Biomarker detection using graphene field effect transistor sensor arrays and multichannel data acquisition

Abstract

We have developed a CMOS based low-power, multichannel data acquisition IC which interfaces directly with an array of GFET sensors for high-throughput, multiplexed evaluation of clinically-relevant biomarkers in a physiological sample. An example device for detecting a target analyte in a biological sample is provided. The device comprises an integrated circuit (IC) operable for evaluation of the biological samples for the target analyte, a two-dimensional array of graphene field effect transistor (GFET) sensor units where each of the GFET sensor units comprises a graphene surface with one or more probes attached to the graphene surface, and a control circuitry configured to simultaneously control a voltage clamping or a current clamping of each of the GFET sensor units. The biological sample is in contact with the one or more probes and the graphene surface, and the one or more probes bind to the target analyte present in the biological sample.

Description

BIOMARKER DETECTION USING GRAPHENE FIELD EFFECT TRANSISTOR SENSOR ARRAYS AND MULTICHANNEL DATA ACQUISITIONCROSS-REFERENCE TO RELATED APPLICATION

[0001] This patent document claims priority to and benefits of U.S. Provisional Application No. 63 / 515,312, entitled “BIOMARKER DETECTION USING GRAPHENE FIELD EFFECT TRANSISTOR SENSOR ARRAYS AND MULTICHANNEL DATA ACQUISITION,” and filed on July 24, 2023. The entire content of the before-mentioned patent application is incorporated by reference as part of the disclosure of this patent document.TECHNICAL FIELD

[0002] This patent document relates to systems, devices and methods for evaluation of biomarker concentrations in biological samples with high specificity and low power consumption.BACKGROUND

[0003] Graphene field effect transistors (GFETs) have emerged as promising platforms for biochemical sensing due to their unique properties, including high carrier mobility, large surface-to-volume ratio, and exceptional sensitivity to surface charge changes. GFET fabrication techniques employed to create sensing devices include the transfer methods of graphene films onto various substrates and the integration of graphene with functional materials such as metallic contacts, polymers, and other nanomaterials etc. Various architectures and designs can be implemented including back-gated, top-gated, co-planar gated, and dual-gated configurations, emphasizing the impact of these designs on device performance and sensing capabilities. Despite the plethora of fabrication techniques, there still exists a need for low-power hardware that enables multimodal GFET biomarker sensor array deployment for point-of-care, wearable, clinical, and resource constrained environments.SUMMARY

[0004] The disclosed technology relates to systems, devices and methods for evaluation of biomarker concentrations in biological samples with high specificity and low power consumption.

[0005] In some example aspects, a device for detecting a target analyte in a biological sample is provided. In some examples, the device comprises an integrated circuit (IC), comprising row channels and column channels, operable for evaluation of the biological sample for the target analyte. In some examples, the device comprises a two-dimensional array of graphene field effect transistor (GFET) sensor units in electrical communication with the IC, the GFET sensor units arranged, respectively, at row positions along the row channels and column positions along the column channels, each of the GFET sensor units comprising a graphene surface with one or more probes attached to the graphene surface. In some examples, each of the GFET sensor units can be chemically functionalized for specific analyte detection and integrated directly with a CMOS based multichannel clamping and sensing integrated circuit, embedded inside of a standard liquid sample well. In some examples, the device comprises a control circuitry, in electrical communication with the IC, configured to simultaneously control a voltage clamping or a current clamping of each of the GFET sensor units . In some examples, each of the GFET sensor units includes a well to receive the biological sample such that the biological sample is in contact with the one or more probes and the graphene surface. In some examples, the one or more probes bind to the target analyte present in the biological sample. The sensor system has a small modular footprint that can be integrated in a variety of applications including point-of-care diagnostics, clinical screening, wearable monitoring, and resource constrained health care provision.

[0006] In some example aspects, methods are provided for detecting target analytes in a biological sample obtained from a subject by contacting the biological sample with the device of the present technology.

[0007] In other example aspects, methods are provided for detecting a target analyte in a biological sample. In an example, the method comprises (a) obtaining a first set ofcurrent-voltage curves associated with a graphene field effect transistor (GFET) sensor unit in contact with the biological sample, (b) extracting, based on the first set of current-voltage curves, a first feature space comprising a Dirac point of a current-voltage curve, (c) inputting the first feature space to a machine-learning module trained to detect the target analyte in the biological sample, and (d) determining, based on an output of the machine-learning module, whether the biological sample includes the target analyte.

[0008] In yet another example aspect, a device that is configured or operable to perform the above-described methods is disclosed.

[0009] The above and other aspects and their implementations are described in greater detail in the drawings, the descriptions, and the claims.BRIEF DESCRIPTION OF THE DRAWINGS

[0010] FIG. 1 shows an overview of an example system comprised of a GFET sensor 2D array in crossbar arrangement and low-power, low-noise multichannel acquisition integrated circuit system-on-chip.

[0011] FIGS. 2A and 2B show examples of single and multichannel sensor-acquisition hardware interface configurations.

[0012] FIG. 3 shows an example fabrication of a multifunction data acquisition integrated circuit (MDAQIC) and GFET biosensor array system.

[0013] FIG. 4 shows an example fabrication of 2D array of GFET sensors (stack up).

[0014] FIG. 5 shows an example biosensor selective functionalization process.

[0015] FIG. 6 shows an example fabrication and circuit design of a GFET sensor array.

[0016] FIG. 7 shows an example design of a GFET sensor array and a schematic of a controller for the GFET sensor array.

[0017] FIG. 8 shows example images of a fabricated GFET sensor array.

[0018] FIG. 9 shows an example comparison of drain-to-source (IDS) current in voltageclamp mode and voltage (VDS) in current-clamp mode as a function of gate-to-source voltage (VGS) in an example GFET sensor array.

[0019] FIG. 10 shows an example of a Dirac voltage investigation across VGS range in alternating IDS measurement mode of control sample with known biomarker concentration.

[0020] FIG. 11 shows an example schematic of a functionalized GFET and an example of a Dirac shift in an l-V curve.

[0021] FIG. 12 shows an example schematic of a GFET.

[0022] FIG. 13 shows an example of feature sets according to some implementations of the disclosed technology.

[0023] FIG. 14 shows example l-V curves of a GFET sensor.

[0024] FIGS. 15A and 15B show example classifications of GFET curves.

