochemical, a metabolite, an electrolyte, an ion, a hormone, a neurotransmitter, a vitamin, a mineral, a drug, a therapeutic agent, a toxin, a pathogen, an infectious agent, an allergen, an enzyme, a protein, a nucleic acid, DNA, and RNA.

[0019] The analyte sensor system is preferably a microneedle or microneedle array, each microneedle component preferably having a vertical extent of 200-2000 μm.

[0020] The microneedle or microneedle array preferably comprises at least one protrusion that is capable of being inserted into the viable epidermis or dermis of a user.

[0021] The first and second electrodes preferably comprise a metal, a metal alloy, a metal oxide, a semiconductor, or a polymer surface.

[0022] The first and second electrodes are confined to the tapered distal regions of the microneedles or elements of the microneedle array.

[0023] The selection recognition element preferably comprises at least one of an enzyme, an aptamer, an antibody, a capture probe, an ionophore, a catalyst, a biocatalyst, DNA, RNA, an organelle, or a cell.

[0024] The product is preferably a chemical, a biochemical, a mediator, a resistance change, an electrical signal, an electrochemical signal, a conductance change, an impedance change, or an absorbance change.

[0025] The membrane is preferably at least one of a polymer, a hydrophilic layer, a biocompatible layer, a diffusion-limiting layer, a hydrogel, a film, and a coating.

[0026] The bias potential or current is preferably either direct or alternating.

[0027] The electrical response preferably includes at least one of potential, current, impedance, conductance, resistance, capacitance, and inductance.

[0028] The mathematical transformation preferably includes at least one of a differencing operation, a denoising operation, regression, deconvolution, Fourier decomposition, background subtraction, Kalman filtering, and maximum likelihood estimation.

[0029] Attenuation preferably includes at least one of eliminating, minimizing, or reducing the duration of the common mode signal.

[0030] The common mode signal preferably includes at least one of a warm-up signal following application of the microneedle array-based analyte sensor to the wearer's skin, pressure-induced signal artifacts, temperature-induced signal fluctuations, and interference signals due to endogenous or exogenous chemical species circulating in the user's physiological fluids.

[0031] The endogenous or exogenous chemical species preferably comprises at least one of a biomarker, a chemical, a biochemical, a metabolite, an electrolyte, an ion, a hormone, a neurotransmitter, a vitamin, a mineral, a drug, a therapeutic agent, a toxin, a pathogen, an infectious agent, an allergen, an enzyme, a protein, a nucleic acid, DNA, and RNA.

[0032] The physiological fluid is preferably at least one of the user's interstitial fluid, skin interstitial fluid, or blood. [Brief explanation of the drawings]

[0033] [Figure 1] FIG. 10 is a block diagram illustrating the use of an analyte-invariant signal in conjunction with an analyte-selective signal to denoise an analyte signal. [Figure 2A] 1 is a raw signal trace resulting from an analyte-invariant sensor and an analyte-selective sensor. [Figure 2B] 1 is a raw signal trace resulting from an analyte-invariant sensor and an analyte-selective sensor. [Figure 2C] 1 is a raw signal trace resulting from an analyte-invariant sensor and an analyte-selective sensor. [Figure 3A] 1 is a raw signal trace resulting from multiple analyte-selective sensors. [Figure 3B] 1 is a raw signal trace resulting from multiple analyte-selective sensors. [Figure 3C] 1 is a raw signal trace resulting from multiple analyte-selective sensors. [Figure 4] 1 is a bar graph illustrating improvement in analyte-selective sensor accuracy. [Figure 5] 1 is a prior art analyte selective sensor block / process flow diagram, where the non-analyte signal is additive to the analyte signal. [Figure 6] FIG. 1 is a block / process flow diagram of an analyte-invariant sensor. [Figure 7] FIG. 1 is a block / process flow diagram of a system for removing perturbations to an analyte signal that are non-analyte in origin (and additive). [Figure 8] FIG. 1 is a block / process flow diagram of a system for removing perturbations to an analyte signal that are non-analyte in origin (and additive). [Figure 9] FIG. 1 is a block / process flow diagram of a system for removing common mode signals resulting from perturbations that are non-analyte in origin. [Figure 10] FIG. 1 is a block / process flow diagram of a system for removing common mode signals resulting from perturbations that are non-analyte in origin. [Figure 11] FIG. 1 is a block diagram of a device for removing signal perturbations that are non-analyte in origin. [Figure 12] 1 is a flow chart of the method of the present invention under a microneedle embodiment. [Figure 13] 1 is a flow chart of the method of the present invention under an electrode embodiment. [Figure 14] 1 is a flow chart of the method of the present invention under analyte-selected and analyte-invariant embodiments. [Figure 15A] FIG. 1 is a top view of an embodiment of a device of the present invention. [Figure 15B] FIG. 1 is a side view of an embodiment of a device of the present invention. [Figure 15C] FIG. 15C is a cross-sectional view of the device of FIG. 15B with the MNA retracted. [Figure 15D] FIG. 15C is a cross-sectional view of the device of FIG. 15B with the MNA released. [Figure 15E] FIG. 1 is a side view of an embodiment of a device of the present invention with the housing removed and the MNA retracted. [Figure 15F] FIG. 1 is a side view of an embodiment of a device of the present invention with the housing removed and the MNA released. [Figure 15G] FIG. 15D is a perspective view of the device of FIG. 15C. [Figure 16] FIG. 1 is a cross-sectional view of skin with subcutaneously implanted microneedles having electrodes. [Figure 17A] FIG. 1 is a diagram of a microneedle array composed of a first electrode having a selected recognition element disposed on the first electrode, a membrane disposed on the selected recognition element, and a membrane disposed on a second electrode. [Figure 17B] FIG. 1 is a diagram of a microneedle array composed of a first electrode with selected recognition elements disposed on the first electrode and a membrane (blanket configuration). [Figure 17C] FIG. 1 is a diagram of a microneedle array composed of a first electrode having a membrane containing a selected recognition element disposed on the first electrode, and a membrane disposed on a second electrode. [Figure 17D] FIG. 1 is a diagram of a microneedle array showing the main components and measurements. [Figure 18A] FIG. 1 is a top view of an embodiment of a device of the present invention. [Figure 18B] FIG. 18B is a side view of the device of FIG. 18A. [Figure 18C] FIG. 18B shows an exploded view rendering of the device of FIG. 18A. [Figure 19] 1 shows the electronic circuitry housed in a prototype wearable device enclosure designed to interface directly with a microneedle-based biosensor device. [Figure 20] FIG. 10 shows another view of the electronic circuitry housed in a prototype wearable device enclosure designed to interface directly with a microneedle-based biosensor device. [Figure 21] Shown is the electronic circuitry housed in a sealed housing, with access to the microneedle device provided via a gold-plated pressure connector located on the visible surface of the housing. [Figure 22] 1 shows a skin-piercing hollow microneedle array with multiple protrusions having a vertical extent of approximately 1000 μm, where each element of the microneedle array is functionalized to provide selective biosensing capabilities. [Figure 23A] 1 shows a hollow, non-functionalized microneedle array. [Figure 23B] 1 shows hollow "filled" functionalized microneedle arrays with selective biosensing capabilities. [Figure 24]FIG. 1 shows an exploded view rendering of the complete microneedle biosensing system, showing all functional components, including the microneedle biosensor and the printed circuit board that houses the electronics required to convert the biochemical signal into digital data that can be wirelessly transmitted to an external device via an implanted wireless transceiver. [Figure 24A] FIG. 25 is an isolated enlarged view of the microneedle biosensor components of FIG. 24. [Figure 25] FIG. 1 shows another view of the wearable microneedle biosensing system. [Figure 26] A rear view of the electronic components is shown. [Figure 27] A detailed block / process flow diagram is shown. [Figure 28] FIG. 1 is a circuit diagram of a stand-alone potentiostat integrated circuit. [Figure 29] FIG. 1 is a circuit diagram of a multi-component potentiostat. [Figure 30] FIG. 1 is a block diagram of a differential amplifier. [Figure 31] FIG. 2 is a signal flow diagram of the present invention. [Figure 32] FIG. 1 is a circuit diagram of an integrated analog front end and sensor interface. [Figure 33] FIG. 1 is a circuit diagram of a mirror differential amplifier and filtering. [Figure 34] FIG. 1 is a circuit diagram of a fixed mirror instrumentation amplifier. [Figure 35] FIG. 1 is a circuit diagram of a digital potentiometer-tunable Miller instrumentation amplifier. [Figure 36] FIG. 1 is a diagram of a handheld analyzer in a large form factor. [Figure 37] FIG. 1 is a diagram of a handheld analyzer in a small form factor. [Figure 38] FIG. 1 is a block diagram of a sample algorithm. [Figure 39] FIG. 1 is a diagram of a handheld analyzer in a small form factor. DETAILED DESCRIPTION OF THE INVENTION

