Electrochemical method for detecting diseases

Abstract

A method of diagnosing a person suspected of having cancer detectable in a biofluid, comprising : obtaining biofluid sample from the patient; acquiring an electrochemical signal generated jointly by redox molecules in the biofluid sample, wherein the signal is acquired with the aid of an electronic tongue; optionally preprocessing the electrochemical signal, to obtain processed data; applying trained chemometric model (s) to the processed or raw data; and diagnosing a type of cancer after the trained chemometric model has classified the processed / raw data.

Description

[0001] Electrochemical Method for Detecting Diseases

[0002] Background of the invention

[0003] Integration of electrochemical tools into diagnosis of diseases is based on the redox nature of many of the metabolites involved. An electrochemical approach offers reduced diagnostic time and potential miniaturization - an important consideration because a biofluid sample taken from a patient often has limited volume.

[0004] Electrochemical sensor arrays are divided into two major categories, utilizing either specific sensing or cross-reactive, partially selective sensing. In specific sensor arrays, each individual electrode is designed to target specifically just one analyte; it will recognize no other. In cross-reactive, partially selective sensor arrays, each individual electrode is purposely designed to sense more than one analyte. The limited selectivity of each individual electrode is compensated through signal processing using chemometric tools.

[0005] Sensors of the latter type, when used for voltammetric measurements, are named voltammetric electronic tongues, or voltammetric e-tongue. Voltammetry is an electrochemical technique which measures the current at a working electrode as the potential applied across the working electrode and a counter electrode is varied with time. When electroactive species are present in a tested sample, they undergo oxidation (or reduction) when the potential on the working electrode is sufficiently positive (or negative). The oxidation / reduction electrochemical reactions are manifested by an increase in the current (anodic or cathodic) measured; that is, creation of an electrochemical signal with magnitude and position characteristic of a given electroactive species . Diseases, in particular some types of cancers, are not necessarily manifested by the presence of a single, easily quantifiable redoxactive biomarker / metabolite in a biofluid sample. Rather, diseases are more likely to be associated with a redox fingerprint generated jointly by a group of redox-active metabolites. It is for this reason that voltammetric e-tongues lend themselves to disease detection in biofluids, owing to their potential to sense and capture redox fingerprint as a whole.

[0006] Figure 1 shows the concept of diagnosis, e.g., of some type of cancer, by running voltammetry measurements in a biofluid sample (e.g., urine sample) using a voltammetric e-tongue, to be inserted into the sample (the black cylinders stand for the working electrodes; counter and reference electrodes are also seen). The electrodes are connected to a potentiostat, which varies the potential across the electrodes to record the voltammograms. The voltammogram are then analyzed with the aid of a chemometric trained model, to determine if the test sample was taken from a cancer patient or from non-cancer patient.

[0007] For example, Pascual, L. et al. [Analyst 141, 4562-4567 (2016)], reported the use of a voltammetric e-tongue consisting of seven metallic working electrodes (iridium, rhodium, platinum, gold, silver, cobalt, and copper) to detect a signature of prostate cancer in urine samples with the aid of multivariate analysis.

[0008] In a recently filed co-assigned international patent application no. PCT / IL2023 / 051060 a different type of voltammetric e-tongue was described for detecting prostate cancer and bladder cancer in urine samples. Instead of working electrodes made of different metals, variation among working electrodes was achieved through the application of different coatings onto the surface of, e.g., carbon electrodes, generating partially-selective, cross-reactive electrodes, capable of holistic sensing of redox states. The invention

[0009] We have found that a voltammetric e-tongue can be designed to detect in blood (plasma / serum) and / or urine samples redox fingerprints assigned to different diseases, namely, different types of cancer and inflammatory bowel disease (IBD). These fingerprints are presumably generated by cell-free circulating redox-active metabolites. These circulating redox-active metabolites are substances that influence disease progression through redox reactions. They play a pivotal role in regulating critical cellular functions, including signal transduction, modulation of metabolic pathways, acceleration of cellular growth, and disruption of genomic integrity. Thus, metabolomic approaches provide a comprehensive snapshot of the body's metabolic state and can be exploited via electrochemical tools to detect various clinical indications, such as early-stage cancer detection, which could significantly improve treatment outcomes and enhance survival rates. Another potential clinical application is in managing Inflammatory Bowel Disease (IBD), a chronic condition characterized by lifelong gastrointestinal tract inflammation, with patients experiencing phases of relapse and remission. Predicting and managing disease relapse is challenging, and sudden worsening of IBD symptoms can occur without warning. Ineffective management of remission is directly linked to increased costs.

[0010] Voltammetric e-tongues

[0011] In its most general form, the voltammetric e-tongues is an assembly of a few sets of working electrodes, each set consisting of electrode (s) of the same type. By electrodes of the same type we mean either bare electrode that are made of the same material (e.g., carbon, noble metals), or electrodes coated with the same film material. The number of electrodes of type i is marked n±. For example, eight types (sets) of working electrodes are listed below: 1) a set consisting of one or more bare electrodes. Bare electrodes are preferably made of carbon or noble metals, e.g., gold, platinum, rhodium, and iridium.

[0012] 2) a set consisting of one or more electrodes coated with polysaccharide (e.g., chitosan) film; typical film thickness is from 1 to 50 pm, e.g., 5 to 20 pm OR a set consisting of one or more electrodes coated with polysaccharide (e.g., chitosan) film with conductive additives incorporated into the film, such as carbon nanotubes; typical film thickness is from 1 to 100 pm, e.g., 5 to 60 pm.

[0013] 3) a set consisting of one or more electrodes coated with reduced graphene oxide film; typical film thickness is 200 to 1,000 nm, e.g., 350 to 550 nm.

[0014] 4) a set consisting of one or more electrodes coated with platinum black film; typical film thickness is from 1 to 50 pm, e.g., 6 to 10 pm.

[0015] 5) a set consisting of one or more electrodes coated with M0S2, which was electrodeposited onto the electrode surface from a deposition solution by cycling a potential range A, which corresponds to the electrochemical double layer (EDL) potential region of the working electrode (above -0.4 V, e.g., from -0.3 to +0.7 V (vs Ag / AgCl)), at a scanning rate of at least 0.05 V-s-1, e.g., from 1.0 to 10.0 V-s-1(labeled MoS2A);

[0016] 6) a set consisting of one or more electrodes coated with M0S2, which was electrodeposited onto the electrode surface from a deposition solution by cycling a potential range B, which extends to more positive potentials than potential range A (extended EDL; 0 to +1.4 V (vs Ag / AgCl)) at a scanning rate of at least 0.05 V-s-1, e.g., from 0.05 to 2.0 V-s-1(labeled MoS2B); ) a set consisting of one or more electrodes coated with WS2, which was electrodeposited onto the electrode surface from a deposition solution by cycling a potential range A, which corresponds to the electrochemical double layer (EDL) potential region of the working electrode (above -0.4 V, e.g., from -0.3 to +0.7 V (vs Ag / AgCl)), at a scanning rate of at least 0.05 V-s-1, e.g., from 1.0 to 10.0 V-s-1(labeled WS2A); and

[0017] 8) a set consisting of one or more electrodes coated with WS2, which was electrodeposited onto the electrode surface from a deposition solution by cycling a potential range B, which extends to more positive potentials than potential range A (extended EDL; 0 to +1.4 V (vs Ag / AgCl)) at a scanning rate of at least 0.05 V-s-1, e.g., from 0.05 to 2.0 V-s-1(labeled WS2B).

[0018] Techniques of surface modification of electrodes to deposit the coatings set out above are described in earlier publications, e.g., in WO 2018 / 225058, WO 2022 / 157753 and WO 2002 / 137236, showing galvanostatic electrodeposition (chronopotentiometry), potentiostatic electrodeposition (chronoamperometry) and cyclic voltammetry electrodeposition. The preparation of precursor deposition solutions, namely, chitosan electrodeposition solution, chitosan-carbon nanotube electrodeposition solution, platinum- black electrodeposition solution, graphene oxide electrodeposition solution, M0S2 electrodeposition solution and WS2 electrodeposition solution are described in detail in the experimental section below. Their application onto electrodes, to fabricate a voltammetric e- tongue, is illustrated below (coating thickness can be measured by atomic force microscopy or profilometry). n± - the number of electrodes in each set (i=l, 2, ..., 8) - is usually up to 3, namely, l<n±<3 (sometimes more than one electrode is applied for repetition, but one electrode per type, i.e., a voltammetric e-tongue with the following composition of working electrodes was found to be useful: (a total of eight working electrodes).

[0019] One compact design of a voltammetric e-tongue with the composition set out above is shown in Figure 2. The voltammetric e-tongue comprises a ring-shaped counter electrode (labeled "AUX" for auxiliary), e.g., Pt ring with outer and inner diameters of 15 to 20 mm (e.g., 18 mm) and 21 to 25 (e.g., 22 mm), respectively. The reference electrode (labeled "REF", e.g., Ag / AgCl coated electrode), is centrically positioned; its surface area is about 30 to 40 mm2e.g., 36mm2. The working electrodes are placed in the annular space between the counter (outer) and reference (inner) electrodes, e.g., equally spaced apart from one another at distance of 4 to 8 mm, e.g., 6 mm from the center of the reference electrode. Each working electrode is disc-shaped with surface area of 25-30 mm2, e.g., 27mm2. The following combination of working electrodes was assembled: a bare (gold or carbon) electrode, a chitosan / CNT- coated gold / carbon electrode, a platinum black-coated gold / carbon electrode, a reduced graphene oxide-coated gold / carbon electrode, a MoS2A-coated gold / carbon electrode, a MoS2B-coated gold / carbon electrode, WS2A-coated gold / carbon electrode and WS2B-coated gold / carbon electrode.