[0025] FIG. 16 shows an example flowchart of a method for detecting a target analyte in a biological sample.DETAILED DESCRIPTION

[0026] Disclosed are methods, devices and systems that pertain to a multimodal graphene field effect transistor (GFET) sensor array chip with complementary metal-oxide- sem iconductor (CMOS) integrated low-power sensing and acquisition IC for high-throughput evaluation of biological samples for clinically relevant biomarkers. The disclosed technology specifies a chip design, layer integration and method for selective functionalization of graphene field effect transistor (GFET) arrays with nucleic acid aptamers or specific antibodies (probes) with high yield. In some embodiments, probe binding with target biomarkers present in the biological sample is highly specific and creates a measurable change in the conductance of the GFET channel. This change in conductance causes a characteristic shift in the Dirac point of the GFET sensor observed in the transfer curves (IDS vs VGS). In some embodiments, the sensing and acquisition system described herein enables precise control of voltage and / or current clamping of an individual GFET sensor inthe array or simultaneous control and acquisition across many sensors in the array (multiplexing). These features are highly tunable to minimize power consumption, maximize sensitivity, decrease detection time and allow assay multiplexing in real time, thereby increasing the statistical robustness of sensing results. Independent GFET sensors can be modulated with different probes that allow multiple biomarkers to be detected from a single GFET array. The integrated clamping and recording acquisition system has a very small footprint and offers many independently addressable channels enabling direct interfacing with an array of GFET sensors.

[0027] The utilization of GFET arrays and multiplexing techniques for simultaneous detection of multiple analytes may be possible, however, challenges such as device stability, reproducibility, and scalability exist. The integration of GFETs with microfluidic systems and the development of portable and miniaturized devices are also gaining interest.

[0028] Strategies for enhancing the performance of GFET -based biochemical sensors can include surface functionalization with specific receptors or probes, incorporation of graphene nanostructures for increased sensitivity, and integration of signal amplification techniques such as enzyme cascades or nanomaterial-based signal amplifiers. Gating (application of channel current controlling voltage to the gate terminal) of GFET sensors is a crucial aspect of its operation as it affects its sensitivity, stability, and overall performance. Proper gating ensures that the GFET operates in the desired region and optimizes its sensitivity for biochemical sensing applications. In addition, application of a voltage clamp for current measurement across the sensors(or current clamp for voltage measurement) is necessary for stable and reproducible results. There are tradeoffs between the two clamping modes with current clamping having an advantage in power savings but with potentially great implementation complexity.

[0029] Few known acquisition systems are able to provide both voltage and current clamp modes with corresponding current and voltage measurements. Far fewer, if any, are available in a small, low-power integrated module, built on an underlying CMOS architecture, offering direct interfacing capabilities to single or array-style GFET sensors.

[0030] While the present disclosure is capable of being embodied in various forms, the description below of several embodiments is made with the understanding that the present disclosure is to be considered as an exemplification of the invention and is not intended to limit the invention to the specific embodiments illustrated.

[0031] Section headings are used in the present document to improve readability of the description and do not in any way limit the discussion or the embodiments (and / or implementations) to the respective sections only.

[0032] Examples of GFET sensor arrays

[0033] In some embodiments, the device of the present technology comprises an array of GFET sensors. Figure 1 depicts an example of a crossbar arrangement of sensors with dual function clamping and sensing channels interfaced along the rows and columns. The arrangement of sensors in this array style allows for the precise application of source-to- drain current and voltage clamps, in addition to corresponding voltage and current measurements made across individual sensors. The crossbar scheme significantly increases the number of individually addressable sensor units by channels of the MDAQIC IC from n to (n / 2)A2. In some embodiments, a MDAQIC channel connected to a column position can supply source current to all GFET sensors within that column but current will only flow through, and therefore create a measurable voltage drop across, the sensor for which the corresponding MDAQIC row channel is selected and configured with an equal but opposite drain current.

[0034] FIG. 2A provides a simplified model for an example operation of the GFET sensor with a VDS clamp voltage and an AC VGS signal applied to the gate. The AC gate signal can be generated external to the system, as depicted in FIG. 2A, or generated on- chip for direct application to the sensor, as depicted in FIG. 2B.

[0035] Examples of devices including GFET sensor arrays

[0036] In some aspects, a device is provided for detecting a target analyte in a biological sample. FIGS. 3 and 4 show examples of a fabrication process for some example embodiments of the GFET sensor array. In some embodiments, the CMOS 32-channelMDAQIC device will be embedded into the carrier package at the center of the device hosting the GFET active sensors and PDMS microfluidic channels. As shown in FIG. 4, a passivation and interconnect scaffold will be patterned in two steps. First, a passivation layer (e.g., Sll-8, polymide, parylene, silicon oxide, silicon nitride, aluminum oxide, silicon dioxide, an epoxy-based negative photoresist, etc.) will be patterned to form vertical cavities on the GFET contacts and on the 32-channel CMOS device. Then, a thick resist photolithography and conformal sputtering metal deposition is used to form metallic vertical interconnects (vias) through the vertical cavities as well as horizontal interconnections between the GFETs and CMOS device. This interconnection is critical to allow for the CMOS device to read and control the array of GFET biosensors for multiplexing sensing. After the interconnects are fabricated, a second passivation layer (e.g., SU-8, polymide, parylene, silicon oxide, silicon nitride, aluminum oxide, silicon dioxide, an epoxy-based negative photoresist, etc.) will be deposited on the interconnects and CMOS device to protect the CMOS device from water and humidity.

[0037] In one such preferred embodiment, graphene is deposited on a silicon wafer with predefined conductive pad and embedded conducting traces that fan out to a standard IC package style such as a ball-grid array (BGA). PBASE is applied to functionalize the surface of the graphene film such that a biomarker-specific DNA-based aptamer or antibody can be subsequently attached. A final passivation step neutralizes areas of unbounded PBase (FIG. 5). Complete sensor array wafers can be bump-bonded onto a BGA footprint onto a flexible or rigid printed circuit board (PCB) substrate followed by a formation of a seal well around the sensor array. Insulated signal traces route from the BGA footprint to the MDAQIC chip located on the underside of the PCB substrate, in another location on the same PCB, or via an interconnect wire to a separate PCB.