[0034] Body-worn analyte-selective sensors, such as continuous glucose monitors, are highly sensitive electrochemical systems configured to sense an analyte or analytes in a selective manner with high accuracy. This accuracy can be unduly affected by various external stimuli, which cause undesirable perturbations in the signal(s) converted from the analyte-selective sensor, thereby introducing errors into the measurement and compromising the ultimate accuracy achievable with such devices. Thus, even the most sophisticated analyte-selective sensors are often subject to external perturbations that may be chemical, electrical, or mechanical in origin. The present innovation aims to mitigate the dominance of excessive physical, chemical, and otherwise exogenous influences on the fidelity of the measurement of a target analyte or analytes. This is achieved through the implementation of at least one analyte-selective sensor and at least one analyte-invariant sensor, whereby the analyte-selective sensor features a selective recognition element, and the analyte-invariant sensor lacks a selective recognition element but is otherwise identical in structure and component configuration to the analyte-selective sensor. Using mathematical transformations, algorithms, or a combination thereof, the common mode signal appearing in both the analyte-selective and analyte-invariant sensors may be minimized, mitigated, or completely eliminated, thereby resulting in an analyte signal of higher fidelity and / or accuracy.

[0035] FIG. 1 shows a block diagram 10 illustrating the use of an analyte-invariant (non-enzyme) signal 11 along with an analyte-selective (Current Ch1, Current Ch2, Current Ch3) signal 13 to denoise an analyte signal 14 (Current Ch1 clean, Current Ch2 clean, Current Ch3 clean). ** What about the temperature signal 12? Processing 1 / RLS filter 15?

[0036] To mitigate non-analyte-derived signal perturbations incident on a body-worn microneedle array-based analyte sensor, the device is configured to feature at least one analyte-selective sensor and at least one analyte-invariant sensor, both located on unique microneedle components of the array, as shown in Figure 17A. Specifically, the analyte-selective sensor is configured to feature an electrode 40a on the surface of a first microneedle 30a of the microneedle array, a selection recognition element 41 disposed on the first electrode 40a, the selection recognition element 41 configured to generate a product resulting from interaction of the selection recognition element 41 with an analyte, and a membrane 42 disposed on the selection recognition element 41. Similarly, the analyte-invariant sensor is configured to feature an electrode 40b on the surface of a second microneedle 30b of the microneedle array and a membrane 43 disposed on the electrode. The analyte-selective and analyte-invariant sensors are arranged in a microneedle array to facilitate sensing at spatially distinct locations within the user's viable epidermis 131 or dermis 132, as shown in FIG. 16, thereby functioning to minimize any undue influence or crosstalk from one sensor to another. As shown in FIG. 11, upon application of the same or unique bias signal (DC or AC potential or current) to both the analyte-selective and analyte-invariant sensors, subsequent electrical responses (potential, current, impedance, conductance, resistance, capacitance, or inductance) are measured from both the first and second electrodes. Mathematical transformations are then applied to the electrical response generated at the first electrode as a function of the electrical response generated at the second electrode to remove common-mode signals incident on both the analyte-selective and analyte-invariant sensors. These mathematical transformations can include differential (subtraction) measurements, deconvolution, Fourier decomposition, background subtraction, Kalman filtering, and maximum likelihood estimation.

[0037] In another embodiment of the present invention, as shown in FIG. 17B, an analyte-selective sensor is configured featuring an electrode 40a on the surface of a first microneedle 30a of the microneedle array, a selection recognition element 41 disposed on the first electrode 40a, and a membrane 42 disposed on the selection recognition element 41 and on a second electrode 40b of a second microneedle 30b.

[0038] In yet another embodiment of the invention, as shown in Figure 17C, a sensor is configured featuring an electrode 40a on the surface of a first microneedle 30a of the microneedle array and a membrane 42 containing a selected recognition element 41 disposed on the first electrode 40a. A membrane 43 is disposed on a second electrode 40b on the surface of a second microneedle 30b.