[0020] Figure 3 is a photograph showing a small plastic housing for the ring-shaped array of electrodes of Figure 2. A screen-printed- electrode (SPE) array with the geometry shown in Figure 2, consisting of non-coated carbon electrodes, is commercially available. The electrodes are coated as previously explained and the surface-modified array is encased in the plastic housing (11), which is provided with a central circular opening (12) at its top. A biofluid sample (urine, plasma, serum, etc.) container (13) can be fitted to the opening, to discharge the liquid sample onto the electrodes. The electrical connectors (14) extending from the bottom of the plastic housing are also shown in Figure 3. In operation the electrodes are electrically connected to a potentiostat (15) or galvanostat which control the potential or current of the working electrodes, respectively, to create a data set of electrochemical signals (e.g., voltammograms) when the electrodes are in contact with the test biofluid sample (the sample may be pretreated, i.e., steps such as freezing and thawing or centrifuging) . The data set of electrochemical signals is analyzed by a processor applying one or more chemometric techniques, e.g., by multivariant analysis of variance.

[0021] Experimental results reported below show that a voltammetric e- tongue with the following composition of working electrodes: can detect various clinical states with high accuracy, sensitivity and specificity, by conducting voltammetry in blood (plasma, serum) and / or urine samples and analyzing the voltammograms against a trained model (which is described in detail below): colorectal cancer (voltammogram recorded in blood (plasma, serum) samples); lung cancer (voltammogram recorded in blood (plasma, serum) samples); early-stage cancer detection, distinguishing blood (plasma, serum) samples from colorectal cancer patients and lung cancer; (voltammogram recorded in blood (plasma, serum) or urine samples); relapse / remission of inflammatory bowel disease (IBD; Crohn's disease, Ulcerative Colitis (UC); voltammogram recorded in urine samples); and bladder cancer (early detection or recurrence; voltammogram recorded in urine samples). The voltammetric e-tongue of the invention is not limited to the design shown in Figure 2. For testing urine samples, one simple and straightforward configuration that is useful is based on the use of an electrochemical measurement cup to hold the urine sample (e.g., not less than 3 ml sample), fitted with a perforated cover; the holes in the cover correspond in number and size to the electrodes, such that individual working electrodes can be inserted into the measurement cup through the holes to be immersed in the sample. Commercial counter electrode (e.g., commercial Pt wire) and commercial reference electrode (Ag / AgCl) are also inserted into the cup.

[0022] Another suitable design of a voltammetric e-tongue was shown in WO 2018 / 225058 and is reproduced below in Figure 4. It was based on a cylindrical body made of silicon, polyvinyl alcohol or polydimethylsiloxane, which was 2 to 5 cm long and with diameter is in the range from 2 to 3 cm. The accessible surfaces of the electrodes were deployed on one base of the tubular body: a discshaped reference electrode positioned concentrically and coaxially in respect to the cylindrical body, a counter electrode at the vicinity of the reference electrode and multiple surface modified working electrodes (3.14 mm2each) positioned in radial direction from the reference and counter electrodes and evenly distributed along the perimeter of the base of the cylindrical body. The opposite base provides the electrical wiring to be connected to potentiostat / galvanostat . When put to use, the electrochemical sensor is immersed in the biofluid sample to be analyzed such that the base of the cylinder, where the electrodes are disposed, is exposed to the sample allowing the electrodes that (optionally) protrude from the base to be dipped into the urine sample, creating the electrochemical cell for the measurements. Alternative designs based on microfabricated configurations can also be considered for the voltammetric e-tongue, especially when small blood samples are available for testing. One example is shown in Figure 5. An electrochemical sensor in the form of a microfabricated 1.5cm x 1.5cm chip (1) on a glass substrate is shown. It can be a portable device or can be placed in the lab. The device dimensions are compatible with the conventional microfabrication techniques where the diameter of the working microelectrodes (4) are ~100 micrometer and the diameter of counter electrode (3) is ~500 micrometer. The chamber (5) is designed to hold small volume samples (10-30 microliter). Reference electrode (2) can be integrated into the array by electroplating one or two microelectrodes with Ag / AgCl as previously described (see Example 6 of WO 2022 / 137236). There are two kinds of chambers, a small chamber for each microelectrode opening (4 and 3) and a bigger chamber to carry the fluid sample (5). The chambers are made of insulating polymer, e.g., SU-8 polymer (6). The contacts pads (7) can be connected via pogo pins (8) and then to the multichannel connection (9) of the potentiostat or galvanostat unit (10 not shown). The device may be powered by a battery or alternatively, can be connected to a main power supply. A control unit (not shown) is designed to serve several purposes, chiefly controlling the potential of the working electrodes or the current flowing through the cell, respectively, according to the chosen electrochemical technique.

[0023] The microsensor described above (which consists of microelectrodes, microchambers encompassing the microelectrodes, all confined within a recessed zone that serves as a receptable for holding the liquid sample) can be created over a substrate by techniques such as etching and photolithography. Briefly, a substrate is cleaned, a first photoresist is applied (either negative, positive or image reversal resist), e.g., by spin coating, spray coating or dip coating, to produce a thin uniform layer on the substrate, followed by soft baking. A first mask is aligned, to transfer the pattern corresponding to electrodes' sites onto the surface of the substrate. The photoresist is exposed through the pattern on the mask with UV light, followed by a development step. Next, bare microelectrodes are deposited in the intended sites, e.g., first titanium which serves as an adhesion layer and then gold followed by lift off procedure that resulted in a gold microelectrode array on glass substrate. In order to define the electrode effective surface area, another lithography step was done using, e.g., SU-8 photoresist. To define the chamber for fluid, another photolithography step was followed with e.g., thick SU-8 resist. Having patterned the microstructures on the substrate, the desired coatings are applied on the microelectrodes, for example, by electrodeposition. The fabrication of such microsensors by photolithography and etching, followed by surface modification by electrodeposition, is described in detail in WO 2022 / 137236 (see Figures 9A, 9B, 9C and 10 and Example 6 of WO 2022 / 137236).

[0024] Having recorded voltammograms with the aid of a voltammetric e- tongue as shown above, in a biofluid sample taken from a person suspected of having cancer or other condition, the sample is classified against a trained model (built separately for each disease mentioned above), to give the outcome of the diagnostic test. The trained model that has been built and stored for use, selects, with respect to each of said diseases, k significant voltage (E) values (k is an odd number), and associates a set of n working electrodes for each of said k potentials, enabling the creation of a first distribution, characteristic of biofluid samples from cancer patients, and a second, well-distinguished distribution, characteristic of biofluid samples from healthy people. The distributions are stored by the trained model. The classification of the test biofluid sample, i.e., to decide if it was taken from a cancer patient or non-cancer patient, comprises: calculating, for each of the k selected potentials, the distance between the observation vector (a current vector) measured in the test sample by a vector of the n chosen working electrodes and the first and second distributions, i.e., calculating Mahalanobis distances : where x is the sample observation vector, and p is the mean vector of either the disease / cancer or the non-disease / healthy group; determining, for a pair of Mahalanobis distances MDdisease and MDnon- disease calculated for a given voltage, which distance is smaller; diagnosing cancer or other disease state as appropriate, if in the majority of the pairs, MDdisease < MDnon-disease.

[0025] A similar approach is taken to determine clinical relapse or remission phases of IBD patients:

[0026] Model built through Multivariant Analysis of Variance (MANOVA)

[0027] To train a model, data is collected from biofluid samples of patients diagnosed with a disease, and a control group, as appropriate. The voltammograms are recorded, e.g., by differential pulse voltammetry, sweeping the potential in one direction, and then in the opposite direction. Hereinafter, we use the term "cyclic differential pulse voltammetry" to describe this method. A cyclic differential pulse voltammogram (consisting of a single cycle) is shown in Figure 6. The current (I) of a working electrode is measured as the potential (E) is swept at an appropriate scan rate from an initial potential to a switching potential in the positive direction, known as anodic sweep, or oxidation scan, and then the direction of the scan is reversed, namely, in the negative direction to more negative potentials, back, e.g., to the initial potential (cathodic sweep / reduction scan), or vice versa, the potential is initially swept negatively and then swept positively.

[0028] For example, cyclic differential pulse voltammetry was conducted from e.g., negative initial potential lying between -0.5 and -0.05V, e.g., -0.1V, to a positive switching potential above +0.7V, e.g., +1.0V (vs. Ag / AgCl), and in the reverse direction, with a scan rate of 0.01 to 0.05 V / sec, e.g., 0.02 V / sec, to produce a cyclic differential pulse voltammogram consisting at least a single cycle for each working electrode. That is, a total of eight voltammograms was produced for each biofluid sample.

