Electrode device, method of forming the same, and signal collection platform

Abstract

Various embodiments relate to an electrode device The electrode device includes a substrate, one or more working electrodes on the substrate, a counter electrode on the substrate, and a reference electrode on the substrate, wherein the one or more working electrodes, the counter electrode, and the reference electrode are spaced from one other, and wherein the one or more working electrodes, the counter electrode, and the reference electrode include carbon nanotubes (CNT) and cellulose, a weight ratio of the carbon nanotubes (CNTs) to the cellulose of a ratio from a range from 1:8 to 1:2. Various embodiments relate to a method of making the electrode device, a signal collection platform, and a method of forming the signal collection platform.

Description

ELECTRODE DEVICE, METHOD OF FORMING THE SAME, AND SIGNAL COLLECTION PLATFORMCROSS-REFERENCE TO RELATED APPLICATION

[0001] This application claims the benefit of priority of Singapore application No. 10202302141W filed July 27, 2023, the contents of it being hereby incorporated by reference in its entirety for all purposes.TECHNICAL FIELD

[0002] Various embodiments of this disclosure mmm realyate to an electrode device. Various embodiments may relate to a method of forming an electrode device. Various embodiments of this disclosure may relate to a signal collection platform. Various embodiments of this disclosure may relate to a method of forming a signal collection platform.BACKGROUND

[0003] Physiological status profiling in complex healthcare scenarios with distributed sensing platforms has gained significance due to recent pandemics and the anticipated ageing of the population. Conventional electrochemical sensors employ diverse calibration methods for biomarker quantification in solutions. However, accuracy diminishes in multi-component biofluid analysis due to the reliance of electrochemical theories on ideal conditions. Moreover, most calibration processes focus on identified electroactive metabolites, overlooking the abundant bioinformation contained in biofluids and thereby curtailing the diagnostic capabilities of these sensors. These limitations are particularly pronounced in peripheral biofluids like saliva and sweat, which are often subject to environmental and individual variability, leading to weaker diagnostic correlations than those achieved with traditional blood or biopsy samples.SUMMARY

[0004] Various embodiments may relate to an electrode device. The electrode device may include a substrate, one or more working electrodes on the substrate, a counter electrode on the substrate, and a reference electrode on the substrate, wherein the one or more working electrodes, the counter electrode, and the reference electrode are spaced apart from one other,and wherein the one or more working electrodes, the counter electrode, and the reference electrode comprise carbon nanotubes (CNT) and cellulose, a weight ratio of the carbon nanotubes (CNTs) to the cellulose selected from a range from 1 : 8 to 1 : 2.

[0005] Various embodiments may relate to a method of making an electrode device. The method of making an electrode device may include forming one or more working electrodes, a counter electrode, and a reference electrode on a substrate, wherein the one or more working electrodes, the counter electrode, and the reference electrode are spaced from one another, and wherein the one or more working electrodes, the counter electrode, and the reference electrode comprise carbon nanotubes (CNT) and cellulose, a weight ratio of the carbon nanotubes (CNTs) to the cellulose selected from a range from 1 : 8 to 1 : 2.BRIEF DESCRIPTION OF THE DRAWINGS

[0006] In the drawings, like reference characters generally refer to the same parts throughout the different views. The drawings are not necessarily drawn to scale, emphasis instead generally being placed upon illustrating the principles of various embodiments. In the following description, various embodiments of the invention are described with reference to the following drawings.

[0007] FIG. 1 is a general illustration of a method of making an electrode device according to one embodiment of the present disclosure.

[0008] FIG. 2 is a general illustration of a method of making an electrode device according to another embodiment of the present disclosure.

[0009] FIG. 3A illustrates a film to be patterned of FIG. 2.

[0010] FIG. 3B illustrates a patterned film of FIG. 3A.

[0011] FIG. 3C illustrates the patterned film of FIG. 3B through which a mixture is filtered and integrated therewith to form electrodes.

[0012] FIG. 3D illustrates a plan view of the electrodes of FIG. 3C transferred to a substrate.

[0013] FIG. 3E illustrates a perspective view of the electrodes on the substrate of FIG. 3D.

[0014] FIG. 3F illustrates a plan view of the electrodes on the substrate of FIG. 3D over which an insulation layer is deposited.

[0015] FIG. 4 illustrates a perspective view of the electrodes on the substrate deposited with the insulation layer of FIG. 3F, with an Ag / AgCl layer and a stabilization layer coated on anelectrode pad of the preliminary reference electrode to form an electrode device, showing electrode pads while omitting other parts for clarity.

[0016] FIG. 5 illustrates a plan view of the coated electrode device of FIG. 4.

[0017] FIG. 6 is a general illustration of a method of forming a signal collection platform according to one embodiment of the present disclosure.

[0018] FIG. 7 is a general illustration of a signal collection platform of the method of FIG. 6.

[0019] FIG. 8 illustrates EC-omics (electrochcmomics) profiling the holistic metabolic dynamics in peripheral biofluids in one example. Insets (a,b) illustrate that an EC-omics approach uncovers metabolic networks from an electrochemical perspective through the electrode-biofluids interfaces (see Inset (b)) that remain unexplored in the traditional moleculecentric frameworks (see Inset (a)). The term “CNT / BC” refers to a carbon nanotube / bacteria cellulose composite as proposed herein. Inset (c) illustrates that EC-omics can provide a holistic view of the metabolic network in peripheral biofluids from an electrochemical perspective with our portable EC-omics platform. Uric acid was used as an illustrative or representative electroactive metabolite, that can be analyzed with voltammetric spectra methods (CV and DPV). With feature extract processes, the electrochemical features could be integrated into comparatively comprehensive EC-omics networks. Inset (d) illustrates examples of main constituents of EC-omics, including, e.g., electroactive species and their associated electrochemical properties. The impact of solution variables (I-III) and kinetic characteristics (IV-VI) on the electrochemical spectra were illustrated according to electrochemical theories.

[0020] FIG. 9A illustrates a method of fabricating saliva profiling electrodes involving vacuum filtration and screen-printing methods in one example. Nylon films (pore size 0.45 μm) were patterned with 30 pm thick polyimide (PI) tape and different ratios of carbon nanotube / bacterial cellulose (CNT / BC) were filtered onto the patterned Nylon films. The resulting CNT / BC electrodes were transferred onto a conformable PI substrate (adhesive force: 200 400 g) and screen printed with Ag / AgCl paste (reference electrode) and a PI insulation layer. The Ag / AgCl was covered with a NaCl / PVB film by drop casting to complete the disposable electrode.

[0021] FIG. 9B arc SEM images showing CNT / BC (1:3 mass ratio) composite electrodes (bottom) are denser than electrodes made with CNTs alone (top) in one example. Both sampleswere sputter-coated with a 10 nm layer of gold to improve conductivity and produce clearer SEM images.

[0022] FIG. 9C illustrates the contact angle (bars, left axis) and sheet resistance (symbols, right axis) of CNT / BC composite films at different mass ratios (wt%) in one example. Inset of FIG. 9C shows images from contact angle tests. Electrodes with higher BC ratios are more hydrophilic and have lower sheet resistance.

[0023] FIG. 10 illustrates performances of CNT / BC electrodes in comparison with commercially available electrochemical electrodes, c.g., electrochemical electrodes fabricated in this work (Inset (a), CNT / BC electrodes) and three representative electrodes obtained by other methods (Insets (b-d)). Inset (b) shows Au electrode fabricated with thermal evaporation method (Gold-Thermal electrodes). Inset (c) shows gold electrodes composed fabricated by screen printing method (Gold paste-SP electrodes). Inset (d) shows carbon paste electrode fabricated with screen printing method (Carbon paste-SP electrodes). Insets (e, f) show CV spectra of different electrodes for PBS solution (Inset (e)) and a mixture solution of uric acid (160 pM) and tyrosine (200 pM) in PBS solution (Inset (f)).

[0024] FIG. 11 A illustrates the disposable electrodes for biofluids analysis containing four working electrodes (WE 1 to WE4) placed symmetrically relative to the Ag / AgCl reference electrode (RE) and the counter electrode (CE), according to one example.

[0025] FIG. 1 IB illustrates the COMSOL analysis of an electrochemical reaction consisting of a reactant (R) and product (P), showing three CV scans consistent among all four working electrodes. Two segments per CV scan were performed: -0.8 to 1.0 V and 1.0 to -0.8 V. Each scan started at -0.8 V and ended at 1.0 V. Redox potential (EP) was set at 0.4 V with a scan rate of 100 mV / s and a voltage range of -0.8 to 1 V. The inset shows a map of simulated concentrations of R (top layer) and P (bottom layer).

[0026] FIG. 1 1C illustrates CV spectra of artificial saliva containing 160 pM uric acid and 200 pM tyrosine obtained using 8 different disposable electrodes (si to s8) showing that measurements from all four working electrodes were consistent and reproducible. Peaks for uric acid and tyrosine could be determined.

[0027] FIG. 1 ID illustrates that in one example, typical EC-omics features (Ep, ip, W1 / 2) for uric acid (0-0.4 V) and tyrosine (0.4-0.8 V) extracted from CV and DPV spectra were greater than 95% and consistent across all eight different sensors. Each symbol is an average of fourworking electrodes. A standard deviation (error bars) of < 7% indicates measurements from the four working electrodes are consistent.

[0028] FIG. 12A illustrates an example of a system-level block diagram showing the signal transduction, processing, and wireless transmission from sensors to the user interface. A 420 mAh lithium polymer battery and voltage regulator powered the system with ± 3.3 V. The preprogrammed microcontroller generates and applied the desired voltage to the electrodes through an external digital-to-analog converter (DAC) smoothened by a low-pass filter (LPF). The resulting current was amplified and converted to voltage by a trans-impedance amplifier (TIA) and read by an analog-to-digital converter (ADC). The final digital voltammetry signal was transmitted to the user interface via Bluetooth.

[0029] FIG. 12B is an image of a portable signal collection circuit connected to the disposable sensing electrodes by connection cables, according to one example.

[0030] FIG. 12C illustrates, according to one example, an application programming interface of an EC-omics software that contains sections for inputting basic clinical information (Profile), adjusting measurement parameters for collecting voltammetric spectra (Collection), and extracting features from voltammetric spectra and producing support vector machine (SVM)-bascd diagnosis (Analysis).

[0031] FIG. 13A is a schematic diagram of the equilibrium of uric acid at the CNT / BC electrode-biofluid interfaces and the impact of solution parameters on the voltammetric spectra, according to one example.

[0032] FIG. 13B illustrates two cyclic voltammetry (CV-1 and CV-2) scans and one differential pulse voltammetry (DPV) obtained by applying programmed stimulation voltage waveforms to the working electrodes in one example. Characteristic peaks in the representative CV-1, CV-2, and DPV spectra of uric acid and tyrosine were obtained by averaging spectra from the four working electrodes.

[0033] FIG. 13C illustrates peaks for uric acid varied according to the concentration of uric acid (Curie and: 40-160 pM), relative electrolytes (Cartificiai saliva: 25-100%, with the full composition of artificial saliva set as 100%), and pH value (5-7) of artificial saliva in one example. Planes of 3D cubes show spectra from a typical artificial saliva solution at constant pH 7, 100% Cartificiai saliva, and 160 pM Curie acid- Working potentials were referenced to Ag / AgCl.

[0034] FIG. 13D shows typical features extracted from CV-1, CV-2, and DPV spectra, including redox potential (Ep), redox current (ip), and half-peak width (W1 / 2), according to one example. Regression lines and markers delineate the data zone of a peak.

[0035] FIG. 13E illustrates contour plots showing ipof uric acid extracted from CV-1 spectra responding differently to Curie acid (40- 160 pM), Cartificiai saliva (25- 100%) and pH (5-7) in artificial saliva in one example. The ipwas averaged from WE 1 to WE4 of one portable sensor. All data are representative of three independent experiments.

[0036] FIG. 13F illustrates contour plots showing Epof uric acid extracted from CV-1 spectra responding differently to Curie and (40-160 pM), Cartificial saliva (25-100%) and pH (5-7) in artificial saliva, according to one example. Epwas average from WE 1 to WE4 of one portable sensor. All data are representative of three independent experiments.