[0039] As shown in FIG. 17D, each microneedle 30 of the microneedle array 20 preferably has a through-silicon via 33 embedded within the microneedle 30. The microneedles 30 preferably have an insulator 34 made of oxide. This allows the sensors to be individually probed as separate components of the microneedle array 20. The microneedle array can preferably be reflow soldered to almost any circuit board, such as an integrated circuit. Each microneedle 30 preferably has an individual sensor 31 confined to the distal tip of the microneedle 30, preferably in a region between 1 and 1500 μm from the distal end of the microneedle 30. The microneedles 30 preferably have a backside metal contact 32, a through-needle via 33, an insulator 34 that electrically insulates the microneedle 30, and a patterned metal contact 35 on the distal tip 36 of the microneedle 30. The back metal contact 32 is preferably constructed of a nickel / gold material with the inner portion 37 constructed of an aluminum material. The microneedle 30 preferably has a through needle via 33 constructed of a silicon material. The distal tip 36 preferably has an oxide portion and a platinum portion. The length Lm of the microneedle 30 is preferably in the range of 200-2000 μm, and most preferably 625 μm. The width Wm of the microneedle 30 is preferably in the range of 100-500 μm, and most preferably 160 μm. The distal tip 36 preferably has a length Ld in the range of 50-200 μm, and most preferably 100 μm.

[0040] The presented devices and methods are capable of determining analytes including at least one of biomarkers, chemicals, biochemicals, metabolites, electrolytes, ions, hormones, neurotransmitters, vitamins, minerals, drugs, therapeutic agents, toxins, enzymes, proteins, nucleic acids, aptamers, DNA, and RNA. Furthermore, these systems use microneedle arrays containing at least two protrusions that can be inserted into a user's living epidermis or dermis, each protrusion having a range of 200-2000 micrometers from proximal to distal end. The electrode constituents discussed above are limited to the distal regions of the aforementioned protrusions and include metal, semiconductor, or polymer surfaces. The selected recognition elements discussed include at least one of enzymes, aptamers, antibodies, capture probes, ionophores, catalysts, biocatalysts, DNA, RNA, organelles, or cells, and are configured to generate a chemical, biochemical, mediator, resistance change, electrical signal, conductance change, impedance change, or absorbance change upon exposure to the analyte. The membrane is at least one of a polymer, a hydrophilic layer, a biocompatible layer, a diffusion-limiting layer, a hydrogel, a film, and a coating.

[0041] Other novel and practical features of the present invention include its inherent ability to negate the effects of crosstalk due to diffusive transport of products from an analyte-selective sensor to an analyte-invariant sensor. The present invention also reduces the impact of analyte-depletion zone or diffusion layer effects, which act to limit the amount of analyte that can diffuse into an analyte-selective sensor.

[0042] Figures 2A-2C show raw signal traces resulting from an analyte-invariant sensor (non-enzyme), as shown in Figure 2A, and an analyte-selective sensor (raw Ch1), as shown in Figure 2B, with common-mode signal perturbations highlighted by red boxes. Performing a mathematical transformation allows for removal of common-mode signal perturbations (red lines), which are non-analyte in origin, as they appear in both the analyte-invariant and analyte-selective sensors, as shown in Figure 2C.

[0043] Figures 3A-3C show raw signal traces resulting from multiple analyte-selective sensors (raw Ch1, raw Ch2, raw Ch3) as shown in Figure 3A and an analyte-invariant sensor (non-enzyme) as shown in Figure 3B, illustrating the 2-hour warm-up period required for sensor equilibration after implantation into tissue. Performing a mathematical transformation allows the apparent warm-up period to be reduced to less than 1 hour, as shown in Figure 3C. The warm-up period is non-analyte in origin, as it appears for both analyte-invariant and analyte-selective sensors.

[0044] FIG. 4 shows a bar graph illustrating improvement in analyte-selective sensor accuracy (evidenced by mean absolute relative difference, MARD) on the first day of sensor use by extending the warm-up period or by implementing an algorithm configured to subtract the analyte-invariant sensor signal from the analyte-selective sensor signal.

[0045] Assuming that the analyte-selective sensor is sensitive to non-analyte signal perturbations (in addition to the analyte signal) and the analyte-invariant sensor is purely a function of the non-analyte signal perturbations (i.e., unaffected by the analyte signal), the true analyte signal is isolated by differential measurement. True analyte signal = analyte-selective sensor signal - analyte-invariant sensor signal

[0046] The above relationship is implemented in a simple digital signal processing routine (such as a subtractor / difference engine) executed in the device's firmware or software, which can equally be realized with simple analog hardware such as a differential amplifier.

[0047] The common mode signal, which appears in both the analyte-selective and analyte-invariant sensors, is separated using several methods: First, it is subtracted from the analyte signal by a subtractive relationship. True analyte signal = [analyte-selective sensor signal + common-mode signal] - [analyte-invariant sensor signal + common-mode signal]

[0048] This is achieved through a differential amplifier and a simple analog signal processing routine.

[0049] Assuming that the common mode signal is not additive, but is present in its entirety at the analyte-invariant sensor and as a modulation of the signal provided by the analyte-selective sensor, the common mode signal is separated ratiometrically by the following relationship:

[0050]

number

[0051] Convolution methods can be employed to separate pure analyte-selective signal components from other noise sources. Assume that the measurement signal [m(x)] from an analyte-selective sensor represents the convolution of the component of the signal [a(x)] that is purely analyte-derived, as measured by an analyte-invariant sensor, with a component [n(x)] contributed by sources of erroneous signal measurements that are non-analyte in origin. m(x)=a(x) * n(x)

[0052] Fourier or wavelet-based decomposition of both the analyte-selective and analyte-invariant signals can be employed to spectrally distinguish between the analyte signal and the excessive influence of any non-analyte-derived signal perturbations.

[0053]

number

[0054] Normalization may be employed to place equal weight on the spectral components. M NORM (jω)=A NORM (jω)N NORM (jω) Or rewrite it A NORM (jω)=M NORM (jω) / N NORM (jω)

[0055] Thus, the spectrally pure tones resulting from the analyte-selective signal are calculated using the above relationship: An inverse Fourier or wavelet transform is then employed to return to the time or data sequence domain.

[0056]

number

[0057] The signal-to-noise ratio (SNR) produced by such a system is calculated as the logarithm (base 10) of the ratio of the analyte-selective sensor signal to the analyte-invariant sensor signal.

[0058]

number

[0059] This allows the calculation of the noise figure (NF) of the system.

[0060]

number

[0061] The common-mode rejection ratio (CMRR) is calculated as the logarithm (base 10) of the ratio of the analyte-selective sensor signal to the analyte-invariant sensor signal.

[0062]

number

[0063] Given the ability to measure non-analyte-derived signal contributions, the following list of routines can be employed to compensate for non-analyte contributions from the signal or to generate optimal estimates of analyte concentration.