[0029] Then each of the eight voltammograms recorded is divided into two I versus E sets of data: oxidation current versus potential and reduction current versus potential, resulting in a total of sixteen sets of I versus E data points, i.e., the currents measured for m potential points during oxidative and reductive scans (m is usually >100, e.g., 120<m<250; in the experiments reported below, m=157 or m=215) . All the data collected from a single biofluid sample using the voltammetric e-tongue of Figure 2 can be arranged in a matrix form as follows (#0 and #R identifies the type of working electrode and the oxidative / reductive scan, respectively; Ej is the index of a potential for which a current Ij was measured {j=l, 2, 3, ...m}, with conventional smoothing and baseline corrections (both are optional) :

[0030] Table A: A biofluid sample from cancer or IBD patients Table B: A biofluid sample from the control group

[0031] For the model, the division of a cyclic differential pulse voltammogram generated by each working electrode into two I(E) curves, resulted in practically doubling the number of working electrodes, as each electrode is assigned with the currents measured during positively sweeping the potential (oxidation of redox species) and negatively sweeping the potential (reduction of redox species) . Thus, for the purpose of the model, a voltammetric e-tongue consisting of eight working electrodes (like the one shown in Figure 2), can be viewed as sixteen electrodes-containing voltammetric e-tongue, labeled with the scan direction (forward / oxidation or backward / reduction).

[0032] Next, the data is split into a training and testing sets (70- 85 / 15-35%, e.g., 80% / 20% or other proportion), and crossvalidated to determine the best hyperparameters (i.e., selecting a subset consisting of k optimal voltage points, wherein k is an odd number; the magnitude of k is about 2 to 25% of m, for example, when m=157, k is an odd number between 3 and 30; and a combination of working electrodes per each selected voltage, variance and covariance).

[0033] During cross-validation, the train data is spilt into, e.g., five folds (or three folds, or other). Four folds are used as internal training data, the fifth fold is used for testing and selecting the model.

[0034] The cross-validation comprises iteration between sets consisting of k voltage values (k is an odd number of voltages, 3<k<30), with n-working electrodes being assigned to each voltage value in the set of k-voltages (based on the design discussed above, 2<n<16). That is, by iteration is meant that all possible permutations of odd k voltage values (k=3, 5, 7, ...19..., 29) and related combinations of n working electrodes are generated (n=2, 3, 4, 5,...,16). For each voltage, the working electrodes are sorted by their ANOVA value calculated between the disease group (or cancer group) and the control group (healthy group or non-disease group) to yield a dictionary such that the keys are the voltages, and the values are a sorted list of the best n-number of working electrodes per voltage, to select the best k-voltages by MANOVA using the n-selected working electrodes.

[0035] For example, the input of the model trained according to the present invention for early diagnosis of colorectal cancer has the following form, with k=5 and n=3 (the selected voltages are identified by their ordinal number in the range from 1 to m and the corresponding E potentials (vs. Ag / AgCl) and the selected working electrodes are designated with the aid of arrows, —* for the oxidative scan and <— for the reductive scan):

[0036] E4 (E=-0.079V):[Pt black—*, chitosan+CNT<—, MoS2A—*]

[0037] E5 (E=-0.074V):[Pt black—*, chitosan+CNT<—, MoS2A—*]

[0038] E9 (E=-0.054V):[Pt black—*, chitosan+CNT<—, MoS2A—*]

[0039] E157 (E=+0.707V):[chitosan+CNT<—, MoS2B—*, Pt black—*]

[0040] Ei58 (E=+0.712V):[chitosan+CNT<—, MoS2B—*, Pt black—*]

[0041] For the group of patients, for each of the selected k potentials, Ixn row vector is created, labelled |i_cancer (or p_disease), with the calculated averages pi, ...pnof the currents measured over each of the n-working electrodes at the given potential (averaging over all members of the cancer group). The corresponding nxn covariance matrix is also produced.

[0042] For the control group, for each of the k selected potentials, Ixn row vector is created, labelled p_healthy (or p_non-disease), with the calculated averages pi, ...pnof the currents measured over each of the n-working electrodes at the given potential (averaging over all members of the control group). The corresponding nxn covariance matrix (labelled COV_healthy is also produced.

[0043] That is, for each of the k selected potentials, the following vectors / matrices are generated and saved: p_cancer (Ixn vector); p_healthy (Ixn vector);

[0044] COV_cancer (nxn matrix); and COV_healthy (nxn matrix).

[0045] Next, the fifth fold is subjected to prediction by the model, by calculating, per each of the selected k potentials, the Mahalanobis distance of each (the prediction of each observation is determined by the minimal distance to the Cancer or Healthy Group). Because k is an odd number, a majority decision can be made.

[0046] Test sample classification:

[0047] For each test sample, the Mahalanobis distance to both the 'cancer group' and the 'healthy group' is calculated for each chosen voltage and selected array of working electrodes. A predictor is then selected for each voltage to minimize the Mahalanobis distance. Subsequently, a majority decision is made based on the outcomes from all predictors across the chosen voltages. where x is the sample observation vector, and p is the mean vector of either the cancer or the healthy group.

[0048] Table C below summarizes the key features of the diagnostic tool provided by the present invention, based on the experimental results reported below:

[0049]

[0050] The metrics for classification models were:

[0051] Accuracy, which is defined by: truepositive+truenegative

[0052] Accuracy = - :- totalpopulation

[0053] Sensitivity, which is defined by: truepositive

[0054] Sensitivity truepositive+falsenegative specificity, which is defined by: truenegative

[0055] Specificity truenegative+falsepositive

[0056] In addition to screening and diagnosing, the method of the invention can be used for cancer monitoring. By monitoring is meant, for example, tracking the recurrence and / or progression of cancer, classification of tumor according to WHO (World Health Organization) or EAU (European Association of Urology) guidelines, and effectiveness of treatment. Accordingly, the invention is primarily directed to a method of diagnosing a person suspected of having cancer detectable in a biofluid (e.g., in blood, plasma or serum samples, such as colorectal cancer and lung cancer), comprising: obtaining a biofluid (e.g., blood) from the patient; acquiring an electrochemical signal generated jointly by redox molecules in the biofluid sample, wherein the signal is acquired with the aid of an electronic tongue comprising at least two sets, e.g., at least three sets (or 4, 5, 6, 7, 8 sets) of working electrodes selected from: a set consisting of one or more bare electrodes; a set consisting of one or more electrodes coated with polysaccharide film, optionally with conductive additives incorporated into the film; a set consisting of one or more electrodes coated with reduced graphene oxide film; a set consisting of one or more electrodes coated with platinum black film; a set consisting of one or more electrodes coated with MoS2Afilm; a set consisting of one or more electrodes coated with MoS2Bfilm; a set consisting of one or more electrodes coated with WS2Afilm; a set consisting of one or more electrodes coated with WS2Bfilm; wherein the superscripts A and B indicate electrodeposited M0S2 and WS2 coatings produced by cycling across potential ranges A and B, respectively, wherein potential range A corresponds to the double layer potential region of the working electrode and potential range B extends to more positive potentials than potential range A; optionally preprocessing the electrochemical signal, to obtain processed data; applying trained chemometric model(s) to the processed or raw data; and diagnosing a type of cancer after the trained chemometric model has classified the processed / raw data. Preferably, the electrochemical signal is acquired by voltammetry, e.g., cyclic differential pulse voltammetry.

[0057] In one embodiment of the invention, the model applied was trained based on cyclic differential pulse voltammetry measurements in biofluid samples from cancer and non-cancer patients, to select k potential (E) values, where k is an odd number, and a group of n working electrodes for each of said k potentials, wherein said n working electrodes were selected based on currents measured as the potential was swept positively forward from Ei to Ef and negatively in reverse direction, or vice versa, to create a first distribution, characteristic of biofluid samples from the cancer patients, and a second, well-distinguished distribution, characteristic of biofluid samples from the non-cancer patients. The classification of the test biofluid sample comprises: calculating, for each of the k selected potentials, the distances between the current vector measured in the test sample by cyclic differential pulse voltammetry and the first and second distributions, given by Mahalanobis distances:

[0058] MDnon-cancer = MahalanobisDistancetoHealthyGroup9 Voltage( ) where x is the sample current vector, and p is the mean vector of either the cancer or the non-cancer group; determining, for a pair of Mahalanobis distances MDdisease and MDnon- disease calculated for a given potential, which of the two distance is smaller; diagnosing cancer, if in the majority of the pairs, MDdisease < MDnon- diseasej and vice versa.

