For the GeNose C19 sensor array, the sensitivity of each sensor during exposure to varying VOC concentrations depends on the used active material. Moreover, the sensor behaviors might be slightly altered when they were tested to the breath samples from different patients, although they were from the same group (either positive or negative COVID-19). This occurrence could be understood because the content and complexity of the exhaled VOCs are diverse, as discovered in another breath analysis study using GCIMS24. Several VOC biomarkers could be identified as the discriminants for distinguishing between positive and negative COVID-19 patients (e.g., ethanal, acetone, acetone/2-butanone cluster, 2-butanone, methanol monomer and dimer, octanal, feature 144, isoprene, heptanal, propanol, and propanal)24. Nonetheless, the compounds observed from two different hospitals (i.e., Edinburgh, the United Kingdom (UK), and Dortmund, Germany) in their study were dissimilar for the same case of COVID-19 patients, which then added more complexity in analyzing the obtained breath data. These limitations were due to uncertainties in the instrument setup, operating conditions, and background contamination levels.
Thus far, a detailed study in those matters has not been performed. Meanwhile, another clinical GCIMS study conducted by researchers in Beijing, China, suggested several other potential breath-borne VOC biomarkers for COVID-19 (i.e., acetone (C3H6O), ethyl butanoate, butyraldehyde, and isopropanol)72. They found that the decrease and increase in acetone (C3H6O) and ethyl butanoate levels, respectively, due to the changes in metabolites resulting from SARS-CoV-2 infections, are distinctive for COVID-19 patients72,73. Moreover, the average measured isopropanol and butyraldehyde for the COVID-19 patients were lower than those for the healthy control and lung cancer and non-COVID-19 respiratory infection patients. The metabolomics of exhaled breaths in critically ill COVID-19 patients were also investigated from a research consortium in France using a proton transfer reaction quadrupole time-of-flight mass spectrometer74. They observed four prominent VOCs (i.e., methylpent-2-enal, 2,4-octadiene, 1-chloroheptane, and nonanal) that could discriminate between COVID-19 and non-COVID-19 acute respiratory distress syndrome patients74. Overall, the reported MS studies in several different countries (i.e., UK, Germany, France, and China) indicate that the distinctive VOC biomarkers for COVID-19 may vary across the world and should be further investigated based on the community, race, and cases with large cohorts75.
In contrast to the MS method that attempts to quantitatively find and identify the exact VOC biomarkers from exhaled breaths, our technique used in GeNose C19 focuses more on the AI-based pattern analysis of integrated sensor responses to complex VOCs, qualitatively resulting from the combined extra-pulmonary metabolic and gastrointestinal manifestations of COVID-1976. Thus, the breath data analysis and decision-making procedure can be performed in a simple way and short time, respectively, with a high detection accuracy. To enable this, besides having a high sensitivity, chemoresistive sensors should ideally be designed to possess a high selectivity to a specific analyte in a gas mixture and zero cross-sensitivity to other compounds in the operating background. Such sensors were normally constructed in hybrid organic/inorganic structures with 3D nano-architectures (e.g., nanofibers, nanowires, and nanofins), enhancing the active surface-area-to-volume ratios77,78. Here, the surfaces of semiconductor nanostructures were often functionalized with certain self-assembled monolayers or polymers to specifically detect the target gas molecules32,34,79. Nevertheless, these organic materials suffer from low robustness. They are all well-known to degrade within a short duration of use (i.e., their chemical compositions will alter downgrading the sensor performance). As a result, pure inorganic materials (metal oxide semiconductors) are still preferably manufactured by sensor companies and widely used in gas sensing applications, including in the GeNose 19 system. Here, a single sensor alone is not sufficient for performing a specific breath pattern recognition because exhaled VOCs might have similar characteristics. This selectivity drawback could be alleviated by employing an array of 10 sensors with different sensitivities and integrating the machine learning-based breath pattern recognition algorithms.
