DEVICE AND METHOD FOR RAPID DETECTION OF VIRUSES

Abstract

The invention proposes an approach utilizing novel and artificially intelligent hybrid sensor arrays with multiplexed detection capabilities for disease-specific biomarkers from the exhaled breath of a subject. The technology provides a rapid and highly accurate diagnosis in various COVID-19 infection and transmission scenarios.

Claims

1-48. (canceled)

49. A method of identifying the presence of a viral infection in a subject, the method comprising: a) exposing a breath sample obtained from the subject to a sensor surface comprising a plurality of nanoparticles surface-associated with a ligand selected from dodecanethiol, hexanethiol, decanethiol, tert-dodecanethiol, butanethiol, 2-ethylhexanethiol, dibutyl disulfide, 2-nitro-4-trifluoromethylbenzenethiol, benzylmercaptane, 4-chlorobenzenemethanethiol, 3-ethoxythiolphenol, 4-tert-methylbenzenethiol and 1-heptanethiol, b) determining a volatile organic compound (VOC) profile indicative of presence one of more VOCs in the breath sample from the subject; and b) comparing the VOC profile of the subject to a VOC profile of a control and/or to a VOC profile obtained from the subject at an earlier time point(s); to thereby determine one or more of (1) presence of a viral infection, (2) absence of a viral infection, (3) reoccurrence of the viral infection, (4) the type of viral infection, or (5) viral load or stage of the viral infection.

50. The method according to claim 49, wherein the sensor surface comprises one or more sensing regions, each of the sensing regions being associated with same or different independently measurable population of nanoparticles.

51. The method according to claim 50, wherein each of the sensing regions comprises a plurality of nanoparticle populations, wherein each of the plurality of population differs from another in at least one of nanoparticle size, nanoparticle morphology, nanoparticle composition, and surface decoration.

52. The method according to claim 49, wherein the nanoparticles are metallic nanoparticles.

53. The method according to claim 52, wherein the metallic nanoparticles comprise or consist a metal selected from gold, silver, nickel, cobalt, copper, palladium, platinum and aluminum or alloy or metal combination thereof.

54. The method according to claim 53, wherein the nanoparticles are gold nanoparticles.

55. The method according to claim 49, wherein the ligand molecules comprise one or more of dodecanethiol, 2-ethylhexanethiol, 4-tert-methylbenzenethiol, decanethiol, 4-chlorobenzenemethanethiol, 3-ethoxytiophenol, tert-dodecanethiol and hexanethiol.

56. The method according to claim 49, wherein the ligand molecules comprise one or more of 4-tert-methylbenzenethiol, tert-dodecanethiol and hexanethiol.

57. The method according to claim 49, wherein the ligand molecules comprise 4-tert-methylbenzenethiol and/or tert-dodecanethiol and/or hexanethiol.

58. The method according to claim 49, wherein the ligand molecules comprise tert-dodecanethiol.

59. The method according to claim 49, wherein the sensor surface comprises a plurality or one or more sensing regions, each of the sensing regions is in the form of a plurality of gold nanoparticles, each of the nanoparticles being surface-associated with ligand molecules selected from dodecanethiol, hexanethiol, decanethiol, tert-dodecanethiol, butanethiol, 2-ethylhexanethiol, dibutyl disulfide, 2-nitro-4-trifluoromethylbenzenethiol, benzylmercaptane, 4-chlorobenzenemethanethiol, 3-ethoxythiolphenol, 4-tert-methylbenzenethiol and 1-heptanethiol.

60. The method according to claim 59, wherein the gold nanoparticles are surface associated with ligand molecules selected from dodecanethiol, 2-ethylhexanethiol, 4-tert-methylbenzenethiol, decanethiol, 4-chlorobenzenemethanethiol, 3-ethoxytiophenol, tert-dodecanethiol and hexanethiol.

61. The method according to claim 49, wherein the sensor surface comprises at least three sensing regions, each of the sensing regions comprising a different population of nanoparticles, each population differing from another in the surface associated ligand molecules.