[0059] The method of the invention may be a multi-cancer early detection method. Another aspect of the invention is a method of managing treatment of IBD patients by determining clinical relapse and remission phases, comprising: obtaining urine sample from the patient; acquiring an electrochemical signal generated jointly by redox molecules in the urine sample, wherein the signal is acquired with the aid of an electronic tongue comprising at least two sets, e.g., at least three sets (or 4, 5, 6, 7, 8 sets) of working electrodes selected from: a set consisting of one or more bare electrodes; a set consisting of one or more electrodes coated with polysaccharide film, optionally with conductive additives incorporated into the film; a set consisting of one or more electrodes coated with reduced graphene oxide film; a set consisting of one or more electrodes coated with platinum black film; a set consisting of one or more electrodes coated with MoS2Afilm; a set consisting of one or more electrodes coated with MoS2Bfilm; a set consisting of one or more electrodes coated with WS2Afilm; a set consisting of one or more electrodes coated with WS2Bfilm; wherein the superscripts A and B indicate electrodeposited M0S2 and WS2 coatings produced by cycling across potential ranges A and B, respectively, wherein potential range A corresponds to the double layer potential region of the working electrode and potential range B extends to more positive potentials than potential range A; optionally preprocessing the electrochemical signal, to obtain processed data; applying trained chemometric model(s) to the processed or raw data; and determining whether the IBD patient is in a clinical relapse phase or in a clinical remission phase, after the trained chemometric model has classified the processed / raw data. Preferably, the electrochemical signal is acquired by voltammetry, e.g., cyclic differential pulse voltammetry.

[0060] The model applied was trained based on cyclic differential pulse voltammetry measurements in urine samples from IBD patients in relapse and remission phases, to select k potential (E) values, where k is an odd number, and a group of n working electrodes for each of said k potentials, wherein said n working electrodes were selected based on currents measured as the potential was swept positively forward from Ei to Ef and negatively in reverse direction, or vice versa, to create a first distribution, characteristic of biofluid samples from IBD patients in a clinical relapse phase, and a second, well-distinguished distribution, characteristic of biofluid samples from IBD patients in a clinical remission phase.

[0061] The classification of the test biofluid sample comprises: calculating, for each of the k selected potentials, the distance between the currents vector measured in the test sample by cyclic differential pulse voltammetry and the first and second distributions, given by Mahalanobis distances: where x is the sample currents vector, and p is the mean vector of either the relapse or remission group; determining, for a pair of Mahalanobis distances MDrelapse and MDremission calculated for a given potential, which of the two distances is smaller; diagnosing clinical relapse phase, if in the majority of the pairs, MDrelapse < MDremission, and vice Versa. Another aspect of the invention is a method of diagnosing a person suspected of having bladder cancer, comprising: obtaining urine sample from the patient; acquiring an electrochemical signal generated jointly by redox molecules in the biofluid sample, wherein the signal is acquired with the aid of an electronic tongue comprising at least two sets, e.g., at least three sets (or 4, 5, 6, 7, 8 sets) of working electrodes selected from: a set consisting of one or more bare electrodes; a set consisting of one or more electrodes coated with polysaccharide film, optionally with conductive additives incorporated into the film; a set consisting of one or more electrodes coated with reduced graphene oxide film; a set consisting of one or more electrodes coated with platinum black film; a set consisting of one or more electrodes coated with MoS2Afilm; a set consisting of one or more electrodes coated with MoS2Bfilm; a set consisting of one or more electrodes coated with WS2Afilm; a set consisting of one or more electrodes coated with WS2Bfilm; wherein the superscripts A and B indicate electrodeposited M0S2 and WS2 coatings produced by cycling across potential ranges A and B, respectively, wherein potential range A corresponds to the double layer potential region of the working electrode and potential range B extends to more positive potentials than potential range A; optionally preprocessing the electrochemical signal, to obtain processed data; applying trained chemometric model(s) to the processed or raw data; and diagnosing bladder cancer after the trained chemometric model has classified the processed / raw data. Preferably, the electrochemical signal is acquired by voltammetry, e.g., cyclic differential pulse voltammetry.

[0062] The model applied was trained based on cyclic differential pulse voltammetry measurements in biofluid samples from bladder cancer and non-cancer patients, to select k potential (E) values, where k is an odd number, and a group of n working electrodes for each of said k potentials, wherein said n working electrodes were selected based on currents measured as the potential was swept positively forward from Ei to Ef and negatively in reverse direction, or vice versa, to create a first distribution, characteristic of biofluid samples from the bladder cancer patients, and a second, well-distinguished distribution, characteristic of biofluid samples from the non-cancer patients.

[0063] The classification of the test biofluid (urine) sample comprises: calculating, for each of the k selected potentials, the distance between the currents vector measured in the test sample by cyclic differential pulse voltammetry and the first and second distributions, given by Mahalanobis distances: where x is the sample currents vector, and p is the mean vector of either the cancer or the non-cancer group; determining, for a pair of Mahalanobis distances MDdisease and MDnon- disease calculated for a given potential, which of the two distance is smaller; diagnosing bladder cancer, if in the majority of the pairs, MDdisease < MDnon-disease, and vice versa. An alternative chemometric model that can be applied, e.g., in diagnosis of colorectal (CRC) patients, is a logistic regression model, e.g., Ll-regularized logistic regression model. For example, the voltammetric electronic tongue of the invention (consisting of eight different working electrodes as shown in Figure 2) can be used to record signals from blood / plasma samples from CRC patients and healthy controls. By sweeping the potential in the positive (oxidation) and negative (reduction) directions (or vice versa) and measuring the current at an appropriate number of distinct voltage points across the potential range, in each direction by each electrode, it is possible to create a set consisting of large number of datapoints (>2000). Then, with the help of Ll-regulazition regression the number of features is reduced. Such a model scored very well, exhibiting high prediction probability, as shown by the results of a study reported below.

[0064] Another aspect of the invention is an electrochemical disease (e.g., cancer) diagnostic / monitoring system, comprising: a voltammetric electronic tongue as described above; a counter electrode and optionally a reference electrode; a potentiostat to which the working electrodes, the counter electrode and optionally the reference electrodes are electrically connected to allow control of the potential and produce voltammograms when the electrodes are immersed in a test sample; a processor programmed to analyze the voltammograms by a trained model that classifies the test sample as described above.

[0065] Preferably, the voltammetric electronic tongue comprises working electrodes placed in an annular space between a ring-shaped counter electrode and a centrically positioned reference electrode.

[0066] The present invention provides a versatile diagnostic tool, because, as shown by the data reported herein, the voltammetric e- tongue, with an array of working electrodes consisting of: Bcarbon = Bchitosan+CNT = Bpt-black = nrGO = BMOS2A = HMOS2B = HWS2A = HWS2B = 1 can be used to conduct voltammetry measurements (with the aid of counter and reference electrodes) in a biofluid sample for diagnostics of wide a range of conditions. In fact, the invention provides a multi-cancer early detection method. That is, once a blood sample (or urine sample) is available, it may be screened in the lab for more than one type of cancer / suspected diseases. For each medical disease / condition, the classification is based on currents measured at each of the k potentials selected for the relevant cancer type, over the subset of selected n-working electrodes, as shown in Table C, to calculate the Mahalanobis distances and provide the outcome of the multi-cancer early detection test.

[0067] In the drawings:

[0068] Figure 1 shows the major components of a diagnostic method that can be applied with a biofluid sample that was taken from a patient and an array of electrodes to be contacted with the sample.

[0069] Figure 2 shows a design of a voltammetric electronic tongue.

[0070] Figure 3 is a photograph showing a plastic housing encasing the voltammetric electronic tongue of Figure 2.

[0071] Figure 4 shows a design of a voltammetric electronic tongue.

[0072] Figure 5 shows a design of a voltammetric electronic tongue as a microsensor .

[0073] Figure 6 shows a typical cyclic differential pulse voltammogram.

[0074] Figure 7 shows I versus E plots measured over MoS2Aelectrode, showing the means and standard deviations measured for CRC patients and healthy volunteers. Voltages values marked by vertical dashed lines show significant ANOVA differences.

[0075] Figure 8 is a graph showing the use of Ll-regularized logistic regression for feature selection in a classification model; at the chosen regularization point, the model retained 44 features.

[0076] Figure 9 is a graph showing the prediction probability of the Ll- regularized logistic regression model classifier. Examples

[0077] Preparations 1-6 Electrodeposition solutions

[0078] 1) Chitosan electrodeposition solution (1 wt.%)

[0079] 9 g chitosan powder was dissolved in 500ml DDW and stirred for 20 min. 10 ml 2M HC1 solution was added to the solution to reach pH of 5.5. The solution was sonicated for 45 minutes and then stirred again for 90 minutes at 700 rpm. The solution was filtrated with a metallic mesh 0.2 mm cylinder.

[0080] 2) Chitosan-carbon nanotube electrodeposition solution (1 wt.%) 200 mg carbon nanotube, multi walled, was added to 20 ml of the 1 wt.% chitosan solution and stirred for 15 minutes in 250-500 rpm. Then the solution was sonicated for one hour. The solution was used during a storage period of one week.

[0081] 3)Platinum-black electrodeposition solution

[0082] Platinum black deposition solution was prepared by mixing 0.5g of chloroplatinic acid and 25mg of lead acetate in 50 ml of DI water. The mixture was then stirred and 3.9 pL of concentrated hydrochloric acid (32%; 10.2 Molar concentration) was added to the solution. The prepared solution was covered with aluminum foil and stored at room temperature.