[0038] In one example design, the GFET biosensor utilizes a single graphene plane modified with probes (antibodies, aptamers etc.) that specifically bind to a target analyte. The presence of a bound analyte modifies the electric field creating a measurable shift in the Dirac point of the sensor (FIG. 5). In general, the Dirac point refers to a feature of the electronic band structure of graphene and represents a point at which the conduction andvalence bands meet. In some examples, the Dirac point corresponds to a local minimum on the l-V curve and is sometimes referred to as the charge-neutrality point (CNP). A biosensor array offers novel improvements on typical single FET sensors in a number of ways. The first is the ability to replicate and scale results in a single test. This can effectively reduce the noise from GFET chip-to-chip variation. For applications involving multiple bioanalyte detection, the PBASE-modified graphene can be covered with a photoresist. Specific areas of the GFET array will be exposed via photolithography and aptamer followed by ethanolamine passivation will occur as depicted in FIG. 5. This will continue in stages of removing regions of the graphene array until the entire array has been functionalized with specific probes. Another example approach would be to apply specific bias voltages across the target GFETs during probe incubation. The positive bias voltage will preferentially draw the ssDNA aptamers to the target GFET sensors for linkage with the PBASE molecules. The bias voltage will then be applied to the next GFET sensors and the next aptamer will be added. FIG. 5 depicts a design for a 4 chamber (this is scalable) microfluidic PDMS structure that can be glued on top of the array of biosensors. Each chamber can be sequentially filled with fluid allowing for individual GFET biosensors to be modified with a different and specific biomarker probe.

[0039] In another example embodiment, a multiplexed GFET array design is tested using an MDAQIC (e.g., a Neural Interface System-on-Chip (NISoC) microcontroller). A model of the MDAQIC and GFET array design includes an interface chip carrier. This allows for a separate reader to house the MDAQIC and other electronic components for signal generation, transmission, and analysis. Having a separate interface for the MDAQIC and the fabricated GFET array allows for a single MDAQIC chip to test multiple GFET multiplex array biosensors. FIG. 6 (top) shows a depiction of the multiplexed crossbar graphene array with fabrication steps. The end array is wire-bonded on to a custom PCB for connection to the MDAQIC board. FIG. 6 (bottom) shows the circuit design of the present example, with top overlay of exposed pads for wire-bonding, gate electrodes and graphene channels.

[0040] In an exemplary test chip design fabricated in pursuit of graphene array design optimization, the chip consists of a gold vertical crossbar layer connecting the drains of thegraphene FET channels, an SiO2 dielectric layer, and a top gold horizontal crossbar and electrode layer. After which there is an SiO2 passivation layer. Graphene is wet-transferred and patterned using 02 plasma etching, then annealed. Afterwards the chip is adhered to the custom chip carrier PCB and wire-bounded to the appropriate pads. The final step is a microfluidic well placed over the top to separate the regions of graphene channels for independent functionalization with different biological probes. The chip design of the present example consists of 10 drain crossbars with 10 source crossbars placed perpendicularly. There are 100 independent GFET channels that can be individually sampled using the MDAQIC . Having a separate device for sensor analysis, and MDAQIC operation will allow for a single MDAQIC to be used in an reusable reader. In some implementations, the presently described version of the GFET biosensor array can be used as a point-of-care device for diagnostic testing, or at-home testing.

[0041] FIG 7 (left) shows a top level schematic of the interface board for the MDAQIC chip, complete with header pins to drive each channel independently for testing and optimization purposes. FIG. 7 (right) shows the chip carrier PCB board with pads for wirebonding fabricated GFET array design. Edge board connection pads are available to connect with the MDAQIC interface board.

[0042] An example fabrication methodology in the present example embodiment is described herein. Using a 4-inch wafer process, the vertical crossbars for the drain connections are deposited (10nm Cr, 100nm Au) and patterned using a liftoff process. The dielectric layer of SiO2 is then deposited on top using a RF sputter technique and a liftoff reveals the vias for contact between the bottom metal layer and the top (100nm). Then the top metal layer of metal with pads and horizontal crossbars are deposited using a E-beam evaporator (10nm Cr, 200nm Au). The exposed traces are passivated using another liftoff approach with SiO2 sputtered to the surface of the chip (100nm) and the photoresist liftoff revealing the pads and electrodes. Graphene is then wet-transferred to the surface of the chip and patterned using a O2 plasma etch process. The final step is to wire-bond the fabricated test GFET array to the surface of the chip carrier PCB and a PDMS microfluidic well is placed on the surface of the chip. The biological probe linker, PBASE (pyrene butyricacid succinymidil ester), is added to each of the separate wells then incubated for 2 hours. The ssDNA aptamer for each target bioanalyte is added to each specific well. Finally, the unbound PBASE molecules are passivated with ethanolamine for 30 minutes.

[0043] FIG 8 shows an example device fabricated following the above-described methodology. FIG. 8 (top and bottom left) show 5x magnification microscope images of the chip using a 4-inch wafer process. The lighter gold area are the top layer gold pattern, the darker gold are the vertical crossbars underneath the dielectric layer. FIG. 8 (bottom right) shows an image of the chips side by side using the full 4-inch wafer process.

[0044] In some example embodiments, the GFET sensor array includes a gate controlled by a common channel that can be cycled to generate a field-effect along the GFET arrays. In some examples, the gate is controlled in a crossbar configuration perpendicular to source crossbar channels with a common drain channel.

[0045] In some example embodiments, an implantable chip or controller (e.g., NiSoC) is integrated with the two-dimensional array of GFET sensor units to form an implantable GFET array device. The implantable GFET array device may be implanted, for example, in a human subject, an animal subject, or a plant. Some advantages of the implantable GFET array device include real-time detection of biomarker changes (e.g., concentration, chemical / physical features, distribution) for (i) ensuing pathology / physiological abnormalities, (ii) progression of diseases / pathologies, and (iii) efficacy of treatment and preventative measures. Additional advantages include rapid and early detection of external biological, chemical, and physical stimuli and exposures.

[0046] In some example embodiments, the biological samples may include human, animal or plant biological samples.

[0047] Example performance of GFET sensor array devices

[0048] In some embodiments of the present technology (e.g., example devices in FIGS. 1 -4), the sensing IC provides comparable performance to conventional voltageclamped current measurements when configured for current-clamped voltage measurement but with the added utility of precise current level control and much lower power consumption.FIG. 9 shows an example comparison of results recorded in both configurations across Vgs levels. The characteristic Dirac shift is observed for both configurations in the presence of a positive control biomarker. In some embodiments, the sensing IC further features a noise cancellation mode in which a reference noise sample is taken to remove correlated low- frequency noise components from the recorded signal of interest. Reduction of noise in the recording hardware boosts the sensitivity of the sensing system overall and the effective noise level is reduced, enhancing (or increasing) the signal-to-noise ratio (SNR).