[0037] FIG. 13G illustrates contour plots showing W1 / 2 of uric acid extracted from CV-1 spectra responding differently to Curie acid (40-160 pM), Cartificial saliva (25-100%) and pH (5-7) in artificial saliva in one example. The W1 / 2 was averaged from WE 1 to WE4 of one portable sensor. All data arc representative of three independent experiments.

[0038] FIG. 14 illustrates a comparison of the proposed portable electrochemical sensing platform with a conventional electrochemical station. CV spectra of mixtures of uric acid (160 pM) and tyrosine (200 pM) in l x PBS solution was collected from the proposed E-omics platform and the conventional electrochemical station (CHI832C).

[0039] FIG. 15 illustrates feature extraction based on the voltammetric spectra of uric acid. Insets (a,b) illustrate contour plots showing ipfeatures extracted from CV-2 (Inset (a)) and DPV (Inset (b)) spectra of uric acid increasing with increasing concentration of uric acid (Curie acid: 40-160 pM), and electrolyte composition of the artificial saliva ( Cartificial saliva: 25-100%) and decreasing pH (5-9). Insets (c,d) illustrate contour plots showing Epfeatures extracted from CV-2 (c) and DPV (d) spectra of uric acid to be constant across all concentrations of uric acid and electrolyte composition of the artificial saliva but decreases with increasing pH values. Insets (e,f) illustrate that contour plots of W1 / 2 features extracted from CV-2 (e) and DPV (Inset (f)) spectra of uric acid display distinct response patterns towards changes in uric acid concentration, electrolyte composition of the artificial saliva, and pH gradients. All data arc representative of three independent experiments.

[0040] FIG. 16 illustrates iPfeatures of uric acid in response to different pH in the artificial saliva in one example. Insets (a,b,c) show fitting curves of ipfeatures of CV-1 (Inset (a)), CV-62 (Inset (b)), and DPV (Inset (c)) spectra of uric acid at different pH values. Inset (d) is a graph showing the sensitivity of electrodes based on CV-1, CV-2, and DPV spectra decreased at higher pH values.

[0041] FIG. 17 illustrates typical Epfeatures of uric acid response to pH value and electrolytes concentration of the artificial saliva in one example. Fitting curves of Epfeatures of CV-1 (Inset (a)), CV-2 (Inset (b)), and DPV (Inset (c)) spectra of uric acid at different pH values and electrolyte concentrations of artificial saliva. The box drawn in dashed lines in Inset (c) highlights the deviation tests at pH 9 in CV-2, which is due to low peak intensity.

[0042] FIG. 18A illustrates contour plots showing ipfeatures extracted from CV-1 spectra of tyrosine increase with increasing concentration of tyrosine (Ctyrosine: 50-200 pM) and decreasing pH (5-9) in one example. A less regular trend is observed when changing the electrolyte composition of the artificial saliva. All data are representative of three independent experiments.

[0043] FIG. 18B illustrates contour plots showing ipfeatures extracted from CV-2 spectra of tyrosine increase with increasing concentration of tyrosine (Ctyrosine: 50-200 pM) and decreasing pH (5-9) in one example. A less regular trend is observed when changing the electrolyte composition of the artificial saliva. All data arc representative of three independent experiments.

[0044] FIG. 18C illustrates contour plots showing ipfeatures extracted from DPV spectra of tyrosine increase with increasing concentration of tyrosine (Ctyrosine: 50-200 pM) and decreasing pH (5-9) in one example. A less regular trend is observed when changing the electrolyte composition of the artificial saliva. All data arc representative of three independent experiments.

[0045] FIG. 18D illustrates contour plots showing that Epfeatures extracted from CV-1 spectra of tyrosine were constant across all concentrations of tyrosine and electrolyte composition of the artificial saliva but decreased with increasing pH values. All data are representative of three independent experiments.

[0046] FIG. 18E illustrates contour plots showing Epfeatures extracted from CV-2 spectra of tyrosine were constant across all concentrations of tyrosine and electrolyte composition of the artificial saliva but decreased with increasing pH values in one example. All data are representative of three independent experiments.

[0047] FIG. 18F illustrates that contour plots show Epfeatures extracted from DPV spectra of tyrosine are constant across all concentrations of tyrosine and electrolyte composition of the artificial saliva but decrease with increasing pH values in one example. All data arc representative of three independent experiments.

[0048] FIG. 18G illustrates that, in one example, contour plots of W / / 2 features extracted from C V- 1 spectra of tyrosine displayed distinct response patterns from uric acid to changes in tyrosine concentration, electrolyte composition of the artificial saliva, and pH gradients. All data arc representative of three independent experiments.

[0049] FIG. 18H illustrates that, in one example, contour plots of W1 / 2 features extracted from CV-2 spectra of tyrosine displayed distinct response patterns from uric acid to changes in tyrosine concentration, electrolyte composition of the artificial saliva, and pH gradients. All data are representative of three independent experiments.

[0050] FIG. 181 illustrates that, in one example, contour plots of W'1 / 2 features extracted from DPV spectra of tyrosine displayed distinct response patterns from uric acid to changes in tyrosine concentration, electrolyte composition of the artificial saliva, and pH gradients. All data are representative of three independent experiments.

[0051] FIG. 19 illustrates iPfeatures of tyrosine in response to different pH in the artificial saliva in one example. Insets (a,b,c) show fitting curves of ipfeatures of CV-1 (a), CV-2 (b), and DPV (c) spectra of tyrosine at different pH values. Inset (d) illustrates that graph shows the sensitivity of electrodes based on CV-1, CV-2, and DPV spectra is higher at pH 6, and decreased with increasing pH value from 6 to 9.

[0052] FIG. 20 illustrates EC-omics features of representative clcctroactivc biomolcculcs with shared tyrosine residues in one example. Insets (a, b) show structural formula (Inset (a)) and respective segmental CV and DPV spectra (Inset (b)) of four representative tyrosine residues containing biomolecules (tyrosine, angiotensin IT, oxytocin, and mucin II) at pH 5, 7, and 9, showing varied responses. Potential values ranged from 0.4 to 0.8 V vs. Ag / AgCl. Inset (c) shows heatmap plots of EC-omics features (Ep, ip, and W1 / 2) extracted from the respective CV-1, CV-2, and DPV spectra of tyrosine, angiotensin II, oxytocin, and mucin II, showing distinct electrochemical properties in response to pH. Each feature is an average of eight data points. Inset (d) shows segmented H NMR spectra of various molecules in artificial saliva at different pH values. Insets (c,f) show principal component analysis (PCA) of extracted EC-omics features (Inset (c)) discriminated tyrosine, angiotensin II, oxytocin, and mucin II better than PCA of original voltammetric spectra (Inset (f)).

[0053] FIG. 21 A illustrates a workflow for constructing a saliva EC-omics and NMR database, according to one example. Saliva samples were collected from 75 subjects using standard spitting procedures. Voltammetric and ’H NMR spectra were obtained, subjected to feature extraction, and integrated into the EC-omics and NMR databases, respectively.

[0054] FIG. 21B illustrates representative CV-1, CV-2, and DPV spectra from a healthy subject used to illustrate EC-omics feature extraction in one example. For peak features, the spectra were classified into three sections based on the redox potential of metabolites: pkpr(0.4-0.8 V) for protein-related peaks, pkua(0-0.4 V) for uric acid and nearby compounds, and pkom (-0.5-0 V) for other electroactive metabolites. For area features, the 0 μA line divides the CV spectra into oxidation area (Sp) and reduction area (Sn) to capture the charge transfer process. The calculated ratio of Spto Sn(Sr) indicates the overall redox states of a biofluid. Working potentials are referenced to Ag / AgCl.

[0055] FIG. 21C illustrates extraction of untargeted NMR peak features (Nnmr) and area features (Snim) of a representative1H NMR spectrum of saliva collected from a healthy subject in one example. The spectrum is divided into six sections (nmrl-6) based on chemical shifts of hydrogen atoms connected to different functional groups. Trimethylsilylpropanoic acid (TSP) was shown.

[0056] FIG. 2 ID illustrates statistical distributions of Pearson correlation coefficient (r) of different EC-omics features with periodontitis-related biomarkers. Interclass correlation shows iPand area correlate positively while Epand W1 / 2 exhibit weak correlation with periodontitis- related molecular biomarkers or untargeted NMR features. The median is indicated by the dashed line.

[0057] FIG. 21 E illustrates statistical distributions of Pearson correlation coefficient (r) of different EC-omics features with untargeted NMR features. Interclass correlation shows ipand area correlate positively while Epand W1 / 2 exhibit weak correlation with periodontitis-related molecular biomarkers or untargeted NMR features. The median is indicated by the dashed line.

[0058] FIG. 22 A illustrates chemical structures and corresponding identified NMR peaks of five periodontitis-related biomarkers in saliva used to calculate biomarker abundances from1H NMR spectra. The NMR peaks with minimal overlap with other metabolites arc indicated by regions (indicated by dashed lines) corresponding to hydrogen atoms.

[0059] FIG. 22B illustrates paired Pearson correlation analysis comparing EC-omics features with periodontitis -related molecular biomarkers. The ipand area correlated positively with periodontitis-related molecular biomarkers and untargeted NMR features, while Epand W1 / 2 exhibited a weaker correlation. Positive and negative correlations and significance of correlation are shown by the size and shading of the circles. Asterisks denote t-test p < 0.05.

[0060] FIG. 22C illustrates paired Pearson correlation analysis comparing EC-omics features with untargeted NMR features. The ipand ar ea correlated positively with periodontitis- related molecular biomarkers and untargeted NMR features, while Epand W1 / 2 exhibited a weaker correlation. Positive and negative correlations and significance of correlation are shown by the size and shading of the circles. Asterisks denote t-test p < 0.05.

[0061] FIG. 22D illustrates correlations between 48 features in the saliva EC-omics database (absolute value less than 0.5 not presented) according to one example. The Pearson correlation coefficient greater than 0.5 indicates substantial correlations among the different EC-omics features (ip, Ep, W1 / 2, num, and area), which could unveil the complex metabolic networks from an electrochemistry perspective, providing information way beyond the scope and dimensionality of information available from the conventional molecule-centric biomarkers. Positive and negative correlations and significance of the correlation arc shown by the size and shading of the circles.

[0062] FIG. 23 illustrates paired Pearson correlation analysis between other EC-omics features and untargeted NMR features. Positive and negative correlations and significance of correlation axe shown by the size and shading of the circles. Asterisks denote t-test p < 0.05.

[0063] FIG. 24A illustrates a scheme in which the metabolic network in saliva changes at the onset of periodontitis.

[0064] FIG. 24B are periapical radiographs of three representative healthy subjects and three subjects with periodontitis. Tooth position is indicated at the lower right corner of the image: the symbol p refers to left mandibular teeth, the symbol refers right maxillary teeth, and the symbol refers to right mandibular teeth. The numbers 6 and 7 in FIG. 24B represent the first molars and the second molars, respectively. Healthy subjects do not show significant alveolar bone resorption, e.g., showing a comparatively shorter distance between the cementoenamel junction and the alveolar crest. Periodontitis subjects had alveolar bone resorption exceeding a third of the root length.

[0065] FIG. 24C illustrates visualizations of metabolic dynamics between the healthy and periodontitis groups with 6-axes radar charts, in which each axis represents the kernel smooth distribution of representative EC-omics features (I) or periodontitis -related molecular biomarkers (II). For each Kernel smooth distribution, the distance from the edge to the axis reflects the estimated probability density. The distinct patterns of kernel smooth distribution (shaded area along the axes) and average intensity (connected circles) of representative iP, Ep, W1 / 2. area features, and abundances of typical periodontitis-related biomarkers indicated the variation of the metabolic networks between healthy and periodontitis groups. An example of feature extraction is illustrated in FIG. 21B, and detailed information for EC-omics features (El to E48) is summarized in Table 2.

[0066] FIG. 24D illustrates the discrimination of healthy subjects (n=34) and periodontitis patients (n=41) indicates dynamic changes in peak features (rhombus, ipfeatures highlighted) and area features (circle) during the onset of periodontitis. Solid symbols indicate significant discrimination. *: p < 0.05, **: p < 0.01, ***: p < 0.001.