Furthermore, to demonstrate the proof of concept ability of GeNose C19 for detecting VOCs in human breaths, we performed additional sensing assessments for acetone vapors in a modified test setup (see Supplementary Fig. 2). However, COVID-19 itself cannot be detected by simply sensing or measuring the acetone alone. This testing was mainly dedicated to demonstrate that the GeNose C19 sensor array can detect one of the VOCs normally contained in human breaths and exhibits different sensitivity levels when exposed to various gas concentration levels, which also mimics the real case of exhaled breaths from different persons or patients. The gas sensing configuration for the acetone testing, which utilizes a microsyringe for vapor injection, has already been used in our former experiments for other VOC sensor types (e.g., nanofiber-functionalized QCMs for sensing trimethylamine and butanol gases)35,80,81. Acetone was chosen as a VOC model in this additional study because it is not only produced in the rebreathed breath (0.8 to 2.0 ppm)82, along with other VOCs (alcohol) and CO, but is also one of the significant breath-borne COVID-19 biomarkers based on the study by Chen et al.72. Moreover, in clinical practices, breath-containing acetone has been extensively examined to diagnose other diseases (i.e., lung cancer, diabetes mellitus, starvation, and ketogenic diet)83.
As shown in Supplementary Fig. 2b, c, the S3 and S7 sensors (or their extracted features of F3 and F7) demonstrated the poorest responses toward acetone vapors. Conversely, the S2, S8, and S9 sensors exhibited higher sensitivities than the others. The sensor output signals given by the GeNose C19 data acquisition system agree well with those measured by a calibrated digital voltmeter. Increasing the acetone vapor concentrations from 0.04 to 0.1L with 0.02 intervals resulted in higher responses of the three sensors (S2, S8, and S9), whereas the S3 and S7 sensors were irresponsive (see Supplementary Fig. 2d). In particular, each vapor concentration was measured 10 times to acquire quantitative results. Lastly, as depicted in Supplementary Fig. 2e, LDA discriminated the output voltages produced by the sensors during their exposures to four different acetone concentrations (i.e., 0.040.1L).
In terms of ambient conditions, temperature and humidity might influence the performances of metal oxide semiconductor sensors84. Thus, to investigate their effect, we also performed cross-sensitivity assessments in respect to the two parameters for all the employed GeNose C19 sensors (see Supplementary Figs. 3 and 4). This testing is important because depending on the sensitivities of the sensors toward temperature and humidity, the obtained sensor results during the breath analysis can be disturbed, leading to a difficult interpretation of the data. Moreover, if the sensors are too reactive to the two ambient parameters, the measured data can then be unreliable to analyze the effect of VOCs in the human breath because changes in the signals were mainly affected by the temperature and humidity, not the target gases. Such a cross-sensitivity is a common reliability test for gas sensors. For GeNose C19, the environmental effect can be minimalized and controlled by performing two main procedures. First, environmental checking needs to be conducted while placing GeNose C19 in the measurement room/area. Here, the selection of the machine placement (analysis on air circulation, humidity, and temperature) plays a key role in maintaining good-quality results. GeNose C19 could sense the environmental humidity and temperature levels by utilizing humidity and temperature sensors integrated inside the system. The measurement was displayed in the program interface. Hence, the user or operator could notice the condition. In a real situation during breath sampling, the machine could only be operated if the humidity and temperature inside the chamber were in the ranges of 3050% and 2642C, as defined by the AI-based program in the system. Such a setting is adjustable to meet future demand and placement environments. Second, after checking the environmental condition, the baseline normalization protocol during the sample analysis can be done (see Methods on the GeNose preconditioning). During the AI interpretation of the VOC patterns, several protocols were employed, including signal baseline normalization. By performing baseline normalization, all the sensors that behaved and started from different baselines in different environments can always be calibrated to the standard normalization. Hence, the adaptability of the machine can be improved in new foreign environments.
In the case of acetone testing, the sensors yielded similar responses from three repeated measurements, indicating their reliable sensing results. The sensor resistance decreased (i.e., a higher output voltage was obtained) when the temperature was ramped up from 40C to 46C, and the humidity was kept stable at (30%1%) RH (see Supplementary Fig. 3). Different from silicon micromechanical resonant sensors that have frequency shift interferences caused by the temperature-induced Youngs modulus change (material softening)37,85, the resistance decrease in the employed metal oxide semiconductor sensors (e.g., n-type SnO2 with a bandgap of 3.6eV) at high temperatures was caused by the increasing number of electrons that have sufficient energy crossing to the conduction band and thus contributing to the conductivity86. Because this is a natural characteristic of semiconductor materials, we could overcome this effect in GeNose C19 by controlling the temperature inside the test chamber at relatively stable values (i.e., (42C2C)) during the sensing phase of the exhaled breath.