[0083] 4)Graphene oxide electrodeposition solution

[0084] 10 ml of 1 mg / mL rGO electrodeposition solution was prepared by diluting 2.5 ml of graphene oxide 4 mg / mL solution, with NaCl 100 mM in 5.5 ml DDW. The graphene oxide (GO) solution was prepared using a modified Hummers' method. A 9:1 ratio of sulfuric acid and phosphoric acid (100 mL) was prepared and stirred for several minutes. A graphite powder (7.5 g / L, 1 wt. eq.) was added to the mixture under stirring conditions. Potassium permanganate (45 g / L, 6 wt. eq.) was slowly added to the solution and the mixture was stirred for 6 h at 30-35 °C until the color turned to dark green. To eliminate the excess of potassium permanganate, hydrogen peroxide 30% w / w (2.5 mL) was added slowly and the mixture was stirred for 10 min, resulting in an exothermic reaction that was left to cool at room temperature. Concentrated 32% hydrochloric acid and DI were sequentially added at a 1:3 volume ratio and the resulting solution was centrifuged at 7000 RDM for 5 min. Residuals of the centrifuged solution were washed 3 times with hydrochloric acid and DI (1:3 v / v). The washed GO solution was dried at 90 °C in an oven (Binder- 9010-0082) overnight, yielding the GO powder.

[0085] 5)MOS2 electrodeposition solution

[0086] 200 ml of 0.1 mg / ml Molybdenum Disulfide 0.IM H2SO4 electrodeposition solution was prepared by diluting 20ml 1.0 mg / ml Molybdenum Disulfide solution, with 0.98 ml H2SO4 in 179.02 ml DDW. The mixture was sonicated for 10 min.

[0087] 6) WS2 electrodeposition solution

[0088] 200 ml of 0.1 mg / ml Tungsten Disulfide 0.IM H2SO4 electrodeposition solution was prepared by diluting 20ml 1.0 mg / ml Tungsten Disulfide solution, with 0.98 ml H2SO4 in 179.02 ml DDW. The mixture was sonicated for 10 min.

[0089] Preparation 7 Fabrication of the electrochemical biosensor

[0090] Surface modification of electrodes

[0091] A commercial screen-printed-electrode (SPE) with 8 working carbon electrodes of 2.95mm diameter was used (Metrohm DropSens). The working electrodes were sharing a silver reference electrode, and a carbon counter electrode. The SPE substrate material was ceramic. The geometrical arrangement of the electrodes is shown in Figure 2. Prior to coating, the surface of the electrode was cleaned by cyclic voltammetry using 0.5M H2SO4 sulfuric acid. The electrodes were further rinsed with double-distilled water. The electrochemical activity of the electrodes was validated by cyclic voltammetry (CV) measurement in 5mM ferrocyanide / ferricyanide solution. Then each electrode was rinsed with double distilled water (DDW) and coated with appropriate coating.

[0092] The following coatings were electrodeposited onto the carbon electrodes with Palmsens4 potentiostat:

[0093] Reduced graphene oxide coated electrodes (Working Electrode 1, Pin

[0094] Cyclic voltammetry (CV) technique was employed for the electrodeposition, cycling 3 times across the potential range of -1.4 V to 1.4 V (vs. Ag / AgCl), at a scan rate = 0.1 V / s. A commercial Ag / AgCl reference electrode (ALS, Japan) was used.

[0095] Molybdenum Disulfide Type A-coated electrodes (Working Electrode

[0096] 2, Pin 2):

[0097] Cyclic voltammetry (CV) technique was employed for the electrodeposition from the solution of Preparation 5, cycling 400 times across the potential range of -0.3 V to 0.7 V (vs. Ag / AgCl), at a scan rate = 1 V / s. A commercial Ag / AgCl reference electrode (ALS, Japan) was used.

[0098] Molybdenum Disulfide Type B-coated electrodes (Working Electrode

[0099] 3, Pin 3):

[0100] Cyclic voltammetry (CV) technique was employed for the electrodeposition from the solution of Preparation 5, cycling 400 times across the potential range of 0 V to 1.4 V (vs. Ag / AgCl), at a scan rate = 1 V / s. A commercial Ag / AgCl reference electrode (ALS, Japan) was used. Carbon electrode (Working Electrode 4, Pin 4)

[0101] Carbon electrode was left untreated. The electrode was rinsed in double-distilled water (milli-Q,18MQ) to remove residuals.

[0102] Tungsten Disulfide Type A-coated electrodes (Working Electrode 5, Pin 7):

[0103] Cyclic voltammetry (CV) technique was employed for the electrodeposition from the solution of Preparation 6, cycling 400 times across the potential range of -0.3 V to 0.7 V (vs. Ag / AgCl), at a scan rate = 1 V / s. A commercial Ag / AgCl reference electrode (ALS, Japan) was used.

[0104] Tungsten Disulfide Type B-coated electrodes (Working Electrode 6, Pin 8):

[0105] Cyclic voltammetry (CV) technique was employed for the electrodeposition from the solution of Preparation 6, cycling 400 times across the potential range of 0 V to 1.4 V (vs. Ag / AgCl), at a scan rate = 1 V / s. A commercial Ag / AgCl reference electrode (ALS, Japan) was used.

[0106] Chitosan-CNT coated electrodes (Working Electrode 7, Pin 9)

[0107] A chronopotentiometry technique was employed to electrodeposit chitosan from the solution of Preparation 2 onto carbon electrode over 5 mins, at cathodic current density of 0.595 mA / cm2, using commercial platinum wire and Ag / AgCl reference electrodes (ASL,Japan). The modified electrodes were rinsed in doubledistilled water (milli-Q, 18 MQ) to remove residuals.

[0108] Platinum-black coated electrodes (Working Electrode 8, Pin 10): A chronopotentiometry technique was employed to electrodeposit platinum black from the solution of Preparation 3 onto carbon electrode over 5 mins, at cathodic current density of 4.8 mA / cm2, using commercial platinum wire and Ag / AgCl reference electrodes (ASL,Japan). The modified electrodes were rinsed in doubledistilled water (milli-Q, 18 MQ) to remove residuals. Example 1 Voltammetry measurements in blood samples of patients diagnosed with colorectal cancer using an array of working electrodes and signal analysis

[0109] The goal of the study was to differentiate between

[0110] A) the redox fingerprints of patients diagnosed with early stages of colorectal cancer; and

[0111] B) the redox fingerprints of healthy volunteers.

[0112] Blood plasma sample preparation

[0113] 1 ml plasma samples were collected from thirty (30) volunteers from a biobank - fifteen (15) colorectal cancer (Stage I) and fifteen (15) healthy volunteers.

[0114] The samples were stored at -80°C. Before the use of samples, samples were defrosted at room temperature.

[0115] Electrochemical measurements

[0116] A drop of 1 ml defrosted plasma was placed on the modified commercial SPE, as shown in Figure 2. 2-ways differential pulse voltammetry was performed in the plasma samples across the potential range of -0.1V to +1.0V (Oxidation), and across the potential range of +1.0V to -0.1V (Reduction) with voltage step of 5mV, pulse mode of 50mV, pulse time of O.lsec, and scan rate of 0.02 V / sec. The electrochemical signal was recorded using Palmsens Emstat Pico MUX16 board.

[0117] Classification between colorectal cancer patients and healthy volunteers : data was split into a train (80%) and test (20%) and validated with stratified 5-fold cross validation. Grid search was performed using the following analysis algorithm.

[0118] MANOVA (Multivariable Analysis of Variance):

[0119] MANOVA is a statistical technique used to compare the means of two or more groups on multiple dependent variables simultaneously. It is an extension of ANOVA to multiple dependent variables. The general form of the MANOVA equation is based on the multivariate general linear model. The total variance-covariance matrix T can be partitioned into the variance-covariance matrix within groups W and the variance-covariance matrix between groups B, such that:

[0120] T=W+B

[0121] Where:

[0122] • T is the total variance-covariance matrix, representing the overall dispersion of the multivariate data.

[0123] • W is the within-groups variance-covariance matrix, representing the dispersion within each group.

[0124] • B is the between-groups variance-covariance matrix, representing the dispersion between the group means.

[0125] The hypothesis testing in MANOVA is focused on whether the mean vectors of the groups are significantly different, which is assessed by comparing the between-groups variance-covariance B to the within-groups variance-covariance W. Various test statistics can be derived from these matrices, such as Wilks' Lambda, which is defined as:

[0126] Here, |W|and |B|represent the determinants of the W and B matrices, respectively. The smaller the value of Wilks' Lambda, the greater the evidence for significant differences between the group means on the combined dependent variables.

[0127] Model built:

[0128] The sample measurement comprises arrays of current and voltage for each type of electrode coating. We determined the optimal number of k voltages (values ranging from 1 to 215) and the optimal number of n working electrodes (values ranging from 1 to 16, consisting of 8 values for oxidation and 8 for reduction) by performing a multivariable analysis of variance (MANOVA) between two groups (cancer patients versus healthy volunteers), ranking the results by either the highest F-statistics or the lowest p-values. For each selected voltage, we recorded the mean vectors for both the 'cancer group' and the 'healthy group', along with the covariance matrices.