[0067] FIG. 24E illustrates statistical results of SVM (500 trials) showing that diagnosis based on EC-omics features is more accurate than those based on ipfeatures, molecular biomarkers, and untargeted NMR features. Each rhombus indicates one SVM trial. Boxplot shows 25% to 75% of the data. The error bar represents the standard deviation.

[0068] FIG. 25 illustrates representativeNMR spectra, voltammetric spectra from saliva database. Insets (a-d) show typicalNMR spectra (Insets (a, b)) and the corresponding CV- 1, CV-2, and DPV spectra (Insets (c, d)) from seven representative healthy subjects and seven periodontitis subjects.NMR spectra arc divided into six sections (nmrl-6) based on chemical shifts of hydrogen atoms connected to different functional groups. CV-1, CV-2, and DPV spectra are divided into three primary regions based on the redox potential of representative electroactive species: pkpr(0.4-0.8 V) representing peaks related to proteins and their associated metabolites, pkua(0-0.4 V) representing peaks for uric acid and nearby compounds, and pkom (-0.5-0 V) representing peaks for other electroactive metabolites.

[0069] FIG. 26A illustrates visualizations of the of untargeted NMR features between the healthy groups and the periodontitis groups with 6-axis radar charts, in which each axis represents the kernel smooth distribution of representative untargeted NMR features. For each Kernel smooth distribution, the distance from the edge to the axis reflects the estimated probability density. The distinct patterns of kernel smooth distribution (shaded area along axes)and average intensity (connected circles) of NMR peak and area features validated the variation of the metabolic networks between healthy and periodontitis groups.

[0070] FIG. 26B illustrates statistics of the diagnosis accuracy for 500 trials using Adaptive Boosting (AdaBoost) showing that diagnosis based on EC-omics features is comparable to untargeted NMR features. Comparable accuracy between EC-omics and untargeted NMR features indicates EC-omics is compatible with various machine learning algorithms.

[0071] FIG. 26C illustrates statistics of the diagnosis accuracy for 500 trials using Neural Network (NcurNct) showing that diagnosis based on EC-omics features is comparable to untargeted NMR features. Comparable accuracy between EC-omics and untargeted NMR features indicates EC-omics is compatible with various machine learning algorithms.

[0072] FIG. 26D illustrates representative CV-1 , CV-2, and DPV spectra before and after added with 10% white Gaussian noise (Signal-to-noise ratio, SNR=10 dB).

[0073] FIG. 26E illustrates substantial diagnosis accuracy using different intelligent algorithms under different levels of white Gaussian noise. The error bar represents the standard deviation of 500 trials.

[0074] FIG. 27 illustrates a two-component plot of a typical support vector machine (SVM) test to discriminate healthy and periodontitis patients.DETAILED DESCRIPTION

[0075] The following detailed description refers to the accompanying drawings that show, by way of illustration, specific details and embodiments in which the invention may be practiced. These embodiments arc described in sufficient detail to enable those skilled in the art to practice the invention. Other embodiments may be utilized and structural, logical, and electrical changes may be made without departing from the scope of the invention. The various embodiments are not necessarily mutually exclusive, as some embodiments can be combined with one or more other embodiments to form new embodiments.

[0076] Features that are described in the context of an embodiment may correspondingly be applicable to the same or similar' features in the other embodiments. Features that are described in the context of an embodiment may correspondingly be applicable to the other embodiments, even if not explicitly described in these other embodiments. Furthermore, additions and / or combinations and / or alternatives as described for a feature in the context of an embodiment may correspondingly be applicable to the same or similar feature in the other embodiments.

[0077] In the context of various embodiments, the articles “a”, “an” and “the” as used with regard to a feature or element include a reference to one or more of the features or elements.

[0078] In the context of various embodiments, the term “about” or “approximately” as applied to a numeric value encompasses the exact value and a reasonable variance, e.g. within 10% of the specified value.

[0079] As used herein, the term “and / or” includes any and all combinations of one or more of the associated listed items.

[0080] Embodiments described in the context of one of the electrode dcvicc / platform arc analogously valid for the other electrode device s / platforms. Similarly, embodiments described in the context of a method are analogously valid for an electrode device / platform, and vice versa.

[0081] In one aspect, the present disclosure provides a method 100 of making an electrode device. The electrode device 390 enables the complementary nature of peripheral biofluid sampling (such as sweat and saliva, but not limited thereto) and electrochemical sensing technology to be leveraged to detect a diverse range of biomarkers, c.g., using electrochemical methods in non-invasive diagnostic systems. FIG. 1 is a general illustration of the method 100 of making an electrode device according to one embodiment of the present disclosure. The method 100 includes forming 102 one or more working electrodes, a counter electrode, and a reference electrode on a substrate, wherein the one or more working electrodes, the counter electrode, and the reference electrode are spaced from one another, and wherein the one or more working electrodes, the counter electrode, and the reference electrode comprise carbon nanotubes (CNT s) and cellulose, a weight ratio of the carbon nanotubes (CNT s) to the cellulose selected from a range from 1 : 8 to 1 : 2. According to various embodiments, the weight ratio of the carbon nanotubes (CNTs) to the cellulose may range from 1 : 8 to 1 : 2. Such weight ratio may provide desirable hydrophilicity and electrical conductivity, and mechanical robustness (provided by cellulose) for biofluids enabling the testing results of biofluid samples to be analyzed with high accuracy and repeatability.

[0082] According to another aspect, the present disclosure provides another method 200 of making an electrode device. FIG. 2 is a general illustration of the method 200 of making an electrode device according to another embodiment of the present disclosure. The method 200 includes mixing 202 the carbon nanotubes (CNTs), the cellulose, and a dispersion medium to form a mixture comprising the carbon nanotubes (CNTs) and the cellulose; filtering 204 themixture through a patterned film, thereby forming the one or more working electrodes, the counter electrode, and the reference electrode on the patterned film; transferring 206 the one or more working electrodes, the counter electrode, and the reference electrode from the patterned film to the substrate; depositing 208 a layer comprising silver (Ag) and silver chloride (AgCl) on an electrode pad of the reference electrode; and depositing 210 an insulation layer over portions of the one or more working electrodes, the counter electrode, and the reference electrode, such that the electrode pads of the one or more working electrodes, the electrode pad of the reference electrode and the curved end portion of the reference electrode arc not covered by the insulation layer. Alternatively, the steps of depositing 208 and 210 may be omitted or optional such that the reference electrode may be configured as a pseudo-reference electrode without any layers coated thereon. In such implementation, the pseudo-reference electrode can continue to provide a stable reference potential for profiling the cyclic voltammetric spectra and differential voltammetric spectra.

[0083] Preferably, a weight ratio of the carbon nanotubes (CNTs) to the cellulose in the mixture is selected from a range from 1 : 8 to 1 : 2. The step of mixing 202 can be performed by a magnetic mixer. In step 202, the carbon nanotubes (CNTs), the cellulose, and a dispersion medium is mixed to form a mixture comprising the carbon nanotubes (CNTs) and the cellulose. The step of filtering 204 may be performed by vacuum filtration. The patterned film is preferably hydrophobic. The step of transferring 206 may be performed such that the patterned film is separated from the electrodes. In other words, the patterned film is not transferred together with the electrodes. The step of depositing 208 may be performed by drop casting a paste of silver (Ag) and silver chloride (AgCl), and drying the paste to form a solid layer on the electrode pad of the reference electrode. The step of depositing 210 may be performed by various methods, such as adhering or sticking via an insulation sheet, screen printing, hot press thermoplastic films, chemical vapor deposition, physical vapor deposition, etc.

[0084] Preferably, the patterned film comprises a nylon film and one or more polyimide tapes on portions of the nylon film. Preferably, the deposition of the layer comprising silver (Ag) and silver chloride (AgCl) on the electrode pad of the reference electrode is canned out using screen printing. Preferably, the deposition of the insulation layer over the portions of the one or more working electrodes, the counter electrode, and the reference electrode is earned out using screen printing. Preferably, the substrate is a polyimidc (PI) substrate.

[0085] In some embodiments, the method 200 further comprises depositing a stabilizing layer over the electrode pad of the reference electrode. The deposition of the stabilizing layer over the electrode pad of the reference electrode may preferably be carried out using drop casting. In some embodiments, the stabilizing layer may include chloride salts (e.g., sodium chloride (NaCl) or potassium chloride (KC1)) and hydrophobic or water-repellent polymers.

[0086] FIG. 3 A illustrates a film 310 to be patterned in the method 200 according to various embodiments. The film 310 may be a nylon film and one or more poly imide tapes on portions of the nylon film. Alternatively, the nylon film may be substituted by one of the group consisting of a PVDF (polyvinylidene difluoride) film, a PTFE (Polytetrafluoroethylene) film, a PES (polyether sulfone) films, a MCE (Mixed cellulose ester) film, and other porous films (such as Polyurethane) fabricated by electrospinning. The nylon film or any substitutes of the nylon film is preferably configured to be porous, the sizes of their pores ranging from 0.1 pm to 20 pm. Alternatively, the polyimide tapes may be substituted by one of the group consisting of PET (Polyethylene terephthalate) tapes, Polypropylene (PP) tapes, and Polyvinyl chloride (PVC) tapes.

[0087] As is shown in FIG. 3B, the film 10 is patterned to form multiple patterned voids 340. The patterned voids 340 include four units 350. Each unit 350 includes a reference void 352, four working voids 354, and a counter void 356. In some embodiments, the four working voids 354 may be disposed on both lateral sides of the reference void 352. In some embodiments, the four working voids 354 may be divided equally and disposed on both sides of the reference void 352. It would be understood that various numbers of the patterned voids 340 may be provided, e.g., not limited to four units 350. In some embodiments, the patterned voids 340 may include fewer than or more than four units 350. Further, each unit 350 may include fewer than four working voids 354. Alternatively, each unit 350 may include more than four working voids 354. Each unit 350 is spaced from one another. Each reference void 352, each working void 354, and each counter void 356 are spaced from one another.

[0088] As is shown in FIG. 3C, the patterned film 310 is used to filter the mixture 320. After filtering 204, the mixture 320 fills and remains in the patterned voids 340, and is integrated with the patterned film 310 to form the integrated film 310. Both the mixture 320 and the filtering 204 may be configured such that the mixture 320 remaining in the patterned voids 340 is substantially flushed with the main surfaces 314 and 316 of the film 310. In other words, the mixture 320 forms a generally planar or flat surface with the main surfaces 314 and 316. Themixture 320 which remains in the patterned voids 340 forms the reference electrodes 322, the working electrodes 324, and the counter electrodes 326.

[0089] As is shown in FIG. 3D and FIG. 3E, a set of the reference electrodes 322, working electrodes 324, and counter electrodes 326 are then transferred 206 to a substrate 330, forming an electrode device 390. The reference electrode 322 includes an electrode trace 3224 connected to an electrode pad 3222 at an end portion. Each working electrode 324 also includes an electrode trace 3244 connected to an electrode pad 3242 at an end portion. The counter electrode 326 similarly includes an electrode trace 3264 connected to an electrode pad 3262 at an end portion. The electrode traces 3224, 3244, 3264 are configured to electrically connect to a signal collection circuit arrangement such that the signal collection circuit arrangement in electrical connection with the set of the reference electrode 322, the working electrodes 324, and the counter electrode 326. In some examples, the thickness of the substrate 330 may range from 10 pm to 10000 pm and the thicknesses of the electrodes 322, 324, 326 may range from 0.1 pm to 10 pm. In some examples, the diameters of the electrode pads 3242 range from 0.2 mm to 20 mm.

[0090] An insulation layer 370 is then deposited 210 on the electrode traces 3224, 3244, 3264 and selected portions of the substrate 330, as is shown in FIG. 3F. The insulation layer 370 may cover the electrode traces 3224, 3244, 3264 as well as selected portions of the substrate 330 and exposes the electrode pads 3222, 3242, 3262. This enables a sample awaiting testing to be positioned or disposed on the insulation layer 370 during test, such that the sample is only in contact with the electrode pads 3222, 3242, 3262 and is insulated from the electrode traces 3224, 3244, 3264. In other words, when the sample is disposed on the insulation layer 370, the sample electrically couples the electrode pads 3222, 3242, 3262. In some examples, the sample may be a biofluid. The sample may be selected from the group consisting of noninvasive saliva, sweat, wound extrudate, tears, urine, and less invasive interstitial fluid, to invasive blood, and any combination thereof.