62. The method according to claim 61, wherein the sensor surface comprises at least 8 sensing regions, wherein gold nanoparticles at each of the at least 8 regions is associated to different ligands selected from dodecanethiol, 2-ethylhexanethiol, 4-tert-methylbenzenethiol, decanethiol, 4-chlorobenzenemethanethiol, 3-ethoxytiophenol, tert-dodecanethiol and hexanethiol.

63. The method according to claim 61, wherein the sensor surface comprises at least 13 sensing regions, wherein gold nanoparticles at each of these at least 13 regions are associated to different ligands selected from dodecanethiol, hexanethiol, decanethiol, tert-dodecanethiol, butanethiol, 2-ethylhexanethiol, dibutyl disulfide, 2-nitro-4-trifluoromethylbenzenethiol, benzylmercaptane, 4-chlorobenzenemethanethiol, 3-ethpxythiolphenol, 4-tert-methylbenzenethiol and 1-heptanethiol.

64. The method according to claim 49, wherein the breath sample is obtained from a subject by direct exhalation of breath into a device configured and operable for carrying out the method, or into to a disposable collecting tube optionally in the form of a soft tube, or by suction.

65. A device, optionally in a form of a handheld device, for diagnosing, screening or monitoring a viral infection in an asymptomatic subject, the device comprising: a sensor surface comprising one or more sensing regions, each of the sensing regions comprising ligand-associated nanoparticles, configured and operable for interacting with one or more VOCs present in the subject's exhaled breath; and a processing unit comprising a learning and pattern recognition analyzer configured for receiving output signals from the one or more of the sensing regions and comparing the signals to a stored data, by utilizing a pattern recognition algorithm; wherein the ligands associated to the nanoparticles are selected from dodecanethiol, hexanethiol, decanethiol, tert-dodecanethiol, butanethiol, 2-ethylhexanethiol, dibutyl disulfide, 2-nitro-4-trifluoromethylbenzenethiol, benzylmercaptane, 4-chlorobenzenemethanethiol, 3-ethoxythiolphenol, 4-tert-methylbenzenethiol and 1-heptanethiol.

66. The device according to claim 65, wherein the sensor surface comprises a plurality or one or more sensing regions, each of the sensing regions is in the form of a plurality of gold nanoparticles, each of the nanoparticles being surface-associated with ligand molecules selected from dodecanethiol, hexanethiol, decanethiol, tert-dodecanethiol, butanethiol, 2-ethylhexanethiol, dibutyl disulfide, 2-nitro-4-trifluoromethylbenzenethiol, benzylmercaptane, 4-chlorobenzenemethanethiol, 3-ethoxythiolphenol, 4-tert-methylbenzenethiol and 1-heptanethiol.

67. The device according to claim 65, wherein the breath sample is obtained from a subject by direct exhalation of breath into a device configured and operable for carrying out the method, or into to a disposable collecting tube optionally in the form of a soft tube, or by suction.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

[0085] In order to better understand the subject matter that is disclosed herein and to exemplify how it may be carried out in practice, embodiments will now be described, by way of non-limiting example only, with reference to the accompanying drawings, in which:

[0086] FIG. 1 depicts a patient enrolment and observational design.

[0087] FIG. 2 shows a representative response of a sensor according to the invention to three different breath samples.

[0088] FIG. 3 is an example of breath collection with a novel handheld breathalyzer system constructed according to the invention, from a patient in Wuhan, China.

[0089] FIGS. 4A-D provide diagnosis of COVID-19 patients based on cumulative breath sample response according to the invention, as explained herein.