[0049] In some implementations, performance of the system in the embodiments is further enhanced with the utilization of the MDAQIC’s charge-balanced current clamping across the GFET sensor. As FIG. 10 depicts, IDS current polarity across the GFET sensor can be switched at rate at least 2 orders of magnitude larger than the VGS frequency. This creates a DS voltage measured across the sensor with a similar square-wave pattern. In alternating current, this measurement mode more effectively reduces the effects of sensor hysteresis, offset, and drift seen in such devices when clamped with a DC current level. Alternating VDS measurements are subtracted from each other to produce an essentially offset free voltage level with a clearly distinguishable Dirac point. By performing noise cancellation in the sensing hardware and noise mitigation with charge balancing in the sensor itself, the overall sensitivity of the device is boosted as a result of enhanced SNR.

[0050] Example of using a GFET sensor device for COVID-19 testing

[0051] In the wake of the COVID-19 pandemic, there has been a need for reliable diagnostic testing. However, state-of-the-art detection methods rely on laboratory tests and also vary in accuracy. Described herein is an example of a machine learning approach to COVID-19 detection via graphene-field-effect-transistor (GFET). The usage of a GFET coupled with machine learning can be a promising alternate diagnostic testing method. In the description that follows, current-voltage data gathered from the GFET sensors is processed to assess information about the presence of COVID-19 in biosamples. A binary classification was performed using the following machine learning algorithms: linear discriminant dnalysis (LDA), support vector machines (SVM) with the radial basis function (RBF) kernel, and K-nearest neighbors (KNN) in conjunction with principal componentanalysis (PCA). It is found that LDA and SVM with RBF are the most accurate in identifying positive and negative samples, with accuracies of 99% and 98.5%, respectively. Based on these results, there is a promise to develop a bioelectronic diagnostic method for COVID-19 detection by combining GFET technology with machine learning.

[0052] Testing oneself for SARS-CoV-2 during the COVID-19 pandemic has been a crucial step toward promoting global health. The two most common state-of-the-art testing methods include laboratory reverse transcription polymerase chain reaction (RT-PCR) and rapid antigen testing. However, one of the limitations of PCR testing is its reliance on centralized laboratories to get results. Additionally, compared to the other methods, rapid antigen testing has a relatively high tendency to spark false negatives. As such, there is a need to develop alternative methods of COVID-19 detection, such as through applications of electronics, artificial intelligence, and other indirect detection techniques.

[0053] In one example implementation, a GFET sensor based on the disclosed technology is used for COVID-19 diagnostic testing. The GFET sensor utilizes an electric field to detect changes in biomolecule concentration. The GFET includes three pins - the source, drain, and gate. By applying a range of gate voltages, the output drain current creates a current-voltage (l-V) curve. For viral detection, the GFET is functionalized with ssDNA (single-stranded DNA) aptamers that bind to a specific target protein on the virus of interest. Through this, the viral analytes bind to the aptamers located on the GFET to cause a change in drain current. This shift in the drain current is known as the Dirac shift.

[0054] FIG. 11 provides an example depiction of the effect of binding a viral analyte onto the aptamer on the l-V curve. Specifically, FIG. 11 shows a schematic diagram of the surface of a functionalized GFET showing an aptamer with selective binding affinity to the analyte, in this case a COVID-19 protein. A binding of an analyte causes a Dirac shift in the l-V curve.

[0055] In some implementations, interfacing a GFET system with machine learning can be an effective diagnostic testing platform for COVID-19 detection.Example Preparation of GFET System

[0056] The GFET sensors in the working example were manufactured in an external facility. The GFETs were functionalized with the aptamer for the SARS-CoV-2 nucleocapsid (N) protein. 10pL of phosphate-buffered saline (PBS) containing 1 M of the aptamer was added to the GFET chips.

[0057] FIG. 12 shows an example image and schematic diagram of the surface of a graphene field-effect-transistor (GFET) showing voltages input at the gate (VGS) and drain (VDS), and current output at the drain (IDS). In accordance with the schematic shown in FIG. 12, the drain-source voltage (VDS) was fixed at 100mV. The gate-source voltage (VGS) was swept from -0.5 V to +1.5 V, referred to herein as a forward sweep, and then swept from +1.5 V to -0.5 V, referred to herein as a backward sweep. This forward / backward sweep sequence was carried out four times (4 forward and 4 backward sweeps total). The drain- to-source current (IDS) was recorded to create the l-V characteristic curve from the VGS voltage sweep. The l-V curve illustrates the Dirac point (global minimum of the l-V curve) of each sweep, which has been used extensively as an indicator of aptamer-target binding events. A visual depiction of the input pins for VDS and VGS, output pin for IDS, and the holding well for the biosample is also shown in FIG. 12.

[0058] As discussed earlier, FIG. 2A shows an example instrumentation diagram of the GFET measurement. As shown therein, a computer-controlled dual-channel source meter was used to drive the VGS waveform and simultaneously record the IDS signal. The GFET chips were tested under the diagram shown in FIG. 2A to obtain their l-V curves. To collect data, a connection from the source measuring unit (SMU) to the GFET sensor was accomplished by ensuring metal contact with the three GFET sensor pins.Example Data Collection

[0059] In some embodiments, machine learning methods were used to differentiate between viral and non-viral samples. The efficacy of the ML methods was demonstrated using l-V curves for 19 test samples. Nine of the samples contained the Omicron strain of COVID-19, and are referred to as positive samples. The other 10 test samples are negative samples that contain no traces of COVID-19.

[0060] Before adding the samples, 0.1X PBS was added to the GFETs and ran through the GFET interface shown in FIG. 2A to obtain their l-V curves. This initial curve generation will be defined as the baseline run. The saliva-virus samples were then added to the GFET, and after a few minutes of incubation for aptamer-target binding, the l-V curves were obtained. This will be defined as the sample run.

[0061] l-V curves were generated for all 19 samples to obtain their baseline and sample runs. 1000 l-V points were collected from each run on a GFET sensor. Here, each run was comprised of 8 voltage sweeps (4 forward sweeps and 4 backward sweeps, as mentioned in the example preparation method of a GFET system described above).

[0062] The data being used was derived from each GFET sensor configured with either the SMU or the PIVOT, a handheld device that can run the GFET sensor and output the curves. The data includes outputs from both baseline and sample runs for the given sensor. All the positive samples used in this study were run on GFET sensors that were functionalized with an aptamer for the Omicron variant of COVID-19.Example Pre-processing and Feature Extraction

[0063] The obtained l-V curves are separated into forward and backward sweeps, and for each sweep, three feature spaces are created for use in the machine learning model.