[0091] A layer 362 including silver (Ag) and silver chloride (AgCl) may be deposited 208 on the electrode pad 3222 before or after the step of depositing the insulation layer 370. The layer 362 may alternatively contain other alternative material such as Iridium Oxide (IrOx), carbon-based electrodes (or other conducting paste) modified with / without Ag / AgCl, Platinum, metal oxides (such as ruthenium oxide, or manganese dioxide). In some examples, a pseudoreference electrode may be provided in place of using Ag / AgCl in said electrode. Astabilization layer 364 may then be deposited on the layer 362 to protect the layer 362. The stabilization layer 364 may include a composition that can be selected from the group consisting of NaCl / PVB (polyvinyl butyral), NaCl / Nafion, NaCl / Poly urethane and any combination thereof. The thickness of the layer 362 may range from 1 pm to 50 pm and the thickness of the stabilization layer 364 may range from 50 nm to 50 pm. FIG. 4 illustrates a perspective view of the electrode pads 3222 / 3242 / 3262 with the layer 362 and the stabilization layer 364 coated on the electrode pad 3222, in which connecting parts to the electrode pads 3222 / 3242 / 3262 arc omitted for clarity.

[0092] FIG. 5 illustrates a plan view of the electrode pads and the layers 362 and 364 of FIG. 4. A lateral axis 3222a and a longitudinal axis 3222b are defined by the electrode pad 3222 of the reference electrode 322. The lateral axis 3222a and the longitudinal axis 3222b extend along the substrate 330 and are perpendicular to each other. The electrode pads 3222, 3242, 3262 are all on a same side of the lateral axis 3222a. The electrode pads 3242 of the working electrodes 324 are equidistantly spaced by di from the electrode pad 3222 of the reference electrode 322 on the substrate 330. This improves the consistency of the testing results of the working electrodes 324 with respect to the reference electrode 322.

[0093] Each of the electrode pads 3242 of the working electrodes 324 is equidistantly spaced by d2 from a neighboring electrode pad 3242 of another working electrode 324. In other words, a first distance between a first neighboring pair of the electrode pads 3242 of the working electrodes 324 is equal to a second distance between a second neighboring pair of the electrode pads 3242 of the working electrodes 324. The equidistant four-working electrode configuration improves the accuracy of the testing results of the working electrodes 324.

[0094] The electrode pad 3242 of each of the working electrodes 324 is spaced from the electrode pad 3262 of the counter electrode 326 with an identical minimum distance ds. This configuration enables each electrode pads 3242 of the working electrodes 324 to form an identical electrical circuit with the electrode pad 3262 of the counter electrode 326 and further improves the consistency of the testing results of the working electrodes 324.

[0095] The layout of the electrode pads, as a whole, ensures repeatability and accuracy of testing results of the electrodes. “Repeatability” herein refers to consistency of testing results of a single sample in multiple measurements. “Accuracy” herein refers to correctness or precision of testing results in a single measurement.

[0096] The electrode pads 3222, 3242, 3262 arc symmetrically disposed about the longitudinal axis 3222b. It should be understood that d1, d2, d3ma nyyot necessarily be identical to one another. Alternatively, d1, d2, d3may also be identical to one another. In some embodiments, selected ones of d1, d2, d3may be identical to each other, for example, di may be identical to ds. In some examples, d1ranges from 0.2 mm to 20 mm. ds ranges from 0.2 mm to 20 mm. ds ranges from 0.2 to 20 mm.

[0097] According to another aspect of the disclosure, the present disclosure provides an electrode device. The electrode device includes a substrate, one or more working electrodes on the substrate, a counter electrode on the substrate, and a reference electrode on the substrate, wherein the one or more working electrodes, the counter electrode, and the reference electrode are spaced apart from one other, and wherein the one or more working electrodes, the counter electrode, and the reference electrode comprise carbon nanotubes (CNT) and cellulose, a weight ratio of the carbon nanotubes (CNTs) to the cellulose selected from a range from 1 : 8 to 1 : 2.

[0098] In some embodiments, an end portion of each of the one or more working electrode comprises an electrode pad. Similarly, an end portion of the reference electrode comprises an electrode pad. In some embodiments, an end portion of the counter electrode is curved.

[0099] In some embodiments, the one or more working electrodes are more than one electrode (in other words, there are multiple working electrodes), with electrode pads of the one or more working electrodes equidistantly spaced from the electrode pad of the reference electrode on the substrate. The electrode pad of each of the one or more working electrodes is spaced from the curved end portion of the counter electrode with an identical minimum distance. A first distance between a first neighboring pair of electrode pads of the one or more working electrodes may be equal to a second distance between a second neighboring pair of electrode pads of the one or more working electrodes.

[0100] In some embodiments, a total number of the one or more working electrodes is even.

[0101] Preferably, a number of the one or more working electrodes on a first side of the reference electrode is equal to a number of the one or more working electrodes on a second side of the reference electrode opposite the first side. For avoidance of doubt, the arrangement of working electrodes modified with different sensing materials relative to the reference electrode may not be limited to this preferred configuration.

[0102] The electrode pads of the one or more working electrodes may be configured with a same shape.

[0103] In some embodiments, the electrode device may further include an insulation layer over portions of the one or more working electrodes, the counter electrode, and the reference electrode. The electrode pads of the one or more working electrodes; the electrode pad of the reference electrode and the curved end portion of the reference electrode arc not covered by the insulation layer.

[0104] Preferably, the insulation layer comprises polyimide (PI), polyethylene terephthalate (PET), polyurethane (PU) or screen printed epoxy resins.

[0105] Preferably, a total area of the electrode pads of the one or more working electrodes is smaller than the curved end portion of the counter electrode.

[0106] In some embodiments, the electrode device may further include a layer comprising silver (Ag) and silver chloride (AgCl) on the electrode pad of the reference electrode.

[0107] Preferably, the electrode device further includes a stabilization layer over the electrode pad of the reference electrode.

[0108] Preferably, the stabilization layer comprises: a chloride salt; and a hydrophobic (or water-repellent) polymer.

[0109] Preferably, the electrode device further includes a reservoir around the electrode pads of the one or more working electrodes, the curved end portion of the counter electrode, and the electrode pad of the reference electrode.

[0110] Preferably, the cellulose is selected from the group consisting of bacterial cellulose, cotton cellulose, wood cellulose, and any combination thereof.

[0111] In some embodiments, the electrode device further includes one or more sensing moieties on the one or more working electrodes, the one or more sensing moieties being selected from the group consisting of polyaniline (PANi), Prussian blue (PB), a metal-organic framework (MOF), copper nanoparticlcs, an electrochemical catalyst, and any combination thereof.

[0112] The sensing moieties may be polyanilinc (PANi) with a weight ratio of the carbon nanotubes (CNTs) to PANi selected from a range from 1 :2 to 4: 1 .

[0113] Preferably, the sensing moieties are Prussian Blue (PB) with a weight ratio of the carbon nanotubes (CNTs) to Prussian Blue (PB) selected from a range from 1: 1 to 4:1.

[0114] In yet another aspect, the present disclosure provides a method 600 of forming a signal collection platform. As is shown in FIG. 6, the method 600 includes electrically connecting 602 a signal collection circuit arrangement to the electrode device as described in earlier section. In some embodiments, a controller is configured to receive incoming wireless signals from a processing device and further configured to control the electrode device to apply one or more stimulation voltages to a sample based on the incoming wireless signals received, and wherein the controller is configured to obtain electrical data of the sample via the electrodes based on the one or more stimulation voltages applied and configured to transmit one or more outgoing wireless signals to the processing device based on the electrical data obtained.

[0115] FIG. 7 is a general illustration of a signal collection platform 700 of the method of FIG. 6. An electrode device 710 as described above, e.g. the electrode device 390 having the layers 362, 364, 370, is electrically connected 602 with a signal collection circuit arrangement 720, forming the signal collection platform 700. The signal collection circuit arrangement 720 includes a controller 722. The controller 722 is configured to receive incoming wireless signals from a processing device and further configured to control the electrode device to apply one or more stimulation voltage waveforms to a sample based on the incoming wireless signals received. The controller 722 is further configured to obtain electrical data of the sample via the electrodes based on the one or more stimulation voltages waveforms applied and configured to transmit one or more outgoing wireless signals to the processing device based on the electrical data obtained.

[0116] The controller 722 may include an analog-to digital converter (ADC). The signal collection circuit arrangement 720 may further include a trans -impedance amplifier (TIA) in electrical connection with the analog-to digital converter (ADC).

[0117] Alternatively, the controller 722 may include a digital-to-analog converter (DAC). The signal collection circuit arrangement 720 may further include a low pass filter (LPF) in electrical connection with the digital-to-analog converter (DAC).

[0118] The signal collection circuit arrangement 720 may further include a voltage regulator configured to provide a regulated voltage to the controller based on a supply voltage provided by a battery.

[0119] The signal collection circuit arrangement 720 may further include a processing device that can be a mobile device. A sample that can be tested by the signal collection circuit arrangement 720 can be a biofluid comprising one or more electroactive species selected fromthe group consisting of tyrosine, angiotensin II, oxytocin or mucin II. The processing device is configured to extract one or more features from the electrical data. The features comprise redox potential (Ep), redox current (zp), half-peak width (W1 / 2), or area features (S). The processing device can be further configured to compare the extracted one or more features with data from a database for making a disease diagnosis. The data from the database is based on different electroactive species under different measurement conditions. The processing device can be further configured to modify the database based on the electrical data of the sample.

[0120] The one or more stimulation voltage waveforms may include a cyclic voltammetry (CV) voltage waveform having a first portion which increases linearly and a second portion which decreases linearly. The one or more stimulation voltage waveforms may include a differential pulse voltammetry (DPV) voltage waveform having a plurality of portions with constant values over predetermined durations of time.

[0121] In yet another aspect, the present disclosure provides a signal collection platform. The signal collection platform includes the electrode device described above and a signal collection circuit arrangement comprising a controller, the signal collection circuit arrangement in electrical connection with the electrode device. The controller is configured to receive incoming wireless signals from a processing device and further configured to control the electrode device to apply one or more stimulation voltage waveforms to a sample based on the incoming wireless signals received, and wherein the controller is configured to obtain electrical data of the sample via the electrodes based on the one or more stimulation voltages waveforms applied and configured to transmit one or more outgoing wireless signals to the processing device based on the electrical data obtained. The signal collection platform may be an electrochemomics platform. In other words, the signal collection platform may make use of electrochemomics. Electrochemomics is an analytical approach that combines electrochemical techniques with data analysis to provide comprehensive profiling of complex biological samples.

[0122] In some embodiments, the one or more stimulation voltage waveforms comprise a cyclic voltammetry (CV) voltage waveform having a first portion which increases linearly and a second portion which decreases linearly.

[0123] In some embodiments, the one or more stimulation voltage waveforms comprise a differential pulse voltammetry (DPV) voltage waveform having a plurality of portions with constant values over predetermined durations of time.

[0124] Preferably, the controller comprises an analog-to digital converter (ADC). The signal collection circuit arrangement comprises a trans -impedance amplifier (TIA) in electrical connection with the analog-to digital converter (ADC).

[0125] Preferably, the controller comprises a digital-to-analog converter (DAC). The signal collection circuit arrangement comprises a low pass filter (LPF) in electrical connection with the digital-to-analog converter (DAC).

[0126] The signal collection platform may further include a voltage regulator configured to provide a regulated voltage to the controller based on a supply voltage provided by a battery.

[0127] Preferably, the signal collection platform further includes a processing device.

[0128] Preferably, the processing device is a mobile device.

[0129] Preferably, the sample comprises a biofluid comprising one or more electroactive species.

[0130] Preferably, the one or more electroactive species comprise tyrosine, angiotensin II, oxytocin or mucin II.

[0131] Preferably, the processing device is configured to extract one or more features from the electrical data.

[0132] Preferably, the features comprise redox potential (Ep), redox current (ip), half-peak width (Wj / 2), or area features (S) at different partitions according to redox properties of biomolecules.

[0133] Preferably, the processing device is configured to compare the extracted one or more features with data from a database for making a disease diagnosis. The data from the database is based on different clcctroactivc species under different measurement conditions.