DETAILED DESCRIPTION OF EMBODIMENTS

[0090] The outbreak of severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2, aka COVID-19) has emerged very rapidly and has invaded more than 197 countries worldwide. About 97.5% of patients develop symptoms of COVID-19 within 11.5 days of exposure, causing late diagnosis and a high infection rate. The molecular tests used so far to confirm COVID-19 are accurate and considered the gold standards for SARS-CoV-2 testing. Nevertheless, they require a swab sample and a time-consuming laboratory procedure. Shipping of samples and overload of laboratory facilities entail a delay of many days until the test results are available, increasing the burden on the healthcare system. Furthermore, the tests are highly sensitive only for those who already have symptoms because of the high virus load; however, it is already known that the disease can be spread by asymptomatic carriers who show only mild symptoms or even none at all. For example, RT-PCR analysis takes two days with approximately 70% sensitivity and a high false-negative rate, implying many type 2 errors, causing sick people to be misdiagnosed, resulting in further spread of the disease. Epidemiological data based on the sequenced viral RNA show that the spread of COVID-19 has resulted from local community transmission, which means that the source of infection cannot be traced back to a known exposure. Thus, healthcare systems worldwide require tests that are non-invasive, rapid, inexpensive, and easy-to-use for diagnosing or ruling out infection at earlier stages, even before COVID-19 symptoms manifest, to decrease the transmission and mortality rates.

[0091] The device and methods of the invention provide a rapid non-invasive approach that could potentially serve as an epidemic control tool. This will allow a quick response to threats and streamlining of the necessary resources. The approach relies on novel and artificially intelligent hybrid sensor arrays with multiplexed detection capabilities for a COVID-19-specific breath-print pattern in exhaled breath. To test the feasibility of the exhaled breath approach, as a pre-screening diagnostic system, a case-control study was conducted at the origin of the COVID-19 outbreak, Wuhan, China, during March 2020.

[0092] A sensor utilized according to the invention is characterized by a surface comprising a plurality of (or one or more) sensing regions, each of the sensing regions is in the form of (or comprises or consists) a plurality of nanoparticles, each of the nanoparticles being surface-associated with ligand molecules. Ligand molecules used in accordance with the invention included one or more of the ligands: dodecanethiol, hexanethiol, decanethiol, tert-dodecanethiol, butanethiol, 2-ethylhexanethiol, dibutyl disulfide, 2-nitro-4-trifluoromethylbenzenethiol, benzylmercaptane, 4-chlorobenzenemethanethiol, 3-ethoxythiolphenol, 4-tert-methylbenzenethiol and 1-heptanethiol.

[0093] In some cases, the nanoparticles were gold nanoparticles.

[0094] For some measurements, the nanoparticles were surface associated with ligand molecules selected from dodecanethiol, 2-ethylhexanethiol, 4-tert-methylbenzenethiol, decanethiol, 4-chlorobenzenemethanethiol, 3-ethoxytiophenol, tert-dodecanethiol and hexanethiol. For other measurements, the gold nanoparticles were surface associated with ligand molecules selected from 4-tert-methylbenzenethiol, tert-dodecanethiol and hexanethiol, or those selected from 4-tert-methylbenzenethiol and/or tert-dodecanethiol and/or hexanethiol.

[0095] In some cases, the nanoparticles were surface associated with tert-dodecanethiol.

[0096] In other cases, the nanoparticles were surface associated with ligand molecules selected from butanethiol, dibutyl disulfide, hexanethiol, 1-heptanethiol, tert-dodecanethiol, 2-ethylhexanethiol, 4-tert methylbenzenethiol, 3-ethpxythiolphenol and 4-chlorobenzenemethanethiol.

[0097] In some other cases, the nanoparticles were surface associated with ligand molecules selected from tert-dodecanethiol, hexanethiol, 2-ethylhexanethiol, 4-tert methylbenzenethiol, 3-ethpxythiolphenol, 4-chlorobenzenemethanethiol, dodecanethiol, and decanethiol, or selected from 2-nitro-4-trifluoro-methylbenzenethiol and benzylmercaptan.