[0129] Test sample classification:

[0130] For each test sample, the Mahalanobis distance to both the 'cancer group' and the 'healthy group' is calculated for each chosen voltage and selected array of coatings. A predictor is then selected for each voltage to minimize the Mahalanobis distance. Subsequently, a majority decision is made based on the outcomes from all predictors across the chosen voltages. where x is the sample observation vector, and p is the mean vector of either the cancer or the healthy group.

[0131] The performance of the model for colorectal cancer detection is tabulated below in Tables 1A and IB:

[0132] Table 1A: Train

[0133] Table IB: Test

[0134] It is seen that plasma samples from stage I colorectal cancer patients of were classified versus healthy volunteers with sensitivity of 100% and specificity of 100%. More data is found in Table 1C below. Example 2 Voltammetry measurements in blood samples of patients diagnosed with lung cancer using an array of working electrodes and signal analysis

[0135] The goal of the study was to differentiate between

[0136] A) the redox fingerprints of patients diagnosed with early stages of lung cancer; and

[0137] B) the redox fingerprints of healthy volunteers

[0138] Blood plasma sample preparation

[0139] 1 ml plasma samples were collected from twenty-six (26) volunteers from a biobank - eleven (11) lung cancer patients and fifteen (15) healthy volunteers.

[0140] The samples were stored at -80°C. Before the use of samples, samples were defrosted at room temperature.

[0141] Electrochemical measurements

[0142] A drop of 1 ml defrosted plasma was placed on the modified commercial SPE. 2-ways differential pulse voltammetry was performed in the plasma samples across the potential range of - 0.IV to +0.7V (Oxidation), and across the potential range of +0.7V to -0.1V (Reduction) with voltage step of 5mV, pulse mode of 50mV, pulse time of O.lsec, and scan rate of 0.02 V / sec. The electrochemical signal was recorded using Palmsens Emstat Pico MUX16 board.

[0143] Classification between lung cancer patients and healthy volunteers : data was split into a train (80%) and test (20%) and validated with stratified 5-fold cross validation. Grid search was performed using the following analysis algorithm:

[0144] MANOVA (Multivariable Analysis of Variance):

[0145] MANOVA is a statistical technique used to compare the means of two or more groups on multiple dependent variables simultaneously. It is an extension of ANOVA to multiple dependent variables. The general form of the MANOVA equation is based on the multivariate general linear model. The total variance-covariance matrix T can be partitioned into the variance-covariance matrix within groups W and the variance-covariance matrix between groups B, such that:

[0146] T=W+B

[0147] Where:

[0148] • T is the total variance-covariance matrix, representing the overall dispersion of the multivariate data.

[0149] • W is the within-groups variance-covariance matrix, representing the dispersion within each group.

[0150] • B is the between-groups variance-covariance matrix, representing the dispersion between the group means.

[0151] The hypothesis testing in MANOVA is focused on whether the mean vectors of the groups are significantly different, which is assessed by comparing the between-groups variance-covariance B to the within-groups variance-covariance W. Various test statistics can be derived from these matrices, such as Wilks' Lambda, which is defined as:

[0152] Here, |W|and |B|represent the determinants of the W and B matrices, respectively. The smaller the value of Wilks' Lambda, the greater the evidence for significant differences between the group means on the combined dependent variables.

[0153] Model built:

[0154] The sample measurement comprises arrays of current and voltage for each type of electrode coating. We determined the optimal number of k voltages (values ranging from 1 to 157) and the optimal number of n electrode coatings (values ranging from 1 to 16, consisting og 8 values for oxidation and 8 for reduction) by performing a multivariable analysis of variance (MANOVA) between two patient groups (cancer patients versus healthy volunteers), ranking the results by either the highest F-statistics or the lowest p-values. For each selected voltage, we recorded the mean vectors for both the 'cancer group' and the 'healthy group', along with the covariance matrices.

[0155] Test sample classification:

[0156] For each test sample, the Mahalanobis distance to both the 'cancer group' and the 'healthy group' is calculated for each chosen voltage and selected array of coatings. A predictor is then selected for each voltage to minimize the Mahalanobis distance. Subsequently, a majority decision is made based on the outcomes from all predictors across the chosen voltages. where x is the sample observation vector, and p is the mean vector of either the cancer or the healthy group.

[0157] The performance of the model for lung cancer detection is tabulated below in Tables 2A and 2B:

[0158] Table 2A: Train

[0159] Table 2B: Test

[0160] It is seen that plasma samples from lung cancer patients were classified versus healthy volunteers with sensitivity of 100% and specificity of 100%. More data is found in Table 2C below. Example 3

[0161] Voltammetry measurements in blood samples of patients diagnosed with colorectal vs. lung cancer using an array of working electrodes and signal analysis

[0162] The goal of the study was to differentiate between

[0163] A) the redox fingerprints of patients diagnosed with early stages of colorectal cancer; and

[0164] B) the redox fingerprints of patients diagnosed with early stages of lung cancer.

[0165] Blood plasma sample preparation

[0166] 1 ml plasma samples were collected from twenty-six (26) volunteers from a biobank - fifteen (15) colorectal cancer (Stage I) and eleven (11) lung cancer patients.

[0167] The samples were stored at -80°C. Before the use of samples, samples were defrosted at room temperature.

[0168] Electrochemical measurements

[0169] 1 ml drop of defrosted plasma was placed on the modified commercial SPE. 2-ways differential pulse voltammetry was performed in the plasma samples across the potential range of -0.1V to +0.7V (Oxidation), and across the potential range of +0.7V to -0.1V (Reduction) with voltage step of 5mV, pulse mode of 50mV, pulse time of O.lsec, and scan rate of 0.02 V / sec. The electrochemical signal was recorded using Palmsens Emstat Pico MUX16 board.

[0170] Classification between colorectal cancer patients and lung cancer patients : data was split into a train (80%) and test (20%) and validated with stratified 5-fold cross validation. Grid search was performed using the following analysis algorithm.

[0171] MANOVA (Multivariable Analysis of Variance):

[0172] MANOVA is a statistical technique used to compare the means of two or more groups on multiple dependent variables simultaneously. It is an extension of ANOVA to multiple dependent variables. The general form of the MANOVA equation is based on the multivariate general linear model. The total variance-covariance matrix T can be partitioned into the variance-covariance matrix within groups W and the variance-covariance matrix between groups B, such that:

[0173] T=W+B

[0174] Where:

[0175] • T is the total variance-covariance matrix, representing the overall dispersion of the multivariate data.

[0176] • W is the within-groups variance-covariance matrix, representing the dispersion within each group.

[0177] • B is the between-groups variance-covariance matrix, representing the dispersion between the group means.

[0178] The hypothesis testing in MANOVA is focused on whether the mean vectors of the groups are significantly different, which is assessed by comparing the between-groups variance-covariance B to the within-groups variance-covariance W. Various test statistics can be derived from these matrices, such as Wilks' Lambda, which is defined as:

[0179] Here, |W|and |B|represent the determinants of the W and B matrices, respectively. The smaller the value of Wilks' Lambda, the greater the evidence for significant differences between the group means on the combined dependent variables.

[0180] Model built:

[0181] The sample measurement comprises arrays of current and voltage for each type of electrode coating. We determined the optimal number of k voltages (values ranging from 1 to 157) and the optimal number of n electrode coatings (values ranging from 1 to 16, consisting of 8 values for oxidation and 8 for reduction) by performing a multivariable analysis of variance (MANOVA) between two patient groups (colorectal cancer patients versus lung cancer patients), ranking the results by either the highest F-statistics or the lowest p-values. For each selected voltage, we recorded the mean vectors for both the 'cancer group A' and the 'cancer group B', along with the covariance matrices.

[0182] Test sample classification:

[0183] For each test sample, the Mahalanobis distance to both the 'cancer group A' and the 'cancer group B' is calculated for each chosen voltage and selected array of coatings. A predictor is then selected for each voltage to minimize the Mahalanobis distance. Subsequently, a majority decision is made based on the outcomes from all predictors across the chosen voltages. where x is the sample observation vector, and p is the mean vector of either the cancer A group or the cancer B group.

[0184] The performance of the model is tabulated below in Tables 3A and 3B:

[0185] Table 3A: Train

[0186] Table 3B: Test

[0187] It is seen that plasma samples from stage I colorectal cancer patients were classified versus lung cancer patients with sensitivity of 100% and specificity of 100%. More data is found in Table 3C below. Example 4

[0188] Voltammetry measurements in urine samples of patients diagnosed with inflammatory bowel disease (IBD) using an array of eight working carbon electrodes and signal analysis

[0189] The goal of the study was to differentiate between the redox fingerprints of inflammatory bowel disease patients in the clinical relapse (flare-up) phase and those in the clinical remission phase using a commercial screen-printed electrode (SPE) featuring eight surface-modified carbon electrodes with a diameter of 2.95 mm. The working electrodes shared a silver reference electrode and a carbon counter electrode. The samples were stored at -80°C. Before use, the samples were thawed at room temperature.

[0190] Urine sample preparation

[0191] A total of thirty-nine (39) defrosted urine samples, each 1 ml in size, were collected. These samples came from seventeen (17) Crohn's disease patients in the clinical relapse (flare-up) phase and twenty-two (22) patients in the clinical remission phase.