[0064] Adaptive Filter: Most signal processing applications 5~7 In

[0003] , non-analyte effects are assumed to be additive, due to the fact that multiplicative models represent a larger problem to solve. In these approaches, the general model at each discrete sample n is: s(n)=a(n)+i(n)+e(n) where s(n) is the total detected signal, a(n) is the desired analyte signal, i(n) is the additive contribution due to non-analyte contributions, and e(n) is the filter residual. To solve the above equation, an adaptive filter adjusts the coefficients of a time-varying filter W(n) to regress the non-analyte signal onto s(n). The cost function is It is defined as min{norm(W'i-s,2)}.

[0065] At sample n, the filter residual is

[0066]

number

[0067] 5 to 10 show block diagrams of the processing flow of the sensor.

[0068] FIG. 5 shows a block diagram of a prior art analyte-selective sensor. The non-analyte signal is additive to the analyte signal. FIG. 6 shows an analyte-invariant sensor. FIG. 7 shows a system for removing perturbations to the analyte signal that are non-analyte in origin (and additive). FIG. 8 shows a system for removing perturbations to the analyte signal that are non-analyte in origin (and additive). FIG. 9 shows a system for removing common-mode signals resulting from perturbations that are non-analyte in origin. In this embodiment, the analyte signal is modulated by the common-mode signal, while the analyte-invariant sensor is directly sensitive to the common-mode signal. FIG. 10 shows a system for removing common-mode signals resulting from perturbations that are non-analyte in origin. In this embodiment, the analyte signal is modulated by the common-mode signal, while the analyte-invariant sensor is directly sensitive to the common-mode signal.

[0069] FIG. 11 shows a block diagram 180 of a device for removing signal perturbations that are non-analyte in origin. In certain embodiments, the analyte-selective sensor includes a membrane having an analyte / bio-recognition element. In other embodiments, the analyte-invariant sensor includes a membrane lacking an analyte / bio-recognition element. In still other embodiments, the analyte-selective sensor and the analyte-invariant sensor are at least two separate electrodes. In other embodiments, the analyte-selective sensor is located on an electrode on at least one microneedle of a microneedle array. In still other embodiments, the analyte-invariant sensor is located on at least one microneedle of a microneedle array. In another embodiment, the analyte sensor is a microneedle array. In other embodiments, the analyte sensor system is an analyte-selective microneedle array sensor. In other embodiments, the analyte-selective microneedle array sensor is body-worn on the skin surface of a user. In still other embodiments, the algorithm is processed internally to the device. In other embodiments, the algorithm is processed on a wirelessly connected device. In still other embodiments, the algorithm is processed on a cloud service. In yet other embodiments, the analyte measurements are provided to a user on a display. In other embodiments, the analyte measurements are used to guide therapeutic intervention in an automatic insulin delivery system. In yet other embodiments, the analyte measurements are delivered to a wirelessly connected device. In yet other embodiments, the analyte measurements are stored on a cloud service.

[0070] A method 200 for mitigating non-analyte-derived signal perturbations incident on a body-worn microneedle array-based analyte sensor is shown in FIG. 12. Step 201 involves positioning a first microneedle and a second microneedle of the microneedle array at spatially distinct locations within a user's living epidermis or dermis. Preferably, the first microneedle features a first electrode, a selected recognition element disposed on the first electrode, the selected recognition element configured to generate a product resulting from interaction of the selected recognition element with an analyte, and a membrane disposed on the selected recognition element, and the second microneedle features a second electrode and a membrane disposed on the second electrode. Step 202 involves applying a bias potential or current to each of the first and second electrodes. Step 203 involves measuring a subsequent electrical response from each of the first and second electrodes. Finally, step 204 is to apply a mathematical transformation to the electrical response generated at the first electrode as a function of the electrical response generated at the second electrode to cause attenuation of the common mode signal.

[0071] Another method 205 for mitigating non-analyte-derived signal perturbations incident on a body-worn analyte sensor is shown in FIG. 13. Step 206 begins with positioning a first electrode and a second electrode of the analyte sensor at spatially distinct locations within the user's living epidermis or dermis. Preferably, the first electrode features a selected recognition element disposed thereon, the selected recognition element configured to generate a product resulting from interaction between the selected recognition element and the analyte, and a membrane disposed thereon, and the second electrode features a membrane disposed thereon. Step 207 is applying a bias potential or current to each of the first and second electrodes. Step 208 is measuring a subsequent electrical response from each of the first and second electrodes. Step 209 is applying a mathematical transformation to the electrical response generated at the first electrode as a function of the electrical response generated at the second electrode to cause attenuation of common-mode signals.

[0072] Yet another method 210 for mitigating non-analyte-derived signal perturbations incident on a body-worn analyte sensor system is shown in FIG. 14. Step 211 is positioning an analyte-selective sensor and an analyte-invariant sensor of the analyte sensor system at spatially distinct locations within the living epidermis or dermis of a user. Preferably, the analyte-selective sensor features a first electrode and a selected recognition element disposed on the first electrode, the selected recognition element configured to generate a product resulting from interaction between the selected recognition element and the analyte. Additionally, a membrane disposed on the selected recognition element, and the analyte-invariant sensor features a second electrode and a membrane disposed on the second electrode. Step 212 is applying a bias potential or current to each of the analyte-selective sensor and the analyte-invariant sensor. Step 213 is measuring a subsequent electrical response from each of the analyte-selective sensor and the analyte-invariant sensor. Step 214 is to apply a mathematical transformation to the electrical response generated at the analyte-selective sensor as a function of the electrical response generated at the analyte-invariant sensor to cause attenuation of the common-mode signal.

[0073] Figure 19 shows the electronics housed in a wearable device enclosure 60 designed to interface directly with a microneedle-based biosensor device. The device electronics comprises a wireless transceiver (preferably BLUETOOTH LOW ENERGY) and a microcontroller with an integrated analog-to-digital converter 61 and high amplification circuitry 62. Figure 20 shows another view of the electronics housed in a prototype wearable device enclosure 60 designed to interface directly with a microneedle-based biosensor device. The electronics comprises a high-sensitivity electrochemical analog front end 63 and filtering circuitry 64.

[0074] 21 shows the electronic circuitry housed in the wearable device enclosure 60, with access to the microneedle device provided via a gold-plated pressure connector 67 located on the visible surface of the wearable device enclosure 60. A connection port 65 is also shown.

[0075] Figure 22 shows a skin-piercing hollow microneedle array 70 comprising multiple protrusions having a vertical extent of approximately 1000 μm, where each element of the microneedle array has been functionalized to impart selective biosensing capability. Figure 23A shows a hollow, unfunctionalized microneedle array 70a. Figure 23B shows a hollow, "filled," functionalized microneedle array 70b with selective biosensing capability.