[0098] Methods

[0099] Study Population #1

[0100] A total of 140 participants were enrolled at multiple centers in Wuhan and Hefei, China, as part of an observational study: 49 COVID-19 patients, 58 healthy controls and 33 non-COVID lung infection controls (FIG. 1). The COVID-19 patients were confirmed by computed tomography (CT), nasal and pharyngeal swab specimens for real-time reverse-transcriptase polymerase-chain-reaction (RT-PCR), and antibody tests. The enrolled COVID-19 patients were sampled at two time points approximately 3-5 days apart in the COVID-19 intensive care ward assigned to the First Affiliated Hospital of USTC entrusted by the State Council of China located in Wuhan Union Hospital, China. The research protocol was approved by the ethics committee of Anhui Provincial Cancer Hospital, West District of The First Affiliated Hospital of USTC. All participants provided written informed consent.

[0101] Breath Analyzer

[0102] Portable hand-held and notebook computer-connected artificially intelligent hybrid sensor arrays with multiplexed detection capabilities were used in the current project. The principle of breath analysis in these devices is the change in resistance of the sensors when they come into contact with a particular mixture of VOCs. This allows the device to be trained to recognise a particular disease, COVID-19 in our case. The sensor array used in the device contained cross-reactive, chemically diverse chemiresistors based on organically stabilized spherical gold nanoparticles (GNPs) developed by TECHNION (Haifa, Israel). The measurement protocol comprised two steps: baseline reading and reading from the real breath sample. Therefore, two sets of data from each of the eight sensors in the array were obtained from each study subject using the two devices.

[0103] Breath Sampling

[0104] Breath samples were collected by the study subjects breathing directly into the aperture of the instrument for at least four seconds, keeping the instrument approximately 1-2 cm from the mouth. Built-in sensor technology advised the study subject when the test was complete. If the breath collection was not satisfactory, the subject was asked to repeat the test.

[0105] Feature Extraction and Statistical Analysis

[0106] Features were extracted from the output files of the breath samples for all sensors in the array. An example of a sensor response to different samples can be found in FIG. 2. The feature was calculated as the change in electrical resistance between the breath signal and the baseline signal divided by the baseline signal. The tested groups were subjected to binary comparison and the data were divided randomly into training sets (70% samples) and test sets (30% samples). The results of discriminant function analysis (DFA) on the training set were validated using the test set. The base analysis between the COVID-19 and control samples used the quadratic DFA model based on the measurements of three sensors. For the other two sub-comparisons (COVID-19 vs other lung infections; COVID-19 first vs second sampling), the same three sensors as for main model were used by applying two binary linear DFA models. The model performance of the training set was first determined by measuring the area under the curve (AUC) of the receiver operating characteristic (ROC) and was used to calculate the cut-off values on the basis of Youden's index, which classifies the tested groups as giving either a positive or a negative result for the test set classification. Subsequently, other parameters of model performance were analyzed including accuracy, sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV). To address the influence of the main confounding factors on breath analysis, age, gender and smoking status were registered and evaluated on the basis of the same model classifier achieved by DFA. The first and second samplings from the same COVID-19 patients (i.e., correlated variables) were compared using paired t-test analysis (by subject) only between patients who had provided samples at the two time points. The matched platform results with a Tukey mean-difference presented the mean difference and 95% confidence interval (CI). Significant differences in the one-way and/or matched pair test were considered at a cut-off p-value of 0.05 between the sub-groups checked, as determined from the results using JMP Pro, version 14.0.0 (SAS Institute Inc., Cary, N.C., USA, 1989-2005). Of the 140 samples, 10 were excluded after collection for technical reasons, either bad sampling or sensor failure during sampling and before statistical analysis.

[0107] In FIG. 2, the normalized response of sensor 7 of the breathalyzer system to three different samples: Patient A, COIVD-19 first sample while sick; Patient A, second sample after determined as cured; and healthy control. The X-axis represents cycle measurement; each unit is one cycle of the sensor. The sick sample had a positive change response while the cured and control showed negative charges.