[0064] The first feature space is referred to as the Dirac set, which involves identifying the l-V point corresponding to the Dirac point of each sweep. The Dirac set is created by extracting the Dirac point from a single baseline sweep and a single sample sweep taken from the same chip and experimental test. Four dimensions constitute the Dirac set: voltage corresponding to the baseline Dirac point, voltage corresponding to the sample Dirac point, current corresponding to the baseline Dirac point, and current corresponding to the sample Dirac point.

[0065] The second feature space, the Curve Estimation set, also involves finding the Dirac point of the sweeps, as well as taking 4 evenly spaced l-V points between the Dirac point and each end of the curve (9 points total). These spaced-out points were used to create a 36-dimensional feature space (9 points multiplied by 4 dimensions in a feature space).Lastly, the All Points set was created, in which all the points from the l-V curve were taken to create the feature space. These points create a 4*N dimensional feature space, where N is the total number of l-V points. FIG. 13 displays a visual overview of all feature sets.Example Model Training and Testing

[0066] Three different models were trained and tested using the above-described feature sets to explore the possibility of classifying positive and negative samples based on GFET curves. The three models that were trained and tested in this paper include Linear Discriminant Analysis (LDA), Support Vector Machine with the Radial Basis Function (RBF) kernel, and K-Nearest Neighbors (KNN) with Principal Component Analysis (PCA). These models were tested using ten-fold cross-validation. Accuracy, precision, and recall for all of the models are reported in the description that follows.Example Visual Inspection of l-V Samples

[0067] FIG. 14 illustrates the visual differences between negative and positive samples. FIG. 14 shows l-V curves from a single run of a positive sample (above) and a negative control sample (below) applied to the GFET sensor.

[0068] As shown in FIG. 14, the entire run of one of the negative and positive samples is plotted. It can be seen from FIG. 14 that the baseline and sample sweep in the negative case overlaps with each other. However, there appears to be a horizontal shifting of the sample sweep from the baseline sweep in the positive case. This change makes sense in the positive case since the addition of the COVID-19 biosample modulates the transfer of charges across the graphene.

[0069] The conventional approach to GFET curve classification entails evaluating the absolute difference in Dirac voltage between the baseline and sample curves, followed by establishing a linear decision boundary for positive and negative sample classification. FIG. 15A shows the manual classification of the absolute value of the Dirac voltage differences between the baseline and sample runs. The manual decision boundary created in FIG. 15A maximizes the testing accuracy. Using this method, an accuracy of 68.4% was achieved.

[0070] FIG. 15B shows classifications of the Dirac and Curve Estimation sets using LDA. In particular, FIG. 15B displays one testing fold of the class labels that LDA predicts for the Dirac set and Curve Estimation set, respectively. Only the Dirac points of the sample curve along their linear discriminant were plotted. Here, a clear separation between positive and negative samples is seen.Example Performance of Machine Learning Models

[0071] Tables I, II, III report the accuracy, precision, and recall from the Dirac, Curve Estimation, and All Points set.TABLE ICLASSIFICATION RESULTS FOR THE DIRAC FEATURE SETTABLE IICLASSIFICATION RESULTS FOR THE CURVE ESTIMATION FEATURE SETTABLE IIICLASSIFICATION RESULTS FOR THE ALL POINTS FEATURE SET

[0072] It can be seen in Table I that SVM with RBF recorded the highest accuracy using the Dirac Feature set at 98.5%. In Table II, LDA recorded the highest accuracy using the Curve Estimation Feature set at 96.2%. Finally, in Table III, LDA recorded the highest accuracy using the All Points Feature set at 99%.Example Viral Sample Classification

[0073] When using LDA, accuracy, precision, and recall all significantly improved as a result of adding additional points to characterize the curves rather than just the Dirac points. This was most likely because LDA assumes that the given classes are linearly separable, and performance generally improves upon inputting additional data. Additionally, SVM interfaced with RBF performed exceptionally well in the Dirac set. This is most likely because fitting a nonlinear decision boundary allows for an increase in the complexity of the model to accommodate the limited feature space.

[0074] Despite the limited dataset used, the high accuracy, precision, and recall scores achieved in all feature sets suggest that GFET sensors interfaced with machine learning could be a promising pipeline for effective COVID-19 diagnostic testing. With further optimization, it may be possible to develop a multi-class classification model that is able to detect distinct mutations in a virus using a single aptamer or detect entirely different viruses besides COVID-19.

[0075] In some embodiments, optimizing the data collection process results in improving the performance of the GFET. For example, as an alternative to maintaining a constant VDS, one could vary VGS while inputting an alternating VDS. Furthermore, instead of using VDS and VGS as inputs, an alternating IDS and VGS could be utilized in some implementations to obtain an output of VDS. These variations could possibly reduce sensor drift, which ultimately would lead to better performance of the sensor. Optimizing GFET parameters could potentially improve classification accuracy on machine learning results.

[0076] In the working example described above, it is shown that GFET sensors interfaced with machine learning have promising potential for effective COVID-19 detection. A binary classification was performed on samples that were positive and negative for COVID-19 using current-voltage data generated from the GFETs on the following algorithms: LDA, SVM with the RBF kernel, and KNN with PCA. It is found that using LDA as well as SVM with RBF achieved accuracies of 99% and 98.5%, respectively.

[0077] Embodiments and implementations of the disclosed technology

[0078] The disclosed technology provides a method and system for the high- throughput evaluation of biomarker concentrations in a sample that has not, and cannot, be achieved using existing implementations and systems. In some aspects, the integration of a functionalized GFET sensor array directly with a multichannel acquisition system offers unique advantages in sensitivity, power consumption, and device size. The interface method presented is a considerable improvement over conventional connections between sensing hardware and sensor arrays. The methods presented for control and sensing in the GFET sensor are unique and optimized for power saving and biomarker detection sensitivity. Specifically, the GFET IDS current when clamped in a sub-milliamp range can offer comparable sensitivity to a VDS voltage clamp but at a fraction of the power. Additionally, offset, noise, and hysteresis in the GFET sensor can effectively be reduced by performing an alternating positive and negative IDS current clamp. Noise cancellation is also performed in the acquisition hardware by various means. Low-frequency noise contributions and DC offset in the acquisition channel are canceled.

[0079] FIG. 16 a flowchart for an example method 160 for detecting a target analyte in a biological sample. The method includes, at operation 162, obtaining a first set of current-voltage curves associated with a graphene field effect transistor (GFET) sensor unit in contact with the biological sample. The method includes, at operation 164, extracting, based on the first set of current-voltage curves, a first feature space comprising a Dirac point of a current-voltage curve. The method includes, at operation 166, inputting the first feature space to a machine-learning module trained to detect the target analyte in the biological sample. The method includes, at operation 168, determining, based on anoutput of the machine-learning module, whether the biological sample includes the target analyte.