[0192] Electrochemical measurements

[0193] 1 ml drop of fresh urine was placed on the modified commercial SPE, as shown in Figure 2. 2-ways differential pulse voltammetry was performed in the urine samples across the potential range of -0.1V to +0.7V (Oxidation), and across the potential range of +0.7V to -0.1V (Reduction) with voltage step of 5mV, pulse mode of 50mV, pulse time of O.lsec, and scan rate of 0.02 V / sec. The electrochemical signal was recorded using Palmsens Emstat Pico MUX16 board.

[0194] Classification between patients in flare-up phase and patients in remitting phase: data was split into a train (85%) and test (15%) and validated with stratified 5-fold cross validation. Grid search was performed using the following analysis algorithm. Model built: The sample measurement comprises arrays of current and voltage for each type of electrode coating. We determined the optimal number of k voltages (values ranging from 1 to 157) and the optimal number of n electrode coatings (values ranging from 1 to 16, consisting of 8 values for oxidation and 8 for reduction) by applying a cost function and ranking the results by the highest scores. For each selected voltage, we recorded both the mean and the variance for the 'flare-up group' as well as the 'remission group'. The selected cost function: where p is the mean, and o is the variance of the relapse or the remission group.

[0195] Test sample classification:

[0196] For each test sample, the Mahalanobis distance to both the 'flare- up group' and the 'remission group' is calculated for each chosen voltage and selected coating. For each voltage, a predictor is selected that minimizes the Mahalanobis distance. Then, a majority decision is made based on the outcomes from all the predictors across the chosen voltages. where x is the sample observation vector, and p is the mean vector of either the relapse or the remission group.

[0197] The performance of the model for inflammatory bowel disease is tabulated below in Table 4A:

[0198] Table 4A: Test

[0199] It is seen that urine samples from patients in relapse (flare-up) phase were classified versus patients in remitting phase with sensitivity of 100% and specificity of 100%. More data is found in Table 4B below. Example 5

[0200] Voltammetry measurements in urine samples of patients diagnosed with bladder cancer using an array of eight working carbon electrodes and signal analysis

[0201] The goal of the study was to distinguish the redox fingerprints of patients whose tumors were pathologically diagnosed with bladder cancer from those whose pathology reports indicated the biopsies were non-malignant or non-cancerous (either showing normal pathology or minor inflammation). This was achieved using a commercial screen-printed electrode (SPE) featuring eight surface- modified carbon electrodes with a diameter of 2.95 mm. The working electrodes shared a silver reference electrode and a carbon counter electrode .

[0202] Urine sample preparation

[0203] A total of sixty-eight (68) fresh (non-fasting) urine samples, each 1-2 ml in size, were collected at one of the leading medical centers in Israel. These samples were obtained from forty-seven (47) patients with tumors pathologically diagnosed as bladder cancer and twenty-one (21) patients whose pathology reports indicated non-malignancy. The control was their hospital pathology biopsy report following a TURBT (Transurethral Resection of Bladder Tumor) procedure.

[0204] Electrochemical measurements

[0205] A drop of 1-2 ml fresh (non-fasting) urine was placed on the modified commercial SPE, as shown in Figure 2. 2-ways differential pulse voltammetry was performed in the urine samples across the potential range of -0.1V to +0.7V (Oxidation), and across the potential range of +0.7V to -0.1V (Reduction) with voltage step of 5mV, pulse mode of 50mV, pulse time of O.lsec, and scan rate of 0.02 V / sec. The electrochemical signal was recorded using Palmsens Emstat Pico MUX16 board. Classification between patients with bladder cancer from patients whose pathology is non-malignant / non-cancerous: data was split into a train (80%) and test (20%) and validated with stratified 5- fold cross validation. Grid search was performed using the following analysis algorithm:

[0206] MANOVA (Multivariable Analysis of Variance):

[0207] MANOVA is a statistical technique used to compare the means of two or more groups on multiple dependent variables simultaneously. It is an extension of ANOVA to multiple dependent variables. The general form of the MANOVA equation is based on the multivariate general linear model. The total variance-covariance matrix T can be partitioned into the variance-covariance matrix within groups W and the variance-covariance matrix between groups B, such that:

[0208] T=W+B

[0209] • T is the total variance-covariance matrix, representing the overall dispersion of the multivariate data.

[0210] • W is the within-groups variance-covariance matrix, representing the dispersion within each group.

[0211] • B is the between-groups variance-covariance matrix, representing the dispersion between the group means.

[0212] The hypothesis testing in MANOVA is focused on whether the mean vectors of the groups are significantly different, which is assessed by comparing the between-groups variance-covariance B to the within-groups variance-covariance W. Various test statistics can be derived from these matrices, such as Wilks' Lambda, which is defined as:

[0213] Here, |W|and |B|represent the determinants of the W and B matrices, respectively. The smaller the value of Wilks' Lambda, the greater the evidence for significant differences between the group means on the combined dependent variables. Model built:

[0214] The sample measurement comprises arrays of current and voltage for each type of electrode coating. We determined the optimal number of k voltages (values ranging from 1 to 157) and the optimal number of n electrode coatings (values ranging from 1 to 16, including 8 values for oxidation and 8 for reduction) by performing a multivariable analysis of variance (MANOVA) between two patient groups (cancer versus non-cancer), ranking the results by either the highest F-statistics or the lowest p-values. For each selected voltage, we recorded the mean vectors for both the 'cancer group' and the 'non-cancer group', along with the covariance matrices.

[0215] Test sample classification:

[0216] For each test sample, the Mahalanobis distance to both the 'cancer group' and the 'non-cancer group' is calculated for each chosen voltage and selected array of coatings. A predictor is then selected for each voltage to minimize the Mahalanobis distance. Subsequently, a majority decision is made based on the outcomes from all predictors across the chosen voltages. where x is the sample observation vector, and p is the mean vector of either the cancer or the non-cancer group.

[0217] The performance of the model for bladder cancer detection is tabulated below in Table 5A:

[0218] Table 5A: Test

[0219] It is seen that urine samples from bladder cancer patients were classified versus non-cancer with sensitivity of 90% and specificity of 100%. More data is found in Table 5B below. Example 6 Voltammetry measurements in blood samples of patients diagnosed with colorectal cancer using an array of working electrodes and signal analysis

[0220] A) Preclinical Data & Sample Annotation:

[0221] 81 biobank plasma samples were tested, consisting of 42 early- stage CRC and 39 healthy controls. Samples were divided as follows (samples were stored at -80°C prior to testing):

[0222] B) Predictive Model Development

[0223] The raw data collected from the sensor of Preparation 7 consisted of 2,512 data points (i.e., voltage values) per measurement, structured as follows:

[0224] - Number of sensing elements: the e-voltammetric tongue includes eight different electrodes.

[0225] - Number of measurement modes: the system records both oxidation and reduction current responses, by cyclic differential pulse voltammetry as described above.

[0226] - Number of potential points at which current is measured: for each electrode and mode of measurement, current is measured across 157 discrete voltage points, capturing fine-grained electrochemical behavior.

[0227] This results in a high-resolution electrochemical fingerprint: 8 electrodes x 2 modes x 157 voltage levels = 2,512 total data points. This rich dataset forms the basis for machine learningbased classification and feature selection, enabling sensitive and specific detection of colorectal cancer biomarkers.

[0228] The prediction model was developed using 80% of the dataset (64 samples) for training, with 5-fold cross-validation applied. The remaining 20% (17 samples) were reserved for blind testing to assess model performance. A logistic regression algorithm was used to build the classification model.

[0229] C) Feature Reduction Using Ll-regularized Regression

[0230] The graph shown in Figure 8 illustrates the use of Ll-regularized logistic regression (also known as Lasso regression) for feature selection in a classification model. Logistic regression is a supervised learning algorithm used to predict the probability of a binary outcome—such as disease presence or absence—based on input features.

[0231] In this case, LI regularization adds a penalty to the regression coefficients, pushing many of them to zero as the regularization strength increases. This not only helps prevent overfitting, but also serves as a tool for feature selection by retaining only the most informative variables. At the chosen regularization point, the model retained 44 features, achieving a high mean AUROC (~0.82) while significantly reducing model complexity.

[0232] D) Model Evaluation - Training Set

[0233] 100-fold validation was performed on the training set. The mean Area Under the Curve (AUG) score on the training set (100-fold validation) was 0.80. The median AUG was 0.81.

[0234] E) Sensitivity & Specificity

[0235] With the aid of the model, the electrochemical sensor of the invention achieved 89% sensitivity, 100% specificity and 94% accuracy (71.3% - 99.85% @ 95% CI) for early-stage CRC. The confusion matrix is tabulated below:

[0236] Receiver Operating Characteristic (ROC) of the Prediction Model Classifier: The Area Under the Curve (AUC) score was calculated as 0.96. The prediction probability of the model classifier is shown in Figure 9, indicating a clear separation between the CRC and control groups.