[0108] Results

[0109] Participants

[0110] The selection of participants from three distinct groups—COVID-19 patients, healthy controls, and non-COVID lung infection controls—is described in FIG. 1. The characteristics of the 140 participants are shown in Table 1. Among the patients, more than 60% had no underlying chronic disease while all the rest suffered at least from hypertension; 10% of those suffered from diabetes mellites. Most patients were nonsmokers (73%), with a mean age of 59 years with 57% females. For the control group, 67% were nonsmokers, and the mean age was 52 years with 46% females. For the lung infection control group, 73% were nonsmokers, and the mean age was 63 years with 44% females.

TABLE-US-00001 TABLE 1 Demographics, baseline characteristics and laboratory testing of study population. Patients positive for COVID-19 were identified prior to breath sampling. Only the COVID-19 group was accessed for molecular and immunological characteristics. IQR-interquartile range, CT-computed tomography, RT-PCR-real-time reverse transcriptase polymerase chain reaction. Lung COVID-19 Control infections group group group Characteristics (N = 49) (N = 58) (N = 33) Median age (IQR)-yr  58    53 (45-59) 64 (53-73) (51.5-64.5) Female sex-no. (%) 28 (57) 26 (45) 15 (45) Active Smoking-no. (%) 13 (27) 19 (33)  9 (27) Coexisting conditions-no. (%) Hypertension 18 (37)  8 (12)  8 (24) Diabetes  5 (10)  2 (3)   4 (12) Heart disease  1 (2)  —  2 (6)  Malignancy — —  8 (24) Bacterial\fungal infection — —  6 (18) Other  1 (2)  — — Means of positive determination COVID-19-no. (%) CT angiogram 33 (67) RT-PCR, positive (throat swab)-no. (%) nCovORFlab 29 (59) 2019nCOV-N 28 (57) 2019nCov antibody tests, median (IQR)-AU/ml 2019nCov IgM  42.85 (13.65-69.24) 2019nCov IgG 164.68 (133.93-182.55) Count of irregularities in immune system component levels-no. (%) CD3 (58.17-84.22%) Count of high  7 (14) Count of low  3 (6)  CD4 (25.34-51.37%) Count of high 13 (27) Count of low  1 (2)  CD8 (14.23-38.95%) Count of high  4 (8)  Count of low  4 (8)  CD4/CD8 (0.41-2.72) Count of high 12 (25) Count of low —

[0111] Breath Analysis

[0112] Breath samples were collected and analyzed using a PoC handheld system (FIG. 3). Statistical analysis involved three binary comparisons: COVID-19 vs. control; COVID-19 vs. other lung infections; and COVID-19 1.sup.st vs. COVID-19 2.sup.nd sample. Seventy percent of the data were used to calculate the DFA models; a ROC analysis was done and the cut-off was determined. The remaining test data were classified on the basis of the cut-off as presented (FIG. 4). Accuracies for the training set varied between 90% and 94% for the three models, and between 76% and 95% for the test set (Table 2). The main comparison between COVID-19 and controls gave 100% sensitivity for both training and test groups with a miss rate of zero, while the false-positive rate was 39%. The paired analysis of the two sample times showed a clear distinction for the measured breath composition score (P<0.001, mean difference (95% CI)=−1.57 (−2.27−−0.87)): in the first sample, all were COVID-19 positive; at the second sample time, all but three were considered cured. Among the three uncured, two were identified by the model as false-negative and one was correctly classified as positive. Confounding effects on the main classification were examined: there were no significant difference in respect of age, gender, smoking status or coexisting conditions. In addition, we tested the plausibility of the main model built to distinguish COVID-19 from controls to classify all subgroups. The model showed significant differences between COVID-19 and control, COVID-19 and cured COVID-19 samples, lung infection and control, and lung infection and cured COVID-19 samples.