[0076] 24 and 24A show exploded view renderings of the complete microneedle biosensing system 120 showing the functional components, including the housing member 125, the microneedle biosensor 130, and the printed circuit board 127 that houses the electronic circuitry required to convert the biochemical signal into digital data that is wirelessly transmitted to an external device via an implanted wireless transceiver.

[0077] 25 shows a top perspective view of a wearable microneedle biosensing system 120 containing an electronic backbone (protrusion) and an adhesive patch. The microneedles are located on the rear surface of the adhesive patch (not shown).

[0078] FIG. 26 shows a rear view of the electronics housing the components 130 of the microneedle-based biosensing system 120 and the skin-mounted adhesive patch housing the microneedle array 127.

[0079] FIG. 27 shows a detailed block / process flow diagram 1200 illustrating the major functional components of a microneedle-based biosensing system and supporting electronics. In block 1201, a microneedle array is utilized to acquire transcutaneous biochemical analytes from the living physiological media (interstitial fluid, blood) occupying the epidermal and dermal layers of a user of the microneedle-based biosensing system. In block 1202, an electrochemical analog front end performs one or more of several electroanalytical techniques, such as voltammetry, amperometry, potentiometry, conductometry, impedance measurement, and polarography, to facilitate control and readout of the electrochemical reactions occurring in the microneedle-based biosensing system. In block 1203, the electrical signal generated at the output of the electrochemical analog front end is directed to an amplifier circuit to increase the signal strength to line level. In block 1204, the output from the amplifier circuit is directed to a low-pass or band-pass filter to extract the signal of interest and remove any undesired noise. In block 1205, the signal then undergoes analog-to-digital conversion in an ADC to convert the analog signal to a digital bit stream. In block 1206, the signal is routed to a wireless transmitter or transceiver (BLUETOOTH, WiFi, RFID / NFC, Zigbee, Ant+) 1207 to transmit the signal (corresponding to the level of the biochemical analyte) to a mobile communication device 1208 for further information processing, interpretation, display, archiving, and trend analysis.

[0080] The electrochemical analog front end preferably includes the Texas Instruments LMP91000 Sensor AFE System, a configurable AFE potentiostat for low-power chemical sensing applications, the Texas Instruments LMP91200 configurable AFE for low-power chemical sensing applications, or the Analog Devices AuDCM350, a 16-bit precision, low-power meter on a chip with Cortex-M3 and connectivity. The wireless transceiver is preferably the BLUEGIGA BLE-113A BLUETOOTH Smart Module or the Texas Instruments CC2540 SimpleLink BLUETOOTH Smart Wireless MCU with USB. The accompanying mobile device is preferably an ANDROID® or iOS™-based smartphone, a Samsung GALAXY GEAR, or an APPLE WATCH™.

[0081] Microneedle array electrochemical biosensors convert biochemical signals from interstitial fluid into useful electrical signals.

[0082] The electrochemical analog front end preferably performs at least one or more of the following: applying a fixed or time-varying potential to the microneedle array to induce an electrochemical reaction, thereby generating a current flow; applying a fixed or time-varying current to the microneedle array to induce an electrochemical reaction, thereby generating a potential; measuring the time-varying open circuit potential generated by the electrochemical reaction or ion gradient; measuring the frequency-dependent impedance generated by the electrochemical reaction or biocompatible reaction with the microneedle transducer; and measuring the resistivity or conductance generated by the electrochemical reaction or biocompatible reaction with the microneedle transducer.

[0083] The electrochemical analog front end is preferably dynamically configured to achieve any one of the above embodiments. Similarly, the inputs are preferably arranged to operate in serial or parallel fashion to expand the sensing capabilities of the system.

[0084] The wireless transceiver wirelessly relays the electrical signal generated by the electrochemical analog front end to the mobile or wearable device using any one of several standardized wireless transmission protocols (Bluetooth, WiFi, NFC, RFID, Zigbee, Ant+). Optionally, the electrical signal generated by the analog front end can be amplified, filtered, and / or subjected to analog-to-digital conversion and further signal processing before being relayed by the wireless transceiver.

[0085] The mobile or wearable device displays the sensor readings to the user in an easily understood format and performs any additional signal processing required.

[0086] As shown in FIG. 28, the adjustable bias analog front end / potentiostat 69 consists of a high input impedance operational amplifier and a digital-to-analog converter, or a stand-alone analog front end (“AFE”) or analog interface integrated circuit package.

[0087] FIG. 29 is a circuit diagram of a multi-component potentiostat 230 equipped with an electrochemical cell 71.

[0088] The method steps for potentiostat operation are as follows: Analog Front End / Potentiostat Operation. The potentiostat / AFE unit consists of two (Figure 28) or three (Figure 29) precision instrumentation operational amplifiers (A1 / OA1, OA2, and TIA / OA3) configured in the following arrangement: the control amplifier A1 / OA1 generates the differential voltage (V in Figure 20) measured between a variable (programmable) bias and ground. x ) is amplified (with a gain A) and supplies a current through the counter electrode (CE). Upon sensing the voltage generated at the reference electrode (RE), A1 / OA1 converts its output voltage to the input (V RE () value. RE is then adjusted, and the output potential / current of A1 / OA2 (a buffer or unity-gain amplifier) ​​is modified accordingly. Thus, the control amplifier functions as a voltage-controlled current source that maintains the reference electrode at a constant potential and delivers sufficient current to regulate the electrochemical reaction. When implementing negative feedback, it is essential that A1 / OA2 be able to swing to extreme potentials to allow full voltage compliance, which is required for chemical synthesis. Furthermore, to draw negligible current, it is important that OA2 have a very high input impedance; otherwise, the reference electrode may deviate from its intended operating potential. In practice, using a precision amplifier with an input bias current of 20 fA (or less) allows for undampened operation down to sub-picoampere levels, suitable for nearly all electrochemical studies. TIA / OA3 accepts the current supplied through the working electrode (WE) and generates a voltage (resistor / capacitor network R) proportional to the amount of current passing through the WE electrode. TIA / C5+R4).