[0134] Preferably, the processing device is configured to construct the database based on the electrical data of the sample.

[0135] Example

[0136] To overcome the challenges in decentralized diagnostic settings, it is advantageous to capture a comprehensive view of the metabolic systems and their dynamics during disease onset, rather than relying solely on a limited subset of disease biomarkers. Complementing this endeavor, omics technologies facilitate the untargeted and unbiased discovery of metabolic signatures associated with diseases. This enables the exploration of metabolic dynamics without being confined to a predefined set of targets, and could significantly augment the diagnostic value of peripheral biofluids (see Inset (a) of FIG. 8). The holistic profiling strategycan surpass molcculc-ccntric frameworks, inspiring a broader view of discasc-rciatcd signatures that are adoptable in decentralized health devices. Notably, in the realm electrochemical sensing, the focus may be expanded beyond predefined biomarkers to include the physicochemical properties and kinetic behaviors of electroactive metabolites at electrode interfaces. This strategy facilitates the identification of novel, untargeted biomarkers, thereby enabling a comprehensive analysis of metabolic networks within peripheral biofluids (see Inset (b) of FIG. 8).

[0137] The present disclosure provides an clcctrochcmomics (EC-omics) approach that leverages the untapped and abundant electrochemical parameters to capture the holistic metabolic dynamics in peripheral biofluids (see Inset (c) of FIG. 8). Alternative to conventional omics technologies that primarily focus on molecular networks, the EC-omics approach of the present disclosure enables a more comprehensive characterization of the electroactive species in a sample, and enables extraction of related electrochemical variables in the peripheral biofluids, which enable construction of an electroactivity-based metabolic network (see Inset (d) of FIG. 8). The EC-omics method of the present disclosure can be realized with a customized portable profiling platform, e.g., using the proposed carbon nanotube / bacterial cellulose (CNT / BC) electrodes interfaced with peripheral biofluids to collect comprehensive voltammetric spectra. EC-omics features such as current (ip), redox potential (Ep), half-peak width (W1 / 2), and area features (S) from the spectra (see Table 1) can be extracted.Table 1. Typical EC-omics features and their relation to electroactive speci3es, or variables at solution and electrode interfaceEC-omics Relation to variables at solution and Relation to electroactive species features electrode interface and their behaviorsNumber of transferred electrons (n) Mechanistic description ofElectrode surface area (A) reactions involving the transfer ofPeak current Electrolyte concentration (c) electrons(ip) Diffusion coefficient (n) Concentration of the reactantsScanning rate (v) Diffusion coefficient of theTemperature (T) reactantsThe redox potential of electrochemicalChemical potential related to the reactionPeak location (Ep) types and concentrations ofNumber of transferred electrons (n) electroactive speciesTemperature (T)Mechanistic description of reactions involving the transfer of electronsMechanistic description ofNumber of transferred electrons (n)Peak width (W1 / 2) reactions involving the transfer ofTemperature (T) electronsThe amount of charge for thePositive area of CV Amount of oxidized species oxidization reaction including the(Sp) involved in a CV cycle surface charging processThe amount of charge for theNegative area of Amount of reduced species reduction reaction including theCV (S,t) involved in a CV cycle surface discharging processThe ratio of The ratio of charge for the oxidizationThe ratio of the numbers of positive to the reaction to that for the reduction oxidized to reduced species negative area (Sr) reaction

[0138] The extracted features may be used to form an EC-omics network that reflects the dynamic changes in the metabolic system. A saliva EC-omics database, which may be validated against and compared to periodontitis-related molecular biomarkers and untargeted nuclear magnetic resonance (NMR) features, may be constructed as an illustrative case to demonstrate the accurate disease diagnosis capabilities of the EC-omics approaches of the present disclosure.

[0139] The electrochemical profiling platform of the present disclosure includes disposable sensing electrodes, a portable signal collection circuit, and a mobile interface for wireless data acquisition, transmission and processing of voltammetric signals (see Inset (c) of FIG. 8). The components of the portable electrochemical profiling platform may include the CNT / BC electrochemical electrodes, fabricated using scalable filtration and screen-printing methods, which are adaptable to various conformable substrates suited to practical scenarios (see FIG. 9A). According to various embodiments, the four working electrodes and the counter electrode are made of single-walled carbon nanotubes (CNTs), which serve as the conductive and catalytic material, and bacterial cellulose (BC), which acts as a binder to enhance the mechanical properties and hydrophilicity of the electrodes. Unexpectedly, conventional gold electrodes produced by thermal evaporation or carbon / gold electrodes fabricated throughscreen printing (see Inset (a-f) of FIG. 10) were not the best performing electrodes for the purpose of the electrochemical profiling platform proposed in the present disclosure. Unexpectedly, the proposed CNT / BC electrode was found to better enable the proposed electrochemical profiling platform. This could be attributed to the combination of high sensitivity and negligible background interference characterizing the CNT / BC electrode. In the following experiments, one example used electrodes with a CNT: BC ratio of 1:3 (w / w), which displayed the best electrochemical sensing performance and mechanical stability (see FIG. 9B and FIG. 9C). To obtain parallel voltammetric spectra, one example positioned the four working electrodes symmetrically to the Ag / AgCl reference electrode (see FIG. 11A). Finite element modelling showed that all four working electrodes collected consistent and reproducible voltammetric spectra (see FIG. 1 1B), which were then experimentally validated by the extracted electrochemical features among eight different electrodes from voltammetric spectra of uric acid in phosphate-buffered saline (PBS) solution and (see FIG. 11C and FIG. 11D).

[0140] In addition, an external digital-to-analog converter was used to apply a tailored voltammogram on the electrodes through a microcontroller that generated the programmed stimulation waveform (sec FIG. 12A and FIG. 12B). The transceiver in the signal collection unit allowed voltammetric data to be wirelessly transmitted to a Bluetooth-enabled mobile phone through a customized application programming interface and shared with cloud servers (see FIG. 12C). Considering the complex and irreversible nature of multicomponent peripheral biofluids, two cyclic voltammetry (CV) scans, CV-1 and CV-2, consecutively with an identical voltage range from -0.8 V to 1.0 V and back to -0.8 V were performed, to gather comprehensive electrochemical data. Additionally, to detect the characteristic peaks of the biomolecules, differential pulse voltammetry (DPV) scans (from -0.8 to 1.0 V, and back to -0.8 V) were conducted to minimize background (charging) currents in the CV scans. It was verified that the spectra from these scans could distinguish electroactive metabolites based on their redox activity (such as uric acid and tyrosine) while simultaneously providing a holistic profile of related solution parameters in the biofluids (see FIG. 13A and FIG. 13B). Moreover, the electrochemical profiling platform could be implemented as a portable system configured to collect, process, and analyze the electrochemical signals in a single platform. Advantageously, the collected voltammetric spectra arc comparable in quality to those obtained from conventional electrochemical workstations of significantly larger footprint (see FIG. 14).

[0141] Extracting EC-omics features from electroactive metabolites

[0142] Uric acid, a typical biomarker for purine-rich foods in the saliva was used to demonstrate the extraction of EC-omics features from the voltammetric spectra. CV and DPV spectra of artificial saliva containing varied concentrations of electrolytes (25-100%, with a full composition set as 100%), uric acid (40-160 pM), and pH value (5-9) were collected. The characteristic peaks of uric acid at 0-0.4 V and their changes under different pH, electrolyte, and uric acid concentrations were visualized in a three-dimensional cubic format (see FIG. 13C). To extract essential EC-omics features such as iP, Ep, and W1 / 2 from the characteristic peaks, a bi-quadratic polynomial baseline subtraction to the CV and DPV spectra (see FIG. 13D) was applied. Contour maps for each EC-omics feature were then plotted and organized into a cubic format to reveal the diverse responses of these features to the different solution parameters.

[0143] The contour maps showed different EC-omics features of uric acid displayed complementary response patterns (see FIG. 13E, FIG. 13F, and FIG. 13G). The peak current ipextracted from the different voltammetric spectra increased linearly with increasing uric acid concentration while its sensitivity decreased with increasing pH (see FIG. 13E and Inset (a-b) of FIG. 15). The varying sensitivities among the different voltammetric scans indicated the complementary electrochemical information contained: CV-1 ranged from 0.95 to 0.25 μA μm cm2. CV-2 ranged from 0.59 to 0.16 μA μm1cm-2, and DPV ranged from 0.66 to 0.08 μA μm-1cm-2(see FIG. 16). These results were calculated using the Randles-Sevcik equation, which was also applied to profiling the electrochemical properties of other metabolites in biofluids (See “Electrochemical theories for feature extraction” below).

[0144] Epfeatures of CV-1, CV-2, and DPV spectra changed little across different concentrations of uric acid and electrolytes but decreased linearly when pH increased from 5 to 9 (see FIG. 13F, Inset (c-d) of FIG. 15, and FIG. 17). With W1 / 2, a distinct response pattern towards changes in uric acid concentration, electrolyte composition of the artificial saliva, and pH value (see FIG. 13G and Inset (e-f) of FIG. 15) was confirmed. W1 / 2 provides additional information such as the number of electrons transferred in the electrochemical reaction that complements ipand Ep. All three features (zP, Ep, and W1 / 2 extracted from the CV-1, CV-2, and DPV spectra displayed similar variation tendencies under various artificial saliva conditions. Further, it was demonstrated that the complementary response patterns of these features todifferent solution parameters could provide a holistic profile of the diverse electrochemical variables associated with uric acid.

[0145] The feature extraction and analysis method was further applied to demonstrate that tyrosine has distinct electrochemical properties from uric acid (see FIG. 18A to FIG. 181). Due to differences in redox activity, charge state, and other solution properties in artificial saliva, z’p, Ep, and W1 / 2 for tyrosine responded to changes in pH and electrolytes differently from uric acid. For example, while ipincreased linearly with the concentration of tyrosine, the sensitivities differed compared to uric acid: for CV-1, it ranged from 0.71 to 0.26 μA μm cm-2while for CV-2 it was 0.44 to 0.13 μA μm-1cm-2, and for DPV, it was 0.27 to 0.08 μA μm cm2(see FIG. 19). These results show that EC-omics features extracted from voltammetric spectra contain rich electrochemical information that may be used to profile electroactive species in biofluids.

[0146] To illustrate the importance of feature extraction, the electrochemical spectra of four representative electroactive molecules found in the protein synthesis pathway were examined: tyrosine (an essential amino acid), angiotensin II and oxytocin (peptides), and mucins II (a glycoprotein). The CV and DPV spectra of all four electroactive species were collected in artificial saliva at pH 5, 7, and 9 (sec Inset (a-b) of FIG. 20). Because these molecules contain segments of tyrosine, their spectra overlap within the redox potential range of 0.4-0.8 V. However, the original voltammetric spectra and the heatmaps of extracted EC-omics features (ip, Ep, W1 / 2 for these molecules show distinct responses to pH changes (see Inset (c) of FIG. 20). The noticeable chemical shifts in the proton nuclear' magnetic resonance (1H NMR) spectra of tyrosine segments demonstrated the distinct equilibrium of these metabolites at different pH levels, further validating the varying orientations of tyrosine segments within different peptide / protein sequences (see Inset (d) of FIG. 20). A principal component analysis on the extracted features was subsequently conducted to distinguish the four electroactive species. By combining the EC-omics features, it was possible to effectively discriminate these four electroactive species (see Inset (e-f) of FIG. 20), a functionality that was not available with the original or conventional methods of feature extraction. These findings underscore the potential of the proposed EC-omics approach to generate a more holistic biomolcculc profile in situ and suggest a broad range of applications in constructing networks for more accurate disease diagnosis.