Similar to the trend shown in the cross-temperature test, the sensor resistance also dropped to a lower value, resulting in a higher output voltage when the relative humidity was raised from 30% to 35% and the temperature was set constant at (40C1C) (see Supplementary Fig. 4). The electrical characteristics of metal oxide semiconductors changed due to the water adsorption on their surface while being exposed to humid air. Two different mechanisms of chemisorption and physisorption processes took place to create the first layer (i.e., chemisorbed layer) and its subsequent films of water molecules (i.e., physisorbed water layers), respectively87. If the first chemisorbed layer has been formed, then the successive layers of water molecules will be physically adsorbed on the first hydroxyl layer. Because of the high electrostatic fields in the chemisorbed layer, the dissociation of physisorbed water can easily occur, producing hydronium ion (H3O+) groups. Here, the conduction mechanism relies on the coverage of adsorbed water on the metal oxide semiconductor. First, in the event only hydroxyl ions exist on the metal oxide surface, the charge carriers of protons (H+) resulting from hydroxyl dissociation will hop between adjacent hydroxyl groups. Second, after the water molecules have been adsorbed but not fully covered the oxide surfaces, the charge transfer will be dominated by H3O+ diffusion on hydroxyl groups, despite the occurring proton transfer between adjacent water molecules in clusters. Finally, once the continuous film of the first physisorbed water has been formed (i.e., full coverage of metal oxide by the physisorbed water layer), proton hopping between neighboring water molecules in the continuous film will be responsible for the charge transport88. More detailed explanations of the sensing mechanism and adsorption of water molecules on metal oxide semiconductor surfaces are described elsewhere84,87,88. Again, in the conducted cross-sensitivity measurements (Supplementary Fig. 4), the signal changes of the GeNose C19 sensors affected by humidity are relatively lower (i.e., <100mV) compared to those exposed to exhaled breaths (i.e., ~1V, as shown in Fig. 3a, b). Thus, temperature and humidity will insignificantly influence the system performance during breath measurements, when GeNose C19 has been well preconditioned.
To confirm the performance of our GeNose C19, RT-qPCR was used as the reference standard on the basis of the health service standard protocol underlined by the Indonesian Ministry of Health. Based on the analysis of the RT-qPCR protocol using Bayes theorem, RT-PCR tests cannot be solely relied upon as the gold standard for SARS-CoV-2 diagnosis at scale. Instead, a clinical assessment supported by a range of expert diagnostic tests should be used. Here, although our study mentioned that RT-qPCR was used as the reference standard, clinical data from each patient were also collected and analyzed.
According to a recently published systematic review study, the need for repeated testing in patients with suspicion of SARS-Cov-2 infection was reinforced because up to 54% of COVID-19 patients might have an initial false-negative RT-qPCR89. Meanwhile, in the case of false-positive rates of RT-qPCR, much lower values (i.e., 016.7% with an interquartile range of 0.84.0%)90,91 were exhibited in several studies, which were affected by the quality assurance testing in laboratories. Public Health England also reported that RT-qPCR assays showed a specificity of over 95%, so up to 5% of cases were false positives91. Moreover, the overall false-positive rate of high throughput, automated, sample-to-answer nucleic acid amplification testing on different commercial assays was only 0.04% (3/7,242, 95% CI, 0.01% to 0.12%)92. False-positive SARS-CoV-2 RT-qPCR results could originate from different sources (e.g., contamination during sampling, extraction, PCR amplification, production of lab reagents, cross-reaction with other viruses, sample mix-ups, software problems, data entry errors, and result miscommunication)93. In our case, all the bought and used reagents were checked and calibrated daily to avoid false positives (i.e., no false positive of RT-qPCR result was found in this study). Meanwhile, the false-negative of the RT-qPCR result was found in three patients in their first examination, but positive results were revealed on the second examination the next day. Again, the detailed test procedure can be found in the Methods.