[0089] Analog front end and applied reference / operating bias. In the systems shown in Figures 28 and 29, the reference voltage (V RE / RE) are held constant at the inverting and non-inverting inputs of operational amplifiers A1 / OA2, respectively, while the operating voltage is varied through a voltage divider, resistor network, or other means to generate an operating bias on the connected sensor. The current flowing from CE to WE is steered to the non-inverting input of a variable gain transimpedance amplifier, which is controlled by the relationship VOUT / Vo=-i cell R 4 / TIA Converts the current flow into a scaled voltage output (with C2 and / or VOUT / Vo) according to

[0090] The differential amplifier stage 75 is shown in FIG. 30. The differential amplifier is configured to accept an applied reference voltage (RE or C1 in the internal IC diagram) and the output from the transimpedance amplifier (with or without a buffer stage). The inputs are juxtaposed between the two amplifiers; that is, the reference input is connected to the positive terminal of one amplifier (for negative applied voltages / currents) and the negative terminal of the other (for positive applied voltages / currents). VOUT is connected to the opposite amplifier input. The unused amplifier (opposite the polarity of the applied current / voltage) has its input driven to zero, but still has ground bias if one is present in the system. The gain of the differential amplifier can be configured through manufacturing or in real time to scale with the amount of voltage / current read by the AFE.

[0091] Filtering Step: The output generated from the differential amplifier pair is then subjected to a filtering circuit to remove external noise. Oscillations or random fluctuations in the signal can be present due to several reasons, including ground bias, RF interference, mains oscillations, input impedance mismatch (from the three-electrode sensor), or other causes.

[0092] Analog-to-Digital Converter Step: The filtered signal finally enters an analog-to-digital converter (ADC), located on an external integrated circuit (IC) or collocated (co-located) within a microcontroller or other IC, and is converted into a representative digital signal. Increased sampling resolution can be implemented to gain additional sensitivity and minimize quantization errors.

[0093] Acquisition Algorithm Steps: To further reduce noise, time-averaged values ​​for both the positive and negative bias lines are collected and calculated by the microcontroller / microprocessor over a period of several seconds (following digitization by the ADC). The active bias amplifier (applied voltage / current) is subtracted from the value of the non-active bias amplifier (ground offset) to remove any bias present in the device. Due to this process, a shielding cage is not required to reach picoamp-level sensitivity. The non-active bias amplifier, time-averaged data collection, and filtering scheme always provide a stable and scalable output to the microcontroller / processor.

[0094] The input of the electrochemical cell or sensor, the analyte, is measured by controlled potential techniques (amperometry, voltammetry, etc.). The output of the sensing system, consisting of the measured voltage and calculated current value (determination of the current flowing through the working and counter electrodes of the electrochemical cell or sensor), corresponds to the concentration of the analyte in the sample.

[0095] FIG. 31 shows a signal flow diagram 80 for detecting current through an electrochemical cell. The current signal from the electrochemical cell 66 is sent to an adjustable bias analog front end 81. This signal is sent to a transimpedance amplifier 82. The signal is sent from both the adjustable bias analog front end 81 and the transimpedance amplifier 82 to a mirror differential amplifier 84. The output generated from the mirror differential amplifier 84 is then subjected to filtering circuits 86 and 87 to remove external noise. Oscillations or random fluctuations in the signal can be present due to several reasons, including ground bias, RF interference, mains oscillations, input impedance mismatch (from the three-electrode sensor), or other causes. In an acquisition algorithm 88, time-averaged values ​​for both the positive and negative bias lines are collected and calculated by the microcontroller / microprocessor over a suitable period, such as several seconds (following digitization by the ADC), to further reduce noise. The active bias amplifier (applied voltage / current) is subtracted from the value of the inactive bias amplifier (ground offset) to remove any bias present in the device. Due to this process, no shielding cage is required to reach picoampere levels of sensitivity. Non-active bias amplifiers, time-averaged data acquisition, and filtering schemes provide a consistently stable and scalable output to the microcontroller / processor / ADC.

[0096] 32 is a detailed circuit diagram of the integrated analog front end 150 and sensor interface. This is a circuit diagram of an integrated AFE available from a manufacturer that communicates with a central microcontroller / microprocessor unit (SCL and SDA lines) and controls the electrochemical sensor via the CE (counter electrode), WE (working electrode), and RE (reference electrode) lines. Configurable circuit components for the transimpedance amplifier (TIA) are present across 9 and 10, forming an integrator as configured in the image.

[0097] Figure 33 is a detailed circuit diagram of the mirror differential amplifier 84' and filtering. Here, a set of mirror differential amplifiers is shown utilizing individual operational amplifier components (left) and a low pass filter on the output (right). AMORP and AMORN are the positive and negative differential signals, and AMOUTN and AMOUTP are the filtered differential signals. The output gain is controlled by passive resistors connected to the amplifiers.

[0098] Figure 34 is a detailed circuit diagram of fixed mirror instrumentation amplifiers 84a and 84b. Here, a mirror differential amplifier is shown using a pair of integrated instrumentation amplifiers. The output gain is controlled by a single resistor connected to the RG terminal.

[0099] Figure 35 is a detailed circuit diagram of a digital potentiometer-adjustable Miller instrumentation amplifier 84c, which is similar to Figure 34 but utilizes a programmable / digitally selectable gain resistor integrated circuit (IC3) rather than passive components.

[0100] FIG. 36 is a diagram of a handheld analyzer 220 in a large form factor.

[0101] FIG. 37 is a diagram of a small form factor handheld analyzer 220a.

[0102] FIG. 39 is a diagram of a handheld analyzer 220b in a small form factor.

[0103] The sampling and measurement algorithms are designed to minimize noise sources that are not compensated for or otherwise removed using circuit hardware. As shown in block diagram 90 of Figure 38, each "sample" involves reading both the positive and negative differential outputs and subtracting one from the other. Multiple samples can be collected and analyzed via statistical operations to obtain measurements. The simplest form is to calculate the mean and variance / standard deviation from a set of individual samples. The sampling period must be selected in a manner that minimizes the possibility of noise from other sources.

[0104] The main noise sources are floating grounds and ground drift, mains power, and high frequency interference.

[0105] Floating grounds and ground drift are compensated for by various means. Floating grounds (DC noise) are compensated for by the presence of a paired differential amplifier. Ground drift is compensated for by averaging multiple samples. When measuring a positive bias / current, the negative output will be equal to the floating ground. Subtracting the negative output from the positive output removes the noise caused by ground drift. The reverse can be done when measuring a negative bias / current. The subtraction step should be performed on each sample rather than using an average of multiple readings.

[0106] Mains power is also compensated for in various ways. Noise caused by the mains power, either when connected to an AC power line or induced by proximity to other AC line-powered equipment, is compensated for by algorithmic sampling period selection. Sampling should never be performed with a delay equal to the period of the line power cycle (16 or 20 ms for 60 Hz and 50 Hz power systems, respectively) or any multiple thereof (i.e., 32-40 ms for multiples of 2, etc.). If the sampling delay is less than the line power cycle (16-20 ms), at least one cycle (50-60 Hz) must be captured by multiple samples. For proper statistical analysis, sufficient samples must be collected to establish an appropriate estimate of standard deviation and to mitigate power line harmonics. For example, for a 95% confidence interval for Type 1 (false positive) and Type 2 (false negative) errors, at least 13 samples must be measured. Although this is application-specific, a minimum of 10 samples is recommended. The maximum number of samples depends on the application (for body-worn sensors, the possibility of sudden changes due to external factors such as movement).