[0147] Parsing a saliva EC-omics database

[0148] In another study, the proposed EC-omics was applied to investigate clcctroactivc species in saliva, a representative peripheral biofluid with potentially abundant diagnostic bioinformation. Saliva samples from 75 human subjects were collected together with their personal and clinical information including age, gender, height, weight, and oral health states (see FIG. 21A). To avoid purine-rich foods that could affect the saliva composition, the samples used spitting methods following standard protocols were collected in the morning. With the user-friendly portable system of the present device, voltammetric spectra of the saliva samples could be promptly collected without additional preparation to form a saliva database (salivaDB) containing personal and clinical information for 75 human subjects (with anonymous identifiers), such as age, gender, height, weight, and oral health status. Correspondingly, the CV spectra (CV-1 and CV-2) and DPV spectra, as well as theNMR spectra of the saliva samples from these subjects, were integrated into the saliva database. The CV, DPV, and ’H NMR spectra were employed for the comparison of EC-omics with untargeted NMR techniques and the accurate diagnosis of periodontitis.). EC-omics features extracted from the voltammetric spectra in the saliva database of 75 samples were then combined to form an EC- omics database that may be used for diagnosis. The EC-omics approach was validated by comparing it with the periodontitis-related biomarkers (lactate, histidine, glycine, tyrosine, and phenylalanine), and untargeted NMR features extracted fromNMR spectroscopy — the gold standard in traditional omics technologies.

[0149] For feature extraction, the CV and DPV spectra were categorized into three primary regions based on the redox potential of the representative electroactive species: pkpr(0.4-0.8 V) representing peaks related to proteins and their associated metabolites, pkua(0-0.4 V) representing peaks for uric acid and compounds with nearby redox potential, and pkmu(-0.5- 0 V) representing peaks for other electroactive metabolites (see FIG. 2 IB). To profile the diverse electrochemical reactions of the electroactive molecules at the working electrodes, the zp, Ep, and W1 / 2 were calculated for each region and the total number of voltammetric peaks (num). In addition, to capture the overall redox reaction and charge transfer process at the electrode solution interface, the oxidation area (Sp), reduction area (Sn), and the ratio of .S'ptoSn(Sr) were extracted from the CV spectra. Averaging the EC-omics features from the four working electrodes, 39 peak-related features (zp, Ep, W1 / 2, and num) and 9 area features (Sp, Sn, and Sn) for all 75 saliva samples were obtained. These features were combined to form a saliva EC-omics database (see Table 2).Table 2. Summary of the EC-omics features from the voltammetric spectra, periodontitis-related molecular biomarkers, and untargeted NMR features from 'H NMR spectra. Asterisks (*) denote pairsample t-tests based on saliva EC-omics to discriminate health and periodontitis patients, p < 0.05.[0150J The1H NMR spectra were utilized to analyse periodontitis-related molecular biomarkers and extract untargeted NMR features. Typical NMR peaks with minimal overlap with other metabolites for lactate, histidine, glycine, tyrosine, and phenylalanine were identified according to their chemical shifts and integral to calculate the abundance of each molecular biomarker (see FIG. 22A). For the untargeted NMR features, theNMR spectra were extracted and classified into six sections (nmr1-6) based on chemical shifts of protons in different metabolites. Each section had two types of features: NMR peak number (Nnmr), whichreflects the diversity of protons contained in the various metabolites, and NMR peak area (Snmr), which relates to their concentration (see FIG. 21C).

[0151] Next, paired Pearson correlation analysis was performed to compare electrochemical features with the abundance of periodontitis-related molecular biomarkers and untargeted NMR features. The abundances of molecular biomarkers were investigated as representatives of molecule-centric diagnostic approaches, and the 12 untargeted NMR features (Snmrl-6 and A'niiirl 6) served as references for the diversity and abundance of metabolites in saliva. As shown in FIG. 22B and FIG. 21D, both the ipand area features had more than 10 correlation pairs with substantial positive correlations (Pearson correlation coefficients > 0.30) when correlated to the abundance of molecular biomarkers. They respectively showed median Pearson correlation coefficients of 0.30 and 0.35, which were positively correlated with the untargeted NMR features (see FIG. 22C and FIG. 2 IE). This indicates that both ipand area are substantially associated with molecular biomarkers in traditional omics technology, which is consistent with conventional electrochemical analysis, where ipand area are typically used to calculate the concentration of analytes in biofluids. In contrast, Ep, VV 1 / 2, and other EC-omics features correlated poorly with molecular biomarkers or untargeted NMR features, suggesting that the EC-omics database contains distinct bioinformation that is potentially complementary to traditional metabolomics databases (see FIG. 22B, FIG. 22C, and FIG. 23) Additionally, by visualizing connections with Pearson correlation coefficients above 0.50, it was confirmed that various EC-omics features were closely interrelated, forming a network of valuable electrochemical variables (see FIG. 22D). This demonstrated that the proposed EC-omics method could extract a more holistic view of metabolic processes and open up opportunities formore diagnostic or other healthcare applications previously inaccessible using conventional molecule-centric methods.

[0152] Leveraging saliva metabolic dynamics for accurate periodontitis diagnosis

[0153] As a proof-of-concept, EC-omics was applied to diagnose periodontitis, a serious oral inflammatory disease that is epidemiologically associated with various systematic inflammation-driven disorders. The onset of periodontitis leads to changes in metabolic networks, and clcctroactivc metabolites serve as indicators of the dynamic states of the metabolic networks (see FIG. 24A), which were clinically diagnosed using radiographs and periodontal pocket probes (see FIG. 24B). To validate the diagnostic performance of theportable EC-omics approach, both voltammetric spectra and!H NMR spectra of saliva samples from 34 healthy and 41 periodontitis subjects were collected and analysed (see FIG. 25).

[0154] To visualize the changes in metabolic dynamics between healthy individuals and periodontitis patients, 6-axis radar charts of representative EC-omics or NMR features were plotted, with each axis embodying the kernel-smoothed distribution of one feature. The distinct patterns of radar charts and embedded kernel distribution for each EC-omics feature (zp, Ep, Wi / 2. and area) demonstrated the obvious variation of the estimated data density among the healthy and periodontitis groups (sec FIG. 24C), which was co-validatcd with the periodontitis- related molecular biomarkers and untargeted NMR features (see FIG. 24C and FIG. 26A). These results demonstrated that the proposed EC-omics method enables a more comprehensive profiling of the metabolic dynamics of saliva with its interconnected features. Moreover, the paired sample l-lesls revealed clear differences (p < 0.05) in 23 out of 48 EC-omics features, 3 of the 5 molecular biomarkers, 8 out of the 12 NMR features between the healthy and periodontitis groups (see FIG. 24D and Table 2). These results indicated the complementary nature of the electrochemical approach to molecule-centric biomarkers and NMR technologies when profiling metabolic dynamics. Advantageously, the proposed EC-omics method provides easy accessibility to a practical electrochemical profiling platform that can find a wide range of potential uses in various healthcare applications.

[0155] To more accurately discriminate between healthy subjects and periodontitis patients, a support vector machine (SVM) was used to comprehensively analyse and compare the concentration-related ipfeatures, periodontitis-related molecular biomarkers, comprehensive EC-omics features, and untargeted NMR features. All samples in each dataset were randomly divided into training (60%) and test (40%) sets for each machine learning trial, and the test accuracy was obtained by averaging the accuracies of 500 independent trials (FIG. 27 illustrates one typical SVM trial). With the data obtained from the same saliva samples, the diagnostic accuracy of the EC-omics features (93%) surpassed that of the untargeted NMR features (89%) and significantly outperformed the concentrated-related ipfeatures (57%) and periodontitis -related biomarkers (70%) (see FIG. 24E). The enhanced diagnostic accuracy of the electrochemical approach stems from the integration both the concentration-related ipfeatures and many additional multifaceted EC-omics features (Ep, W1 / 2, num, and area). This allowed a more holistic view of the metabolic dynamics indicative of periodontitis thus improving the diagnostic capabilities. Moreover, the EC-omics approach was alsodemonstrated to be compatible with many other machine learning algorithms, such as Adaptive Boosting (AdaBoost), and Neural Network (NeurNet), achieving a comparable diagnostic accuracy to that based on the NMR database (sec FIG. 26B and FIG. 26C). To further verify, varying levels of white Gaussian noise was introduced to the voltammetric spectra in the saliva database to simulate environmental noises (see FIG. 26D). The EC-omics approach could nevertheless provide consistent diagnostic accuracy across diverse intelligent algorithms, demonstrating robust diagnosis capability in external disturbances (see FIG. 26E).

[0156] Electrochemical theories for feature extraction

[0157] The CV-DPV data may be described as the “fingerprint” data from which various EC-omics features can be extracted based on electrochemical theories. The following are examples of EC-omics features that could be extracted from the cyclic voltammetry (CV) and differential pulse voltammetry (DPV) spectra obtained using the proposed CNT / BC electrodes: peak information (Ep, ip, W1 / 2) and area information (5P, Sn, Sr). Here, the physics of each feature in different characterization techniques is summarized.

[0158] In an electrochemical reaction approaching the equilibrium, the electrode potential φφeqis correlated to the concentration of the oxidizing agent and the concentration of reducing agent through the Ncrnst relation.

[0159] For CV spectra, the voltage is linearly changed with a constant scan rate and the corresponding current response. When the electrochemical process is reversible, the wave- shape of the voltammogram can be characterized by:Where E1 / 2 corresponds to the potential at which half the peak current is observed, n is the number of transferred electrons, and Epis the voltage at which the peak occurs. The peak current can be calculated as follows:Where F is the Faraday constant, A is the surface area of the test electrode, c is the concentration of the reactants, D is the diffusion coefficient of the reactants, v is the scanning rate, R is the gas constant, and T is the temperature. The peak current is strongly correlated to the number of transferred electrons, electrode surface area, diffusion coefficients, scanning rates, andtemperature. As the electrode surface, scanning rates, and temperature arc given, peak current mainly includes information from the number of transferred electrons and diffusion coefficients. The difference between the peak potential and half-peak potential is:

[0160] Thus, the half-width at half-height is strongly correlated to temperature and number of transferred electrons.

[0161] For DPV spectra, it features by its eliminations of the contribution of non-faradaic (capacitive) processes, which are effectively subtracted out. Moreover, it is useful for resolving the voltammetric signals due to two species with close half-wave potentials, producing easily quantifiable-peak-shaped responses. For a reversible system, the peak current:WhereThe peak occurs at a potential:The half- width at half-height:

[0162] It correlates to the number of transferred electrons in the electrochemical reaction and temperature through a constant of 3.52R / F. Therefore, it may be used to identify the average number of transferred electrons in the electrochemical reactions to figure out the reaction paths.

[0163] Materials and reagents

[0164] Uric acid, L-tyrosine, L-tryptophan, sodium chloride, guanosine, potassium chloride, sodium hydroxide, disodium phosphate, potassium dihydrogen phosphate, urea, hydrochloric acid, sodium dodecyl sulfate (SDS), dimethyl formamide (DMF), polyvinyl butyral (PVB), Angiotensin IT (human), Oxytocin, and Mucin IT (from porcine stomach) were sourced from Sigma Aldrich and used without further purification. Polyimide (PI) tape (25 pm, adhesion force 200-400 g) with a release layer (50 pm) was obtained from Kaptape. Nylon filtering film (pore size 0.45 pm) was purchased from Haining Zhongli Filtering Equipment Factory. Signalwall carbon nanotubcs (CNT, XFN15) were purchased from XFNano. Bacterial cellulose aqueous suspension solution (0.73 wt.%) was purchased from Sciencek. For screen printing inks, the Ag / AgCl paste (AGCL-675) was purchased from Kayaku Advanced Materials, and carbon and Au paste for screen printing working electrodes was purchased from Poten Technology Co., Ltd. The Au source (purity of 99.99%.) for thermal evaporation was sourced from Kurt J. Lesker.

[0165] Fabrication of the CNT / BC electrodes

[0166] For the patterning of disposable electrodes, PI tape (25 pm) with a release layer (50 pm) was patterned with a 50 W CO2 laser cutter (Universal Laser System) with optimized parameters: power 25%, speed 60%, points per inch (PPI) 1,000 in vector mode. The patterned PI tape was then adhered to the Nylone film (pore size 0.45 pm). Meanwhile, CNT powders were washed successively with 1 M HC1 aqueous solution, toluene, and DMF to remove the impurities in CNT under tip sonification. The CNT (0.15 g L1) and SDS surfactant (2.25 g L" ') in DI water were homogenized using tip sonification (UP200 Ht, Hielscher, parameters: power 100 W, interval: 60%, amplification 60%). The CNT / SDS dispersion was then centrifuged (2000 rpm, 5 min) to remove the particles and filtered using the Pl-patterned Nylon filtering film with a customized filtration setup (quartz sand pore size 90-150 pm, vacuum above 50% atm) to obtain the CNT / BC (as working and counter) electrodes at different weight ratios. Next, the Ag / AgCl paste (AGCL-675) was applied to the symmetrical centre of the four working electrodes, and the Laser cutter patterned PI tape was used to standardize the working electrodes into 2.8 mm diameter circles. Finally, a 6 pL solution of Menthol containing PVB (79 g L-1) and NaCl (50 g L-1) was carefully dropped onto the Ag / AgCl reference electrode to achieve a stable reference potential across the solution electrolytes.