[0237] Table 1C Table 2C

[0238] Table 3C

[0239] Table 3C: CRC / LC Parameters

[0240] Table 4B

[0241] Table 4B: Flare / Remission Parameters

[0242] Table 5B: Bladder Cancer / Healthy Parameters

Claims

Claims1) A method of diagnosing a person suspected of having cancer detectable in a biofluid, comprising: obtaining biofluid sample from the patient; acquiring an electrochemical signal generated jointly by redox molecules in the biofluid sample, wherein the signal is acquired with the aid of an electronic tongue comprising at least two sets of working electrodes selected from: a set consisting of one or more bare electrodes; a set consisting of one or more electrodes coated with polysaccharide film, optionally with conductive additives incorporated into the film; a set consisting of one or more electrodes coated with reduced graphene oxide film; a set consisting of one or more electrodes coated with platinum black film; a set consisting of one or more electrodes coated with MoS2Afilm; a set consisting of one or more electrodes coated with MoS2Bfilm; a set consisting of one or more electrodes coated with WS2Afilm; a set consisting of one or more electrodes coated with WS2Bfilm; wherein the superscripts A and B indicate electrodeposited M0S2 and WS2 coatings produced by cycling across potential ranges A and B, respectively, wherein potential range A corresponds to the double layer potential region of the working electrode and potential range B extends to more positive potentials than potential range A; optionally preprocessing the electrochemical signal, to obtain processed data; applying trained chemometric model(s) to the processed or raw data; and diagnosing a type of cancer after the trained chemometric model has classified the processed / raw data.2) A method according to claim 1, for detecting colorectal cancer or lung cancer, wherein the biofluid sample is a blood, serum or p1asma samp1e.3) A method according to claim 1 or 2, wherein the electrochemical signal is acquired by cyclic differential pulse voltammetry.4) A method according to claim 3, wherein the model applied was trained based on cyclic differential pulse voltammetry measurements in biofluid samples from cancer and non-cancer patients, to select k potential (E) values, where k is an odd number, and a group of n working electrodes for each of said k potentials, wherein said n working electrodes were selected based on currents measured as the potential was swept positively forward from Ei to Ef and negatively in reverse direction, or vice versa, to create a first distribution, characteristic of biofluid samples from the cancer patients, and a second, well-distinguished distribution, characteristic of biofluid samples from the noncancer patients.5) A method according to claim 4, wherein the classification of the test biofluid sample comprises: calculating, for each of the k selected potentials, the distances between the current vector measured in the test sample by cyclic differential pulse voltammetry and the first and second distributions, given by Mahalanobis distances:where x is the sample current vector, and p is the mean vector of either the cancer or the non-cancer group;determining, for a pair of Mahalanobis distances MDdisease and MDnon- disease calculated for a given potential, which of the two distance is smaller; diagnosing cancer, if in the majority of the pairs, MDdisease < MDnon- disease, and vice versa.6) A method according to claim 2 or 3, for diagnosing colorectal cancer, wherein the trained model is a logistic regression model.7) A method according to claim 6, wherein the model is Ll- regularized logistic regression model.8) A method according to any one of claims 1 to 5, for detecting colorectal cancer or lung cancer, which is a multi-cancer early detection method.9) A method of managing treatment of IBD patients by determining clinical relapse and remission phases, comprising: obtaining urine sample from the patient; acquiring an electrochemical signal generated jointly by redox molecules in the urine sample, wherein the signal is acquired with the aid of an electronic tongue comprising at least two sets of working electrodes selected from: a set consisting of one or more bare electrodes; a set consisting of one or more electrodes coated with polysaccharide film, optionally with conductive additives incorporated into the film; a set consisting of one or more electrodes coated with reduced graphene oxide film; a set consisting of one or more electrodes coated with platinum black film; a set consisting of one or more electrodes coated with MoS2Afilm; a set consisting of one or more electrodes coated with MoS2Bfilm;a set consisting of one or more electrodes coated with WS2Afilm; a set consisting of one or more electrodes coated with WS2Bfilm; wherein the superscripts A and B indicate electrodeposited M0S2 and WS2 coatings produced by cycling across potential ranges A and B, respectively, wherein potential range A corresponds to the double layer potential region of the working electrode and potential range B extends to more positive potentials than potential range A; optionally preprocessing the electrochemical signal, to obtain processed data; applying trained chemometric model(s) to the processed or raw data; and determining whether the IBD patient is in a clinical relapse phase or in a clinical remission phase, after the trained chemometric model has classified the processed / raw data.10) A method according to claim 9, wherein the electrochemical signal is acquired by cyclic differential pulse voltammetry.11) A method according to claim 9, wherein the model applied was trained based on cyclic differential pulse voltammetry measurements in urine samples from IBD patients in relapse and remission phases, to select k potential (E) values, where k is an odd number, and a group of n working electrodes for each of said k potentials, wherein said n working electrodes were selected based on currents measured as the potential was swept positively forward from Ei to Ef and negatively in reverse direction, or vice versa, to create a first distribution, characteristic of biofluid samples from IBD patients in a clinical relapse phase, and a second, well-distinguished distribution, characteristic of biofluid samples from IBD patients in a clinical remission phase.12) A method according to claim 11, wherein the classification of the test biofluid sample comprises:calculating, for each of the k selected potentials, the distance between the currents vector measured in the test sample by cyclic differential pulse voltammetry and the first and second distributions, given by Mahalanobis distances:where x is the sample currents vector, and p is the mean vector of either the relapse or remission group; determining, for a pair of Mahalanobis distances MDrelapse and MDremission calculated for a given potential, which of the two distances is smaller; diagnosing clinical relapse phase, if in the majority of the pairs, MDrelapse < MDremission, and vice Versa.13) A method of diagnosing a person suspected of having a bladder cancer, comprising: obtaining urine sample from the patient; acquiring an electrochemical signal generated jointly by redox molecules in the urine sample, wherein the signal is acquired with the aid of an electronic tongue comprising at least two sets of working electrodes selected from: a set consisting of one or more bare electrodes; a set consisting of one or more electrodes coated with polysaccharide film, optionally with conductive additives incorporated into the film; a set consisting of one or more electrodes coated with reduced graphene oxide film; a set consisting of one or more electrodes coated with platinum black film; a set consisting of one or more electrodes coated with MoS2Afilm; a set consisting of one or more electrodes coated with MoS2Bfilm;a set consisting of one or more electrodes coated with WS2Afilm; a set consisting of one or more electrodes coated with WS2Bfilm; wherein the superscripts A and B indicate electrodeposited M0S2 and WS2 coatings produced by cycling across potential ranges A and B, respectively, wherein potential range A corresponds to the double layer potential region of the working electrode and potential range B extends to more positive potentials than potential range A; optionally preprocessing the electrochemical signal, to obtain processed data; applying trained chemometric model(s) to the processed or raw data; and diagnosing bladder cancer after the trained chemometric model has classified the processed / raw data, wherein: the electrochemical signal is acquired by cyclic differential pulse voltammetry, and the model applied was trained based on cyclic differential pulse voltammetry measurements in biofluid samples from bladder cancer and non-cancer patients, to select k potential (E) values, where k is an odd number, and a group of n working electrodes for each of said k potentials, wherein said n working electrodes were selected based on currents measured as the potential was swept positively forward from Ei to Ef and negatively in reverse direction, or vice versa, to create a first distribution, characteristic of biofluid samples from the bladder cancer patients, and a second, well-distinguished distribution, characteristic of biofluid samples from the non-cancer patients, and wherein the classification of the test biofluid sample comprises: calculating, for each of the k selected potentials, the distance between the currents vector measured in the test sample by cyclic differential pulse voltammetry and the first and second distributions, given by Mahalanobis distances:where x is the sample currents vector, and p is the mean vector of either the cancer or the non-cancer group; determining, for a pair of Mahalanobis distances MDdisease and MDnon- disease calculated for a given potential, which of the two distance is smaller; diagnosing bladder cancer, if in the majority of the pairs, MDdisease < MDnon-disease, and vice versa.14) An electrochemical diagnostic and monitoring system, comprising : a voltammetric electronic tongue which comprises: a set consisting of one or more bare electrodes; a set consisting of one or more electrodes coated with polysaccharide film, optionally with conductive additives incorporated into the film; a set consisting of one or more electrodes coated with reduced graphene oxide film; a set consisting of one or more electrodes coated with platinum black film; a set consisting of one or more electrodes coated with MoS2Afilm; a set consisting of one or more electrodes coated with MoS2Bfilm; a set consisting of one or more electrodes coated with WS2Afilm; a set consisting of one or more electrodes coated with WS2Bfilm; wherein the superscripts A and B indicate electrodeposited M0S2 and WS2 coatings produced by cycling across potential ranges A and B, respectively, wherein potential range A corresponds to the double layer potential region of the working electrode andpotential range B extends to more positive potentials than potential range A; a counter electrode and optionally a reference electrode; a potentiostat to which the working electrodes, the counter electrode and optionally the reference electrodes are electrically connected to allow control of the potential and produce voltammograms when the electrodes are immersed in a test sample; a processor programmed to analyze and classify the voltammograms of the test sample by a chemometric model defined in previous claims.15) An electrochemical diagnostic system according to claim 14, wherein the voltammetric electronic tongue comprises working electrodes placed in an annular space between a ring-shaped counter electrode and a centrically positioned reference electrode.

Images