[0080] Embodiments of the disclosed technology support inter alia the following technical solutions that solve the technical problem of accurately detecting a target analyte in biological samples.

[0081] 1 . A device for detecting a target analyte in a biological sample, comprising an integrated circuit (IC), comprising row channels and column channels, operable for evaluation of the biological sample for the target analyte, a two-dimensional array of graphene field effect transistor (GFET) sensor units in electrical communication with the IC, the GFET sensor units arranged, respectively, at row positions along the row channels and column positions along the column channels, each of the GFET sensor units comprising a graphene surface with one or more probes attached to the graphene surface, anda control circuitry, in electrical communication with the IC, configured to simultaneously control a voltage clamping or a current clamping of each of the GFET sensor units, wherein each of the GFET sensor units includes a well to receive the biological sample such that the biological sample is in contact with the one or more probes and the graphene surface, wherein the one or more probes bind to the target analyte present in the biological sample.

[0082] 2. The device of solution 1 , wherein the target analyte is a biomarker or a pathogen.

[0083] 3. The device of solution 1 , wherein at least one of the one or more probes is a nucleic acid aptamer.

[0084] 4. The device of solution 1 , wherein at least one of the one or more probes is an antibody.

[0085] 5. The device of solution 1 , wherein the target analyte is a COVID-19 protein.

[0086] 6. The device of solution 1 , wherein the row channels and the column channels intersect with one another to form the two-dimensional array of GFET sensors such that the two-dimensional array of GFET sensor units has a crossbar configuration.

[0087] 7. The device of solution 1 , wherein the binding of the one or more probes to the target analyte creates a change in a conductance of at least one of the row channels or the column channels.

[0088] 8. The device of solution 7, wherein the change in the conductance causes a shift in a Dirac point of a current-voltage curve associated with at least one of the GFET sensor units.

[0089] 9. The device of solution 1 , wherein the IC includes at least one channel operable to connect to at least one of the column positions and supply electrical current to the GFET sensor units within the column channel corresponding to the at least one of the column positions such that a voltage drop is created across a selected GFET sensor unit.

[0090] 10. The device of solution 1 , wherein a 1 -pyrenebutanoic acid succinimidyl ester (PBASE) is applied to the graphene surface to functionalize the graphene surface and enable attachment of the one or more probes to the graphene surface.

[0091] 11 . The device of solution 1 , wherein the control circuitry is further configured to individually sample each of the GFET sensor units and detect signals thereof.

[0092] 12. The device of solution 1 , wherein the two-dimensional array of GFET sensor units is located on a chip carrier comprising a plurality of electrically conductive pins that provide electrical pathways from the row channels and the column channels to the controller device, wherein the two-dimensional array of GFET sensor units is in electrical communication with the controller device via electrical connection of the chip carrier to a printed circuit board of the controller device.

[0093] 13. The device of any one of solutions 1 to 12, wherein the graphene surface is functionalized with an aptamer that binds to a target bioanalyte comprising a virus, a protein, a toxin, or a molecule.

[0094] 14. The device of solution 13, wherein the aptamer is a single-stranded DNA.

[0095] 15. The device of any one of solutions 13 to 14, wherein the protein is aCOVID-19 protein.

[0096] 16. The device of any one of solutions 13 to 14, wherein the virus is a coronavirus.

[0097] 17. The device of solution 1 , wherein the biological sample is a liquid.

[0098] 18. The device of solution 1 , wherein the current clamping is performed using an alternating positive and negative current clamp.

[0099] 19. The device of solution 18, wherein the alternating positive and negative current clamp reduces one or more of an offset, a noise, or a hysteresis of the GFET sensor units.

[0100] 20. The device of solution 1 , wherein the control circuitry is configured to control the voltage clamping by applying a voltage clamp to at least one of the GFET sensor units, wherein the control circuitry is configured to control the current clamping by applying a current clamp to at least one of the GFET sensor units.

[0101] 21 . The device of solution 20, wherein the applying the voltage clamp facilitates a measurement of current across the at least one of the GFET sensor units, wherein the applying the current clamp facilitates a measurement of voltage across the at least one of the GFET sensor units.

[0102] 22. The device of solution 21 , wherein the current clamp is in a sub-milliamp range.

[0103] 23. The device of solution 1 , wherein the control circuitry is configured to perform a noise cancellation operation on a signal produced by the device by removing a low-frequency noise contribution and a DC offset from the signal.

[0104] 24. The device of solution 23, wherein the noise cancellation increases a sensitivity of the device by increasing a signal-to-noise ratio of the device.

[0105] 25. The device of solution 1 , wherein the control circuity is integrated with the two-dimensional array of GFET sensor units, thereby making the device implantable.

[0106] 26. The device of solution 1 , wherein each of the GFET sensor units includes a source terminal, a drain terminal, and a gate terminal, wherein the two-dimensional array of GFET sensor units has a crossbar configuration comprising drain crossbars in communication with drain terminals of the GFET sensor units and source crossbars in communication with source terminals of the GFET sensor units.

[0107] 27. The device of solution 26, wherein the gate terminal of each of the GFET sensor units is associated with a gate electrode, and wherein gate electrodes of the GFET sensor units and the source crossbars are located in a same plane of the device.

[0108] 28. The device of solution 26, wherein the source terminal, the drain terminal, and the gate terminal are located in a same plane of the device.

[0109] 29. The device of claim 26, wherein the gate terminal of each of the GFET sensor units is associated with a gate electrode, and wherein gate electrodes of the GFET sensor units and the source crossbars form another crossbar configuration.

[0110] 30. The device of solution 1 , wherein each of the GFET sensor units includes a source terminal, a drain terminal, and a gate terminal, wherein the two-dimensional array of GFET sensor units has a crossbar configuration comprising gate electrodes associated with gate terminals of the GFET sensor units and source crossbars in communication with source terminals of the GFET sensor units, wherein the gate electrodes and the source crossbars are perpendicular to one another and share a drain connection.

[0111] 31 . A method of detecting target analytes in a biological sample obtained from a subject, the method comprising contacting the biological sample with the device of any one of solutions 1 to 30.

[0112] 32. A method for detecting a target analyte in a biological sample, comprising obtaining a first set of current-voltage curves associated with a graphene field effect transistor (GFET) sensor unit in contact with the biological sample, extracting, based on the first set of current-voltage curves, a first feature space comprising a Dirac point of a current-voltage curve, inputting the first feature space to a machine-learning module trained to detect the target analyte in the biological sample, and determining, based on an output of the machine-learning module, whether the biological sample includes the target analyte.