[0107] High frequency interference, noise caused by radio transmissions and other high frequency signals is completely eliminated by hardware filtering, especially low pass filtering.

[0108] Neural Networks: A wide variety of neural networks (NNs) can be used to both fuse multi-channel signal measurements and reject undesired signals. Inputs to the NN include input measurements, and the network is trained on the desired signal measurement (i.e., interstitial blood glucose level). The network is trained using either supervised or unsupervised learning methods to develop a mathematical model mapping between the signal (i.e., current), temperature, non-analyte and other interference sources, and the target desired analyte signal. Different forms of deep and shallow neural networks can be constructed with combinations of the following layers: recurrent neural networks, convolutional neural networks.

[0109] Convex optimization: In some embodiments, real-time convex optimization is employed to deconvolve the undesired effects by constructing regression cost functions with additional penalty factors in their cost functions to apply prior knowledge of the smoothness of the cross-reference signal or other frequency-based knowledge.

[0110] Projection technology: Projection techniques such as linear and nonlinear (kernel) Principal Component Analysis (PCA) and Independent Component Analysis (ICA) are also employed in selected embodiments for blind source separation. In this case, the input matrix X includes all signals, including analyte-selected and non-analyte-selected signals, as well as any extraneous signal readouts such as temperature. These approaches create a rotation matrix A that maximizes variance (in the case of PCA) and independence (in the case of ICA), resulting in the separation of the input sources.

[0111] Continuous wavelet transform: The Continuous Wavelet Transform (CWT) of the analyte-selective and analyte-invariant sensor measurements is calculated in certain embodiments to construct two-dimensional time-frequency correspondences of the non-analyte and "contaminant" analyte signal measurements. Corresponding frequency coefficients that are time-correlated between the reference non-analyte signal and the contaminant analyte signal are set to zero to remove their effects.

[0112] Non-analyte-derived signal perturbations observable in analyte-selective sensors can claim origin from many physiochemical processes, some of which are endogenous to the biological environment, while others arise due to exogenous effects caused by the wearer of the sensor. Indeed, body-worn analyte-selective sensors often succumb to pressure-induced signal irregularities due to the inadvertent application of pressure or force to the sensor enclosure or housing, referred to as pressure-induced sensor attenuations (PISA). This is often caused by induced changes in perfusion to the sensor or local depletion of the analyte or cofactor of interest, such as oxygen. Disruption of the diffusion layer (nanometer to millimeter range) is also a cause of PISA events, as the sensing enabled by such analyte-selective sensors is inherently diffusion-limited. Performing analyte-invariant measurements allows for the identification of these cases, especially in acute settings, since it is generally understood that the response of the analyte-invariant sensor is largely unaffected by PISA events. Furthermore, all electrochemical sensors undergo a non-Faradic process immediately after excitation by an electrical stimulus, and the subsequent signal response is not proportional to the analyte concentration by the Cottrell relation, but rather to the charging of the double layer capacitance through the solution resistance. This is always present upon excitation of an electrochemical sensor by voltage or current stimuli, R s C dl decays to a negligible level within a finite time according to a time constant, where R s is the solution resistance, and C dlis the double-layer capacitance. Analyte-invariant sensors that undergo the same non-Faradaic signal decay as analyte-selective sensors can be employed in a differential configuration to separate the true analyte signal from the non-Faradaic signal response. Similarly, implanted analyte-selective sensors require a specific duration, referred to as a "warm-up time" or "burn-in," before measuring an accurate indication of the analyte level. The warm-up or burn-in process is a complex physiochemical interaction governed by the interplay between the hydration of the sensor membrane, the establishment of equilibrium between the sensor membrane and the surrounding pore medium, and the adsorption of circulating endogenous proteins (which occupy the pore space) on the sensing surface of the analyte-selective sensor. Analyte-invariant sensors that undergo the same warm-up process as analyte-selective sensors can be employed in a differential configuration to separate the true analyte signal from the non-Faradaic signal response, thus obtaining measurements more timely after sensor application, as shown in Figures 3A-3C and 4.

[0113] Preferred embodiments of the present invention include removing such non-analyte signal perturbations in the system's analog front end, sensor front end, embedded computer, microprocessor, microcontroller, in a wirelessly connected mobile device such as a smartphone, smartwatch, or tablet, or in a cloud service. In other embodiments, the geometry and / or element configuration of the analyte-invariant sensor is identical to the geometry and / or element configuration of the analyte-selective sensor, except that the biorecognition element (i.e., enzyme, antibody, aptamer) is absent. In yet other embodiments, the geometry and / or element configuration of the analyte-invariant sensor is identical to the geometry and / or element configuration of the analyte-selective sensor, except that the biorecognition element (i.e., enzyme, antibody, aptamer) is inactive or has been rendered inactive during the manufacturing process. In yet other embodiments, a system includes multiple analyte-selective sensors and a single analyte-invariant sensor. In yet other embodiments, a system includes multiple analyte-selective sensors, each selective for a unique analyte, and at least one analyte-invariant sensor. In yet other embodiments, the readout from the analyte-invariant sensor is utilized to isolate and remove the temperature dependence of the analyte-selective sensor. In yet other embodiments, the readout from the analyte-invariant sensor is utilized to isolate and remove interference from co-circulating analytes to which the analyte-selective sensor may exhibit partial sensitivity. In yet other embodiments, current methods of mitigation of non-analyte signal perturbations incident on the analyte-selective sensor are employed in sensor fusion algorithms to improve the reliability and / or accuracy of the measurement of the analyte of interest. In yet other embodiments, the analyte-selective sensor and the analyte-invariant sensor occupy the same microneedle material within a microneedle array.

[0114] An "array" is a microneedle or microneedle array-based electrochemical, electro-optical, or fully electronic device configured to measure endogenous or exogenous biochemicals, metabolites, drugs, pharmacological, biological, or pharmaceutical agents within the dermal interstitium that are indicative of a specific physiological or metabolic state in a user's physiological fluid. Specifically, the microneedle array includes a plurality of microneedles having a vertical extent of 200-2000 μm and configured to selectively quantify the level of at least one analyte within the viable epidermis or dermis and located near the papillary, subpapillary, or dermal plexus. The microneedle array is housed in and / or attached to an enclosure or housing that houses a power source, electronic measurement circuitry, a microprocessor, and a wireless transmitter. The sensor is configured with a skin-facing adhesive (sensor adhesive) intended to adhere the sensor for the desired wear period.