[0167] Characterization of the CNT / BC electrodes

[0168] The scanning electron microscopy (SEM) images of the CNT and CNT / BC electrodes were collected by using a field-emission SEM (JEOL, FESEM-7600F). The contact angle tests were conducted on a contact angle analyser (LMS Technologies Pte Ltd), where a drop of H2O (2 pL) was applied to the horizontal surface of the CNT or CNT / BC electrodes. The sheet resistance was measured with an M-3 Mini type four-probe tester (Suzhou Jinggc Electronic Co., Ltd).

[0169] Custom printed circuit board (PCB) and mobile app design

[0170] A custom mobile application was developed with the cross-platform Flutter framework. The mobile application was able to wirelessly communicate with wearable devices via Bluetooth to send commands and acquire, process, and record the salvia profiling spectra. The application established a secure Bluetooth connection to the wearable sensor, where the user can switch over to the measurement page that recorded both the cyclic voltammetric (CV) spectra and differential voltammetric spectra (DPV). These measurement data were added to the list of historic analyte levels on the home page. The entire system operated at 3.3 V, enabling a simple and efficient design. A single LD39050 voltage regulator converted the power from a 3.7 V Li-ion battery to a stable 3.3 V. During data acquisition and transmission, the PCB drew 27 mA from the 3.7 V Li-ion battery. Using a 400-mAh battery, the device could perform approximately 5 hours of continuous measurements. The parameters of CV and DPV analysis were optimized at an electrochemical workstation (CHI 832C), and the spectra were compared.

[0171] Reproducibility of the disposable electrodes

[0172] To assess the reproducibility of the disposable electrodes, simulations were conducted to evaluate the consistency of the four working electrodes using COMSOL Multiphysics, a finite clement analysis and solver software. These simulations were performed within the Electroanalysis cyclic voltammetry module, utilizing the actual geometric dimensions of the disposable electrodes. Three continuous cyclic voltammetric scans were applied to the four working electrodes, with the potential ranging from -0.8 V to 1.0 V, and then back to -0.8 V. In addition, CV-1, CV-2, and DPV spectra of a PBS solution containing uric acid (160 μM) and tyrosine (200 pM) arc collected using the electrochemical sensing platform. From these spectra, the redox potential (Ep), redox current (ip), and half-peak width (W1 / 2) were extracted from the four working electrodes. The standard deviation of these values across the eight sensors was then calculated to investigate the reproducibility of the disposable electrodes.

[0173] Electrochemical properties of uric acid and tyrosine in artificial saliva

[0174] Uric acid and tyrosine were selected as model electroactive molecules for analysis by the electrochemical sensing platform with the working electrodes. First, artificial saliva containing disodium hydrogen phosphate (0.6 g L '), potassium chloride (0.4 g L1), sodium chloride (0.4 g L-1), and urea (4.0 g L-1) was prepared based on the reference with minor modifications. This composition was chosen as the 100% electrolyte concentration and wasdiluted to various electrolyte concentrations. Next, different solutions with gradients of pH (5- 9), electrolyte concentration (25-100%), and concentrations of uric acid (40-160 μM) or tyrosine (50-200 pM) arc prepared. The corresponding CV-1, CV-2, and DPV spectra were collected using the electrochemical sensing platform. The parameters for CV-1 and CV-2 were as follows: quiet time, 2 s, scan rate, 0.1 mV s’1; scan voltage from -0.8 V to 1.0 V and back to -0.8 V (CV-1), followed by a CV-2 scan from -0.8 V to 1.0 V and back to -0.8 V. The parameters for DPV were as follows: initial E, -0.8 V; final E, 1.0 V; incremental E, 0.008 V; amplitude, 0.05 V; pulse period, 0.5 s. and the second half-scan: initial E, 1.0 V; final E, -0.8 V; decremental E, 0.008 V; amplitude, 0.05 V; pulse period, 0.5 s. Feature extraction was conducted with MATLAB to analyze the uric acid (potential range 0-0.4 V) and tyrosine (potential range 0-0.4 V). Ep, zp, and Wm features were extracted from CV-1 , CV-2, and DPV spectra. These values were then plotted into contour maps and organized with a cubic format to illustrate the responses of features to different solution parameters.

[0175] Characterization of representative electroactive species in metabolic networks

[0176] Seven representative clcctroactivc species involved in the central dogma and protein synthesis were analysed. Artificial saliva solutions of tyrosine, angiotensin II, oxytocin at 200 pM, and mucin II at 2 g L’1were adjusted to different pH (5, 7, 9), and tested with the electrochemical sensing platform. Meanwhile, H NMR spectra for these solutions were analysed following the procedures outlined in the "NMR Characterization and Spectra Analysis" section. Feature extraction was carried out using MATLAB to get the Ep, zp, and W1 / 2 features for the features of tyrosine, angiotensin II, oxytocin, and mucin II at 0-0.4 V were also extracted. The obtained Ep, zP, and W1 / 2 features from CV-1, CV-2, and DPV were plotted as heatmaps and further subjected to principal component analysis (PCA) to discriminate these four biomolecules.

[0177] Human subject recruitment

[0178] The validation and evaluation of the electrochemical sensors were performed using human subjects in compliance with all the ethical regulations under protocols approved by the relevant institutional review board. All of the participating subjects (aged over 18 years) recruited gave written informed consent before study participation. All saliva samples were collected using the spitting method, utilizing saliva collection tubes equipped with funnels (Biorisc). The saliva collection tubes were sterilized using an autoclave (Hirayama HG-50). Saliva naturally accumulates in the floor of the mouth without additional stimuli, and subjectsspit the saliva into the collection tubes, repeating the process until approximately 3 mL of saliva samples are obtained. The collected samples are then subjected to the collection of voltammetric spectra and1H NMR spectra.

[0179] rH NMR characterization and spectra analysis

[0180] Solution preparation: The testing solution was prepared according to the reference with few modifications, and filled the NMR tube (600 pL). As example of the testing solution includes 540 pL of saliva plus 60 pL of D2O solution (containing 10 mM TSP as reference peak, 30 mM NaN , for stabilizing saliva). Certain contents in the saliva may undergo degradation when stored at room temperature for extended periods. Therefore, the saliva samples were kept in the -20 °C fridge before the sample preparation. The1H NMR spectra of samples were measured using a specified water suppression pulse sequence such as I D NOESY-presat, which employed the first increment of a NOESY pulse sequence with water irradiation during the relaxation delay and also during the mixing time (on a Bruker instrument, 600 M). The protocol followed the sequence RD-90p-t-90"-tm-90n-ACQ, where RD represented the relaxation delay, t was a short delay of approximately 3 ps, 90° represented a 90-dcgrcc RF pulse, tm was the mixing time, and ACQ was the data acquisition period. The mixing time was set to tm = 100-150 ms, the spectral width = 20 p.p.m., the number of time-domain data points = 32,768, the relaxation delay was RD - 2.0 s, the acquisition time was 1.36 s, and the number of scans was n = 128. The obtained H NMR spectra were processed with baseline correction, phase correction, and calibration to the internal TSP reference peak (chemical shift at 0.0 ppm) for the next procedures.

[0181] Next, theNMR peaks of periodontitis-related biomarkers were identified according to the different chemical shifts of hydrogens on different chemical structures: lactate (1.31-1.35 ppm), histidine (3.11-3.17 ppm), glycine (3.52-3.58 ppm), tyrosine (6.88-6.92 ppm), and phenylalanine (7.30-7.45 ppm). The peak areas of these NMR peaks were calculated to obtain the relative abundances of these five biomarkers. For the untargeted NMR feature extraction, theNMR spectra were extracted and classified into six sections based on chemical shifts of protons in different metabolites: nmr 1 (0.5 ppm-1.5 ppm), nmr 2 (1.5 ppm- 2.5 ppm), nmr 3 (2.5 ppm-3.5 ppm), nmr 4 (3.5 ppm-4.5 ppm), nmr 5 (5.2 ppm-6.5 ppm), and nmr 6 (6.5 ppm-9.0 ppm). The NMR peak number (Nnmr), and NMR peak area (Snmr) were extracted for each section and combined to compare with EC-omics features.

[0182] Feature extraction to build an EC-omics database

[0183] Integration of electrochemical and systems biology knowledge into the measured electrochemical spectra (CV-1, CV-2, and DPV) was realized via a feature extraction process. As different metabolites have different compositions that oxidizcd / rcduccd at different potentials during the electrochemical measurements, the measured CV-1, CV-2, and DPV spectra of saliva samples from 75 subjects were roughly divided into three regions first: proteins and their associated metabolites (O.4-O.8 V), uric acid, and nearby compounds (0-0.4 V), and other electroactive metabolites (-0.5-0 V). Then the theory of electrochemistry was adopted. When the baseline type was selected or clicked during feature extraction, the software will automatically find the max current value in the selected zone according to the respective MATLAB codes. Data points used for the peak analysis were within a zone starting at 35% (of an area defined by the first cursor and the max value) and ending at the location of a second cursor. The data point zone for the analysis was shown in the Selection frame by the extrema. The results of the analysis were displayed in the Results frame. Several values for the peak were displayed: the redox potential (Ep), the redox current (ip), the half-peak width (W1 / 2), and other parameters were calculated. The acquired EC-omics, i,> features, periodontitis-related biomarkers, and untargeted NMR features were analyzed with the paired correlations, Kernel smooth distribution, radar plots with the software Origin 2020b. Next, these features were subjected to support vector machine (SVM), Adaptive Boosting (AdaBoost), and Neural Network (NeurNet) algorithms for periodontitis diagnosis. Furthermore, to assess the antiinterference capabilities of the EC-omics strategy, varying levels of white Gaussian noise were introduced. The Signal-to-Noise ratio (SNR) was calculated as SNR = lOxlog (Signal / Noise), and the same feature extraction processes were carried out for periodontitis diagnosis under these conditions.

[0184] Acquisition of periapical radiographs of healthy and periodontitis subjects for auxiliary examination

[0185] Periapical radiographs of patients as an auxiliary examination were obtained using the intraoral X-ray system (Gendex epert® DC, KaVo, Biberach, Germany) with exposure parameters of 65 kV, 7 mA, and 0.02-0.50 s according to the age and teeth position of the patients.

[0186] Statistical analysis.

[0187] For statistical analyses, MATLAB, Origin, and Excel were used. Machine learning for the discrimination was conducted on Python. All replicate numbers, error bars, P-values, and statistical tests were indicated in the figure legends.

[0188] The present disclosure provides an EC-omics approach that captures the holistic metabolic dynamics in peripheral biofluids by harnessing the electrochemical information at the electrode-biofluids interfaces for disease diagnosis. A portable EC-omics platform was developed and utilized to analyse human saliva. EC-omics features can be extracted from the obtained voltammetric spectra and integrated into a holistic EC-omics network, which significantly enhanced the accuracy of periodontitis diagnosis. Paired Pearson correlation analysis between the saliva EC-omics database and the NMR database shows the EC-omics database contains a wide range of features that align, diverge, and potentially complement traditional omics databases. Upon being reinforced with SVM, the EC-omics approach can discriminate periodontitis with 93% accuracy (based on 500 tests), superior to untargeted NMR features (89% in 500 tests), and significantly outperforms methods that rely solely on concentration-related iPfeatures (57% accuracy in 500 tests) and periodontitis-related molecular biomarkers (70% accuracy in 500 tests). Given electrochemical sensing techniques arc well-established, the EC-omics may be applied to a wide range of disease biomarkers by modifying the working electrodes with additional catalysts, biorecognition moieties, or enzymes. By customizing electrode substrates and their functionalities on working electrodes, the platform may be tailored to a wide range of biofluids and transformed into various sensor formats, offering new possibilities for disease diagnosis, the EC-omics approach, which expands the omics technologies from an electrochemical perspective, is expected to deepen the understanding of biomarkers and accelerate the transition toward a patient-centric, decentralized healthcare system that would benefit the masses and underserved populations.