[0113] 33. The method of solution 32, wherein a training set for the machine-learning module comprises a second feature space extracted from a second set of current-voltage curves associated with the GFET sensor unit in contact with a reference biological sample containing the target analyte and a third feature space extracted from a third set of current-voltage curves associated with the GFET sensor unit in contact with only the reference biological sample.

[0114] 34. The method of solution 33, wherein the second feature space comprises a set of datapoints located on at least one current-voltage curve from the second set of current-voltage curves.

[0115] 35. The method of solution 34, wherein the datapoints are evenly spaced apart from one another between a Dirac point of the at least one current-voltage curve and an end point of the at least one current-voltage curve.

[0116] 36. The method of solution 32 or 33, wherein obtaining the first set of currentvoltage curves comprises performing a forward sweep of voltage applied to a gate terminal of the GFET sensor unit and performing a backward sweep of voltage applied to the gate terminal of the GFET sensor unit.

[0117] 37. The method of solution 36, wherein the forward sweep and the backward sweep are performed an equal number of times.

[0118] 38. The method of solution 32, wherein the machine-learning module comprises a linear discriminant analysis (LDA) operation, a support vector machine (SVM) with a radial basis function (RBF) kernel, k-nearest neighbors (KNN) with principal component analysis (PCA) implementation, or a deep-learning neural network.

[0119] 39. A device comprising one or more processors, wherein the one or more processors are configured to cause the device to perform the method in one or more of claims 32 to 38.

[0120] Conclusion

[0121] The above detailed description of embodiments of the technology are not intended to be exhaustive or to limit the technology to the precise forms disclosed above. Although specific embodiments of, and examples for, the technology are described above for illustrative purposes, various equivalent modifications are possible within the scope of the technology as those skilled in the relevant art will recognize. For example, although steps are presented in a given order, alternative embodiments may perform steps in adifferent order. The various embodiments described herein may also be combined to provide further embodiments.

[0122] Implementations of the subject matter and the functional operations described in this patent document can be implemented in various systems, digital electronic circuitry, or in computer software, firmware, or hardware, including the structures disclosed in this specification and their structural equivalents, or in combinations of one or more of them. Implementations of the subject matter described in this specification can be implemented as one or more computer program products, i.e., one or more modules of computer program instructions encoded on a tangible and non-transitory computer readable medium for execution by, or to control the operation of, data processing apparatus. The computer readable medium can be a machine-readable storage device, a machine-readable storage substrate, a memory device, a composition of matter effecting a machine-readable propagated signal, or a combination of one or more of them. The term “data processing unit” or “data processing apparatus” encompasses all apparatus, devices, and machines for processing data, including by way of example a programmable processor, a computer, or multiple processors or computers. The apparatus can include, in addition to hardware, code that creates an execution environment for the computer program in question, e.g., code that constitutes processor firmware, a protocol stack, a database management system, an operating system, or a combination of one or more of them.

[0123] A computer program (also known as a program, software, software application, script, or code) can be written in any form of programming language, including compiled or interpreted languages, and it can be deployed in any form, including as a stand-alone program or as a module, component, subroutine, or other unit suitable for use in a computing environment. A computer program does not necessarily correspond to a file in a file system. A program can be stored in a portion of a file that holds other programs or data (e.g., one or more scripts stored in a markup language document), in a single file dedicated to the program in question, or in multiple coordinated files (e.g., files that store one or more modules, sub programs, or portions of code). A computer program can be deployed to beexecuted on one computer or on multiple computers that are located at one site or distributed across multiple sites and interconnected by a communication network.

[0124] The processes and logic flows described in this specification can be performed by one or more programmable processors executing one or more computer programs to perform functions by operating on input data and generating output. The processes and logic flows can also be performed by, and apparatus can also be implemented as, special purpose logic circuitry, e.g., an FPGA (field programmable gate array) or an ASIC (application specific integrated circuit).

[0125] Processors suitable for the execution of a computer program include, by way of example, both general and special purpose microprocessors, and any one or more processors of any kind of digital computer. Generally, a processor will receive instructions and data from a read only memory or a random access memory or both. The essential elements of a computer are a processor for performing instructions and one or more memory devices for storing instructions and data. Generally, a computer will also include, or be operatively coupled to receive data from or transfer data to, or both, one or more mass storage devices for storing data, e.g., magnetic, magneto optical disks, or optical disks. However, a computer need not have such devices. Computer readable media suitable for storing computer program instructions and data include all forms of nonvolatile memory, media and memory devices, including by way of example semiconductor memory devices, e.g., EPROM, EEPROM, and flash memory devices. The processor and the memory can be supplemented by, or incorporated in, special purpose logic circuitry.

[0126] While this patent document contains many specifics, these should not be construed as limitations on the scope of any invention or of what may be claimed, but rather as descriptions of features that may be specific to particular embodiments of particular inventions. Certain features that are described in this patent document in the context of separate embodiments can also be implemented in combination in a single embodiment. Conversely, various features that are described in the context of a single embodiment can also be implemented in multiple embodiments separately or in any suitable subcombination. Moreover, although features may be described above as acting in certain combinations andeven initially claimed as such, one or more features from a claimed combination can in some cases be excised from the combination, and the claimed combination may be directed to a subcombination or variation of a subcombination.

[0127] Similarly, while operations are depicted in the drawings in a particular order, this should not be understood as requiring that such operations be performed in the particular order shown or in sequential order, or that all illustrated operations be performed, to achieve desirable results. Moreover, the separation of various system components in the embodiments described in this patent document should not be understood as requiring such separation in all embodiments.

[0128] Only a few implementations and examples are described and other implementations, enhancements and variations can be made based on what is described and illustrated in this patent document.

Claims

CLAIMSWhat is claimed is:1 . A device for detecting a target analyte in a biological sample, comprising: an integrated circuit (IC), comprising row channels and column channels, operable for evaluation of the biological sample for the target analyte; a two-dimensional array of graphene field effect transistor (GFET) sensor units in electrical communication with the IC, the GFET sensor units arranged, respectively, at row positions along the row channels and column positions along the column channels, each of the GFET sensor units comprising a graphene surface with one or more probes attached to the graphene surface; and a control circuitry, in electrical communication with the IC, configured to simultaneously control a voltage clamping or a current clamping of each of the GFET sensor units, wherein each of the GFET sensor units includes a well to receive the biological sample such that the biological sample is in contact with the one or more probes and the graphene surface, wherein the one or more probes bind to the target analyte present in the biological sample.