[0113] FIGS. 4A, B and C show data classification from cumulative sensor responses to breath samples as represented by the canonical variable of the discriminant analysis. Box plots of the first canonical score of the training set (70% of samples) and test set (30% of samples). The horizontal dashed line in the box plots represents the cut-off value of the model. True positive (TP), True negative (TN), False positive (FP), False negative (FN). FIG. 4A: COVID-19 patients (41 participants) and healthy controls (57 participants). FIG. 4B: COVID-19 patients (41 participants) and other lung infection\condition controls (32 participants). FIG. 4C: COVID-19 patients at first sampling (41 participants) and second sampling (21 participants); and COVID-19 patients at second sampling uncured (three participants), and controls who had never smoked and were not identified as having chronic bronchitis by either questionnaire (58 participants). P values are for the comparisons of the training set for each of two binary classifications. The horizontal line in the boxes represents the median, the cross represents the mean, and the bottom and top of the boxes represent the 25th and 75th percentiles, respectively. I bars represent the upper 90 percentile and the lower 10 percentile, and the square-dots outliers. All P values were adjusted for multiple comparisons using the Tukey-Kramer method. For FIG. 4C, the P value is also adjusted for paired analysis. FIG. 4D shows receiver-operating-characteristic (ROC) curves for the cumulative breath-sensor response in patients with defined COVID-19 (Co) infection compared with controls (black curve; area under the curve [AUC], 0.81 [0.70 to 0.89]), in COVID-19 infection compared with other lung infection\conditions (red curve; AUC, 0.97 [0.92 to 0.99]), and in COVID-19 infection first sample compared to COVID-19 infection second sample (AUC, 0.87 [0.67 to 1.00]).

TABLE-US-00002 TABLE 2 Breath test outcomes for the study population. Training set Testing set.sup.‡ Covid Covid vs. Covid1.sup.st Covid Covid vs. Covid 1.sup.st vs. Lung vs. vs. Lung vs. Statistics Control.sup.† Infection.sup.§ Covid 2.sup.nd§ Control Infection Covid 2.sup.nd Accuracy (%) 94 90 90 76 95 88 Sensitivity (%) 100 90 100 100 100 83 Specificity (%) 90 91 69 61 90 100 PPV (%) 88 93 86 61 92 100 NPV (%) 100 87 100 100 100 71 TP (cases) 30 26 32 11 12 10 TN (cases) 35 20 11 11 9 5 FP (cases) 4 2 5 7 1 0 FN (cases) 0 3 0 0 0 2 .sup.†Classification based on QDA, .sup.§Classification based on LDA, .sup.‡Classification based on the ROC cut-off.

DISCUSSION

[0114] The results show excellent sensitivities for the three binary comparisons with a minimal false-negative rate, while specificities were above average. The assumption that COVID-19 patients can be differentiated not only from controls but also from other lung conditions/infections was tested. Indeed, the results showed 95% accuracy in classifying the latter conditions. A follow-up of the same patients at a second-time point, when most were already cured, showed 90% accuracy, though two of the three uncured patients were wrongly determined as cured. This could be attributed to the prolonged healing time; it can take a few weeks to reach a definitely cured state. The results are comparable to current published data from COVID-19 studies, which suggest 82%-98% accuracy for abnormal CT findings and 51%-70% sensitivity of the RT-PCR test. The breath test could be affected by such confounding factors as gender, age, smoking status, and coexisting conditions, so it was important to test their influence on results; but no significant influence was found, similar to previous reports on this technology.

[0115] It is expected that real-time methods such as the exhaled breath approach reported here will significantly reduce unnecessary exposure to contagious persons and support the fight against the COVID-19 pandemic. Moreover, it will reduce the number of unnecessary confirmatory tests and lower the burden on hospitals, while providing individuals with a screening solution that can be performed at home or in PoC facilities. During hospitalization or home isolation, a combination of a sensing patch and a breath analyzer will serve to monitor treatment success and disease regression.

[0116] Study Population #2

[0117] The technology is based on a sensor array that measures Volatile Organic Compounds (VOCs) emitted in the exhaled breath, which are biomarkers for a variety of diseases. Artificial Intelligence analyzes the signal pattern to reveal unique “VOC print” for each disease. This test does not require any user special skills and therefore can be accessible at all point of care, and even for home tests.