[0115] An analyte-selective sensor ("selective sensor") comprises an electrode on the surface of at least one microneedle of the microneedle array, a selective recognition element disposed on the electrode configured to generate a product resulting from interaction of the selective recognition element with an analyte indicative of a particular physiological or metabolic state in the physiological fluid of a user, and a membrane disposed on the selective recognition element. The analyte comprises at least one endogenous or exogenous biochemical, metabolite, drug, pharmacological, biological, or pharmaceutical agent.

[0116] The analyte-invariant sensor ("invariant sensor") is an electrode on the surface of at least one microneedle of the microneedle array that is different from the "selection sensor" and a membrane disposed over the electrode.

[0117] The algorithm ("Algorithm") is a mathematical transformation applied to the electrical response generated at the "Selected Sensor" as a function of the electrical response generated at the "Invariant Sensor" to remove common-mode signals present at both sensors.

[0118] In the method, measurements are recorded at a "selected sensor." Qualitative or quantitative determination of levels of target biomarkers, chemicals, biochemicals, metabolites, electrolytes, ions, hormones, neurotransmitters, vitamins, minerals, drugs, therapeutic agents, toxins, enzymes, proteins, nucleic acids, DNA, or RNA circulating in the user's physiological fluid is performed. An "algorithm" is then applied to the measurements recorded at the "selected sensor" and the "invariant sensor." Thus, a mathematical transformation is applied to the electrical response generated at the "selected sensor" as a function of the electrical response generated at the "invariant sensor" to remove common-mode signals present at both sensors. The algorithm may include at least one of a differencing operation, a noise reduction operation, regression, deconvolution, Fourier decomposition, background subtraction, Kalman filtering, and maximum likelihood estimation.

[0119] Inputs of the present invention include analyte measurements and analyte-invariant measurements. An analyte measurement is a qualitative or quantitative determination of the level of a target biomarker, chemical, biochemical, metabolite, electrolyte, ion, hormone, neurotransmitter, vitamin, mineral, drug, therapeutic agent, toxin, enzyme, protein, nucleic acid, DNA, or RNA circulating in a user's physiological fluid. The measurement is provided by a "selective sensor." An analyte-invariant measurement is a qualitative or quantitative determination of any endogenous or exogenous, stochastic or non-stochastic, physical and / or chemical process, non-analyte-related in origin, incident on an analyte-selective electrochemical sensor. These processes often act to corrupt the measurement signal provided by the analyte-selective sensor. The measurement is provided by an "invariant sensor."

[0120] The output of the present invention is an analyte measurement with the common mode signal removed, which is a qualitative or quantitative measurement of the endogenous level of a particular analyte of interest.

[0121] 18A-18C are diagrams of the sensor. Figure 18C is an exploded rendering of the microneedle sensor 100 showing the main components including the cover 109, the main board 108 with battery, the connector board 107, the microneedle array 110, the base 106 with seal, and the adhesive patch 105.

Claims

**Claim 1**: A device, wherein the device comprises: a microneedle array, the microneedle array comprising a substrate and a plurality of microneedles extending from a first side of the substrate, the plurality of microneedles comprising: a first microneedle comprising a first electrode; a second microneedle comprising a second electrode; a selective recognition element disposed on the first electrode, the selective recognition element being configured to generate a first electrical response resulting from an interaction between the selective recognition element and an analyte; a microneedle array; a processor; wherein: the first microneedle and the second microneedle are configured to be positioned within a living epidermis or dermis of a user; the processor is configured to measure the first electrical response including the first impedance from the first electrode and a second electrical response including the second impedance from the second electrode, the first electrical response and the second electrical response being caused by the application of a bias potential applied to each of the first electrode and the second electrode; the processor is configured to determine a resulting signal representing the first electrical response generated at the first electrode and the second electrical response generated at the second electrode based on a ratio of the first electrical response to the second electrical response. **Claim 2**: The device of claim 1, wherein the first electrode further comprises a first film disposed on the selective recognition element. **Claim 3**: The device of claim 2, wherein the second electrode comprises a second film disposed on the second electrode. **Claim 4**: The device of claim 3, wherein the second film is disposed directly on the second electrode. **Claim 5**: The device of claim 4, wherein the first film and the second film comprise the same material. **Claim 6**: The device of claim 1, wherein the first impedance includes a first frequency-dependent impedance and the second impedance includes a second frequency-dependent impedance. **Claim 7**: The device of claim 1, wherein the bias potential applied to the first electrode and the bias potential applied to the second electrode are the same.

8. The device according to claim 1, wherein the bias potential applied to the first electrode and the bias potential applied to the second electrode comprise an AC potential.

9. The device according to claim 1, wherein the resulting signal includes an analyte signal representing the concentration of the analyte at the first electrode.

10. The device according to claim 1, wherein each of the first electrode and the second electrode is disposed at a distal end of the respective microneedle.

11. A method, the method comprising: providing a first microneedle having a first electrode and a second microneedle having a second electrode, each configured to be inserted into the living epidermis or dermis of a user, and a selective recognition element disposed on the first electrode, the selective recognition element being configured to generate a first electrical response resulting from an interaction between the selective recognition element and an analyte; applying a bias potential to each of the first electrode and the second electrode; measuring the first electrical response from the first electrode and a second electrical response from the second electrode, the first electrical response including a first impedance from the first electrode and the second electrical response including a second impedance from the second electrode; determining a resulting signal representing the first electrical response generated at the first electrode and the second electrical response generated at the second electrode based on a ratio of the first electrical response to the second electrical response. A method comprising the above.

12. The method according to claim 11, wherein the first electrode further comprises a first membrane disposed on the selective recognition element.

13. The method according to claim 12, wherein the second electrode comprises a second membrane disposed on the second electrode.

14. The method according to claim 13, wherein the second membrane is disposed directly on the second electrode.

15. The method according to claim 14, wherein the first membrane and the second membrane comprise the same material.

16. The method according to claim 11, wherein the first impedance includes a first frequency-dependent impedance and the second impedance includes a second frequency-dependent impedance. The method according to claim 11, wherein the bias potential applied to the first electrode and the bias potential applied to the second electrode are the same. The method according to claim 11, wherein the bias potential applied to the first electrode and the bias potential applied to the second electrode comprise an AC potential. The method according to claim 11, wherein the resulting signal includes an analyte signal representing the concentration of the analyte at the first electrode. The method according to claim 11, wherein each of the first electrode and the second electrode is disposed at a distal end of the respective microneedle.

Images