2. The device of claim 1 , wherein the target analyte is a biomarker or a pathogen.

3. The device of claim 1 , wherein at least one of the one or more probes is a nucleic acid aptamer.

4. The device of claim 1 , wherein at least one of the one or more probes is an antibody.

5. The device of claim 1 , wherein the target analyte is a COVID-19 protein.

6. The device of claim 1 , wherein the row channels and the column channels intersect with one another to form the two-dimensional array of GFET sensors such that the two-dimensional array of GFET sensor units has a crossbar configuration.

7. The device of claim 1 , wherein the binding of the one or more probes to the target analyte creates a change in a conductance of at least one of the row channels or the column channels.

8. The device of claim 7, wherein the change in the conductance causes a shift in a Dirac point of a current-voltage curve associated with at least one of the GFET sensor units.

9. The device of claim 1 , wherein the IC includes at least one channel operable to connect to at least one of the column positions and supply electrical current to the GFET sensor units within the column channel corresponding to the at least one of the column positions such that a voltage drop is created across a selected GFET sensor unit.

10. The device of claim 1 , wherein a 1 -pyrenebutanoic acid succinimidyl ester (PBASE) is applied to the graphene surface to functionalize the graphene surface and enable attachment of the one or more probes to the graphene surface.11 . The device of claim 1 , wherein the control circuitry is further configured to individually sample each of the GFET sensor units and detect signals thereof.

12. The device of claim 1 , wherein the two-dimensional array of GFET sensor units is located on a chip carrier comprising a plurality of electrically conductive pins that provide electrical pathways from the row channels and the column channels to thecontroller device, wherein the two-dimensional array of GFET sensor units is in electrical communication with the controller device via electrical connection of the chip carrier to a printed circuit board of the controller device.

13. The device of any one of claims 1 to 12, wherein the graphene surface is functionalized with an aptamer that binds to a target bioanalyte comprising a virus, a protein, a toxin, or a molecule.

14. The device of claim 13, wherein the aptamer is a single-stranded DNA.

15. The device of any one of claims 13 to 14, wherein the protein is a COVID-19 protein.

16. The device of any one of claims 13 to 14, wherein the virus is a coronavirus.

17. The device of claim 1 , wherein the biological sample is a liquid.

18. The device of claim 1 , wherein the current clamping is performed using an alternating positive and negative current clamp.

19. The device of claim 18, wherein the alternating positive and negative current clamp reduces one or more of an offset, a noise, or a hysteresis of the GFET sensor units.

20. The device of claim 1 , wherein the control circuitry is configured to control the voltage clamping by applying a voltage clamp to at least one of the GFET sensor units, wherein the control circuitry is configured to control the current clamping by applying a current clamp to at least one of the GFET sensor units.21 . The device of claim 20, wherein the applying the voltage clamp facilitates a measurement of current across the at least one of the GFET sensor units, whereinthe applying the current clamp facilitates a measurement of voltage across the at least one of the GFET sensor units.

22. The device of claim 21 , wherein the current clamp is in a sub-milliamp range.

23. The device of claim 1 , wherein the control circuitry is configured to perform a noise cancellation operation on a signal produced by the device by removing a low- frequency noise contribution and a DC offset from the signal.

24. The device of claim 23, wherein the noise cancellation increases a sensitivity of the device by increasing a signal-to-noise ratio of the device.

25. The device of claim 1 , wherein the control circuity is integrated with the two- dimensional array of GFET sensor units, thereby making the device implantable.

26. The device of claim 1 , wherein each of the GFET sensor units includes a source terminal, a drain terminal, and a gate terminal, wherein the two-dimensional array of GFET sensor units has a crossbar configuration comprising drain crossbars in communication with drain terminals of the GFET sensor units and source crossbars in communication with source terminals of the GFET sensor units.

27. The device of claim 26, wherein the gate terminal of each of the GFET sensor units is associated with a gate electrode, and wherein gate electrodes of the GFET sensor units and the source crossbars are located in a same plane of the device.

28. The device of claim 26, wherein the gate terminal of each of the GFET sensor units is associated with a gate electrode, and wherein gate electrodes of the GFET sensor units and the source crossbars form another crossbar configuration.

29. The device of claim 26, wherein the source terminal, the drain terminal, and the gate terminal are located in a same plane of the device.

30. The device of claim 1 , wherein each of the GFET sensor units includes a source terminal, a drain terminal, and a gate terminal, wherein the two-dimensional array of GFET sensor units has a crossbar configuration comprising gate electrodes associated with gate terminals of the GFET sensor units and source crossbars in communication with source terminals of the GFET sensor units, wherein the gate electrodes and the source crossbars are perpendicular to one another and share a drain connection.31 . A method of detecting target analytes in a biological sample obtained from a subject, the method comprising contacting the biological sample with the device of any one of claims 1 to 30.

32. A method for detecting a target analyte in a biological sample, comprising: obtaining a first set of current-voltage curves associated with a graphene field effect transistor (GFET) sensor unit in contact with the biological sample; extracting, based on the first set of current-voltage curves, a first feature space comprising a Dirac point of a current-voltage curve; inputting the first feature space to a machine-learning module trained to detect the target analyte in the biological sample; and determining, based on an output of the machine-learning module, whether the biological sample includes the target analyte.

33. The method of claim 32, wherein a training set for the machine-learning module comprises: a second feature space extracted from a second set of current-voltage curves associated with the GFET sensor unit in contact with a reference biological sample containing the target analyte; and a third feature space extracted from a third set of current-voltage curves associated with the GFET sensor unit in contact with only the reference biological sample.

34. The method of claim 33, wherein the second feature space comprises a set of datapoints located on at least one current-voltage curve from the second set of currentvoltage curves.

35. The method of claim 34, wherein the datapoints are evenly spaced apart from one another between a Dirac point of the at least one current-voltage curve and an end point of the at least one current-voltage curve.

36. The method of claim 32 or 33, wherein obtaining the first set of current-voltage curves comprises: performing a forward sweep of voltage applied to a gate terminal of the GFET sensor unit; and performing a backward sweep of voltage applied to the gate terminal of the GFET sensor unit.

37. The method of claim 36, wherein the forward sweep and the backward sweep are performed an equal number of times.

38. The method of claim 32, wherein the machine-learning module comprises a linear discriminant analysis (LDA) operation, a support vector machine (SVM) with a radial basis function (RBF) kernel, k-nearest neighbors (KNN) with principal component analysis (PCA) implementation, or a deep-learning neural network.

39. A device comprising one or more processors, wherein the one or more processors are configured to cause the device to perform the method in one or more of claims 32 to 38.

Images