[0118] The aim of the study was defined as: Collecting and evaluating data of potential volatile biomarkers in the exhaled air of subjects with and without Covid-19 by the novel sensors of the invention. COVID-19 positive and negative subjects were enrolled. Classification to the 2 study arms was based on a PCR test result. Three medical centers participate in the study: Shamir Health Corporation (“Assaf Harofeh”) in Israel; Northwell Health, Inc. in the United States (“Northwell”); Zayed Military Hospital Abu Dhabi (“Zayed Hospital”).

[0119] The study was performed with the sensors installed in 2 devices: 1. The first-generation device with single use units that include the sensors. 2. a device with multi use sensors. The collected data from the devices were analyzed independently by two distinct methods.

[0120] The first dataset was collected with the first-generation device with singe use units that include the sensors of the invention. The dataset included subjects tested with the device at two sites: 35 samples from Northwell N.Y., and 31 samples from Shamir medical center IL. Each test file consisted of responses from duplicated sensor array, and therefore each test file was split into two sample files, based on the sensor sets. Some of the sensors failed to respond, and therefore datasets that included failed sensors were discarded. The total number of sample files that were analyzed after the error-prone samples were discarded is: Northwell—35 sample files (representing 24 tested subjects—17 positives, 7 negatives) and Shamir medical center—31 sample files (representing 21 tested subjects—14 positives, 7 negatives). The data was analyzed by Brainchip with a Spiking Neural Network, the adjacent confusion matrix shows the results on the test set. The test set included 31 samples-21 positives and 10 negatives from 21 tested subjects. Zero out of 21 positive samples were identified correctly which represents 100% sensitivity and 4 out of 10 negative samples were identified correctly which represents 40% specificity. The overall accuracy was 80.65% The second study was performed with the multiuse Nallose sensors installed in Sniffphone device. The dataset included 165 samples taken from 141 subjects tested with Sniffphone device at Zayed Military Hospital—65 samples from 65 COVID-19 positive subjects and 100 samples from 76 COVID-19 negative subjects (Several negative subjects were sampled two or three times). A Linear discriminative analysis was performed. The adjacent confusion matrix shows the results on the test set that that was completely blind to the training and validation of the model. The test set included 37 samples—8 positive and 29 negative samples from 27 tested subjects. Seven out of eight positive samples were identified correctly which represents 87.5% sensitivity, and 25 out of the 29 negative samples were identified correctly which represents 86.2% specificity. The overall accuracy was therefore 86.5%.

[0121] The same data set was analyzed also by the SNN methodology. To make the SNN most efficient, 34 samples were discarded due to noise or improper vector dimensionality. Thus, the dataset included 131 samples taken from 126 subjects tested with Sniffphone device at Zayed Military Hospital-62 samples from 62 COVID-19 positive subjects and 69 samples from 64 COVID-19 negative subjects (Several negative subjects were sampled two or three times). The adjacent confusion matrix shows the results on the test set that that was completely blind to the training and validation of the model. The test set included 53 samples—20 positive and 33 negative samples from 53 tested subjects. Nineteen out of 20 positive samples were identified correctly which represents 95% sensitivity and 29 out of 33 negative samples were identified correctly which represents 87.87% specificity. The overall accuracy was therefore 90.5%.

[0122] Two different analysis methods were applied on the dataset and both showed excellent results for the differentiation between COVID positive and COVID negative. While the multiuse sensors achieved a much better specificity (˜87%) compared to the single use sensors (40%), this is more likely a result of the vast difference between the datasets: the dataset of the multiuse sensors included 165 samples from 141 subjects while the dataset of the single-use sensors included 66 samples from 45 subjects. During the Clinical study with COVID19 patients the company further improved the 4 components of the device: the mechanical design including the breath collection mechanism, the electronics, the sensors and the classifying algorithm.