[0189] From the aspect of accessibility and mass deployment in disease diagnosis. Conventional omics technologies in systems biology rely on large-footprint instruments or sophisticated procedures, which hinder their wide deployment among the general population. In contrast, the proposed electromics profiling and analysis platform of the present disclosure is patient-centric, sustainable, and suitable for mobile deployment, including but not limited to remote healthcare applications.

[0190] From the aspect of electrode performance, the proposed CNT / BC electrode of the present disclosure with unique structural, chemical, and catalytic properties, is characterizedby smooth voltammetric signals. As demonstrated with PBS solution, high resolution redox peaks in the presence of two model analytes (Uric acid and tyrosine) could be obtained, exhibiting advantageous performances over conventional thermally evaporated electrodes, commercial screen-printed carbon or gold electrodes.

[0191] From the aspect of information abundance, conventional devices (such as glucose or uric acid meter) and previously reported methods could only target limited biomarkers and focused on the concentration information. In contrast, the electromics profiling platform of the present disclosure contains two continuous CV curves for a high stoichiometric resolution and one circle of DPV spectra for a high voltammetric (spatial) resolution, which can provide much abundant and comprehensive bioinformation about the whole electroactive species and their inherent electrical features in the bulk biofluids.

[0192] From the aspect of diagnostic accuracy, the machine learning reinforced electromics strategy of the present disclosure experimentally demonstrated 79% accuracy on rhinitis diagnosis, which is comparable to that of NMR features (79%).

[0193] As mentioned above, the signal collection platform may be an clcctrochcmomics platform. The platform may be configured to apply one or more stimulation voltage waveforms to a sample to obtain voltammetric spectra from the sample, such as that illustrated in FIG. 13B. Further, the platform may be configured to obtain electrical data of the sample via the electrodes based on the one or more stimulation voltages waveforms applied, and further configured to extract one or more features from the electrical data. In other words, the platform may be configured to extract electrochemical features from the voltammetric spectra, including parameters such as redox potential (Ep), redox current (ip), half-peak width (W1 / 2), and / or area features (S), e.g., that illustrated in FIG. 13D. The signal collection platform may additionally be configured to integrate these electrochemical features into an interconnected network that can provide a holistic profile of biofluids or specific analytes within the sample, such as that illustrated in FIG. 22D. In other words, the platform may be configured to construct the database based on the electrical data of the sample. The abovementioned approach may allow for the simultaneous analysis of multiple components in complex biological matrices, potentially offering insights into the overall biochemical state of the system being studied.

Claims

Claims1. An electrode device, comprising: a substrate; one or more working electrodes on the substrate; a counter electrode on the substrate; and a reference electrode on the substrate; wherein the one or more working electrodes, the counter electrode, and the reference electrode are spaced apart from one other, and wherein the one or more working electrodes, the counter electrode, and the reference electrode comprise carbon nanotubes (CNT) and cellulose, a weight ratio of the carbon nanotubes (CNTs) to the cellulose selected from a range from 1 : 8 to 1 : 2.

2. The electrode device as recited in claim 1, wherein an end portion of each of the one or more working electrodes comprises an electrode pad, wherein an end portion of the reference electrode comprises an electrode pad, and wherein an end portion of the counter electrode is curved.

3. The electrode device as recited in claim 2, wherein the one or more working electrodes are more than one electrode, with electrode pads of the one or more working electrodes equidistantly spaced from the electrode pad of the reference electrode on the substrate, wherein the electrode pad of each of the one or more working electrodes is spaced from the curved end portion of the counter electrode with an identical minimum distance, and wherein a first distance between a first neighboring pair of electrode pads of the one or more working electrodes is equal to a second distance between a second neighboring pair of electrode pads of the one or more working electrodes.

4. The electrode device as recited in claim 2 or claim 3, wherein a total number of the one or more working electrodes is even.

5. The electrode device as recited in claim 4, wherein a number of the one or more working electrodes on a first side of the reference electrode is equal to a number of the one or more working electrodes on a second side of the reference electrode opposite the first side.

6. The electrode device as recited in any one of claims 2 to 5, wherein the electrode pads of the one or more working electrodes are configured with a same shape.

7. The electrode device of any one of claims 2 to 6, further comprising an insulation layer over portions of the one or more working electrodes, the counter electrode, and the reference electrode; wherein the electrode pads of the one or more working electrodes; the electrode pad of the reference electrode and the curved end portion of the reference electrode are not covered by the insulation layer.

8. The electrode device as recited in claim 7, wherein the insulation layer comprises polyimide (PI), polyethylene terephthalate (PET), polyurethane (PU) or screen printed epoxy resins.

9. The electrode device as recited in any one of claims 2 to 8, wherein a total area of the electrode pads of the one or more working electrodes is smaller than the curved end portion of the counter electrode.

10. The electrode device as recited in any one of claims 2 to 9, further comprising: a layer comprising silver (Ag) and silver chloride (AgCl) on the electrode pad of the reference electrode.

11. The electrode device as recited in claim 10, further comprising: a stabilization layer over the electrode pad of the reference electrode.

12. The electrode device as recited in claim 11, wherein the stabilization layer comprises: a chloride salt; and a hydrophobic polymer.

13. The electrode device as recited in any one of claims 2 to 12, further comprising a reservoir around the electrode pads of the one or more working electrodes, the curved end portion of the counter electrode, and the electrode pad of the reference electrode.

14. The electrode device as recited in any one of claims 1 to 13, wherein the cellulose is selected from the group consisting of bacterial cellulose, cotton cellulose, wood cellulose, and any combination thereof.

15. The electrode device as recited in any one of claims 1 to 14, further comprising one or more sensing moieties on the one or more working electrodes, the one or more sensing moieties being selected from the group consisting of polyaniline (PANi), Prussian blue (PB), a metalorganic framework (MOF), copper nanoparticles, an electrochemical catalyst, and any combination thereof.

16. The electrode device as recited in claim 15, wherein the sensing moieties arc polyanilinc (PANi) with a weight ratio of the carbon nanotubes (CNTs) to PANi selected from a range from 1:2 to 4: 1.

17. The electrode device as recited in claim 15, wherein the sensing moieties are Prussian Blue (PB) with a weight ratio of the carbon nanotubes (CNTs) to Prussian Blue (PB) selected from a range from 1 : 1 to 4: 1.

18. A method of making an electrode device, the method comprising: forming one or more working electrodes, a counter electrode, and a reference electrode on a substrate, wherein the one or more working electrodes, the counter electrode, and the reference electrode are spaced from one another, and wherein the one or more working electrodes, the counter electrode, and the reference electrode comprise carbon nanotubes (CNTs) and cellulose, a weight ratio of the carbon nanotubes (CNTs) to the cellulose selected from a range from 1 : 8 to 1 : 2.

19. The method as recited in claim 18,wherein the method of forming the one or more working electrodes, the counter electrode, and the reference electrode on the substrate comprises: mixing the carbon nanotubes (CNTs), the cellulose, and a dispersion medium to form a mixture comprising the carbon nanotubes (CNTs) and the cellulose; filtering the mixture through a patterned film, thereby forming the one or more working electrodes, the counter electrode, and the reference electrode on the patterned film; and transferring the one or more working electrodes, the counter electrode, and the reference electrode from the patterned film to the substrate.

20. The method as recited in claim 19, wherein the patterned film comprises a nylon film and one or more polyimide tapes on portions of the nylon film.

21. The method as recited in any one of claims 18 to 20, further comprising: depositing a layer comprising silver (Ag) and silver chloride (AgCl) on an electrode pad of the reference electrode.

22. The method as recited in claim 21, wherein the deposition of the layer comprising silver (Ag) and silver chloride (AgCl) on the electrode pad of the reference electrode is carried out using screen printing.

23. The method as recited in claim 21 or claim 22, further comprising: depositing an insulation layer over portions of the one or more working electrodes, the counter electrode, and the reference electrode, such that the electrode pads of the one or more working electrodes, the electrode pad of the reference electrode and the curved end portion of the reference electrode are not covered by the insulation layer.

24. The method as recited in claim 23, wherein the deposition of the insulation layer over the portions of the one or more working electrodes, the counter electrode, and the reference electrode is carried out using screen printing.

25. The method as recited in any one of claims 21 to 24, further comprising: depositing a stabilizing layer over the electrode pad of the reference electrode.

26. The method as recited in claim 25, wherein the deposition of the stabilizing layer over the electrode pad of the reference electrode is carried out using drop casting.

27. The method as recited in any one of claims 18 to 26, wherein the substrate is a polyimide (PI) substrate.

28. A signal collection platform comprising: the electrode device of any one of claims 1 - 17; and a signal collection circuit arrangement comprising a controller, the signal collection circuit arrangement in electrical connection with the electrode device, wherein the controller is configured to receive incoming wireless signals from a processing device and further configured to control the electrode device to apply one or more stimulation voltage waveforms to a sample based on the incoming wireless signals received, and wherein the controller is configured to obtain electrical data of the sample via the electrodes based on the one or more stimulation voltages w aveforms applied and configured to transmit one or more outgoing wireless signals to the processing device based on the electrical data obtained.

29. The signal collection platform as recited in claim 28, wherein the one or more stimulation voltage waveforms comprise a cyclic voltammetry (CV) voltage waveform having a first portion which increases linearly and a second portion which decreases linearly.

30. The signal collection platform as recited in claim 28 or claim 29, wherein the one or more stimulation voltage waveforms comprise a differential pulse voltammetry (DPV) voltage waveform having a plurality of portions with constant values over predetermined durations of time.

31. The signal collection platform as recited in any one of claims 28 to 30, wherein the controller comprises an analog-to digital converter (ADC), and wherein the signal collection circuit arrangement comprises a trans-impedance amplifier(TIA) in electrical connection with the analog-to digital converter (ADC).

32. The signal collection platform as recited in any one of claims 28 to 31, wherein the controller comprises a digital-to-analog converter (DAC); and wherein the signal collection circuit arrangement comprises a low pass filter (LPF) in electrical connection with the digital-to-analog converter (DAC).

33. The signal collection platform as recited in any one of claims 28 to 32, further comprising: a voltage regulator configured to provide a regulated voltage to the controller based on a supply voltage provided by a battery.

34. The signal collection platform as recited in any one of claims 28 to 33, further comprising: a processing device.

35. The signal collection platform as recited in any one of claims 28 to 34, wherein the processing device is a mobile device.

36. The signal collection platform as recited in any one of claims 28 to 35, wherein the sample comprises a biofluid comprising one or more electroactive species.

37. The signal collection platform as recited in any one of claims 28 to 36, wherein the one or more electroactive species comprise tyrosine, angiotensin II, oxytocin or mucin 11.

38. The signal collection platform as recited in any one of claims 28 to 37, wherein the processing device is configured to extract one or more features from the electrical data.

39. The signal collection platform as recited in claim 38, wherein the features comprise redox potential Ep), redox current (ip), half-peak width W1 / 2 , or area features (S) at different partitions according to redox properties of one or more biomolcculcs in the sample.

40. The signal collection platform as recited in claim 38 or claim 39, wherein the processing device is configured to compare the extracted one or more features with data from a database for making a disease diagnosis, and wherein the data from the database is based on different electro active species under different measurement conditions.

41. The signal collection platform as recited in claim 40, wherein the processing device is configured to construct the database based on the electrical data of the sample.

42. The signal collection platform as recited in any one of claims 28 to 41, wherein the signal collection platform in an electrochemomics platform.

43. A method of forming a signal collection platform, the method comprising: electrically connecting a signal collection circuit arrangement to the electrode device of any one of claims 1 - 17, the signal collection circuit arrangement comprising a controller, wherein the controller is configured to receive incoming wireless signals from a processing device and further configured to control the electrode device to apply one or more stimulation voltages to a sample based on the incoming wireless signals received, and wherein the controller is configured to obtain electrical data of the sample via the electrodes based on the one or more stimulation voltages applied and configured to transmit one or more outgoing wireless signals to the processing device based on the electrical data obtained.

Images