Currently, diagnostic methods used to screen COVID-19 are antigen test, rapid molecular test, and chest CT scan. Antigen tests have an average sensitivity of 56.2% (95% CI: 29.579.8%) and average specificity of 99.5% (95% CI: 98.199.9%)94. The average sensitivity and specificity for the rapid molecular tests are 95.2% (95% CI: 86.798.3%) and 98.9% (95% CI: 97.399.5%), respectively94. Meanwhile, chest CT scan possesses an average sensitivity and specificity of 87.9% (95% CI: 84.690.6%) and 80.0% (95% CI: 74.984.3%), respectively95. Nonetheless, these diagnostic methods have their drawbacks. The average sensitivity of antigen tests is not high, as shown by the study above, and it declines when the viral load decreases, which often happens to COVID-19 patients. Moreover, the sample collection is invasive (by a nasopharyngeal or oropharyngeal swab). Rapid molecular testing also employs an invasive sample collection method (by a nasopharyngeal or oropharyngeal swab), and the turnaround time of point-of-care rapid molecular tests still takes at least 20 min96. Moreover, chest CT scan exposes patients to radiation and is not specific.
Compared to these diagnostic methods, GeNose C19 has the potential to be a screening test. A breath test with the portable GeNose C19 is noninvasive and easy to use because it only requires patients to breathe into a sampling bag with minimal preparation, has a fast analysis time, and does not have radiation concerns. Similar to other biological samplings in several laboratory examinations (e.g., blood glucose sampling and chemical blood analysis), GeNose C19 also requires preparation of subjects before breath sampling, such as fasting (i.e., refraining from eating, smoking, or drinking anything other than water at minimum 1h before sampling). However, the duration of the analysis starting from the breath sampling to the test result decision only takes ~3min. The sensitivity and specificity results of GeNose C19 from the profiling tests show that combining GeNose C19 with an optimum machine learning algorithm can accurately distinguish between positive and negative COVID-19 patients. Hence, it opens an opportunity for using this developed breathalyzer as a rapid, noninvasive COVID-19 screening device based on exhaled breath-print identification.
Several factors may influence breath-prints, i.e., pathological and disease-related conditions (smoking, comorbidities, and medication), physiological factors (age, sex, food, and beverages), and sampling-related issues (bias with VOCs in the environment)97. A previous study revealed that older age altered breath-prints in patients with lung cancer98. There were concerns that several other respiratory diseases may present similar VOC patterns to those from the COVID-19. Several studies reported that several comorbid and confounding factors (e.g., chronic obstructive pulmonary disease, asthma, tuberculosis, and lung cancer) might affect the composition of VOCs99,100. Thus, patients with other respiratory diseases can have different patterns of VOCs that result in different sensor signals, suggesting that the electronic nose may still determine the COVID-19 infection to a certain degree by continuing to train its AI database in reading VOCs from confirmed positive COVID-19 patients. Our studies showed no significant difference in the detected sensor signal patterns of patients with comorbidities compared to those without comorbidities. Nonetheless, due to the few comorbid cases obtained in our subjects, which could be considered the limitation in our current study, the influence of existing comorbidities on the VOC pattern cannot be concluded and will be further evaluated in the next research.
Food and beverages (e.g., poultry meat and plant oil) can influence breath-prints, whereas smoking may increase the levels of benzene, 2-butanone, and pentane and simultaneously decrease the level of butyl acetate in exhaled breaths101,102,103. In our study, none of the patients was smoker. The comorbidities were also comparable between the case and control groups. There was no significant difference in the consumption of food and beverages between the two groups. The measurements were conducted in the same environment for all the participants. Thus, there was no bias with other interfering VOCs.
However, the possible presence of physiological variations resulting from physiological and biochemical changes in the body due to alterations in the respiratory rhythm affected by the manipulated breathing technique should also be considered61. Therefore, in our work, breath sampling was performed in such a defined protocol to collect only the third exhaled end-tidal breath. Accordingly, the natural breathing pattern and rhythm can be preserved, resulting in minimal changes in VOCs. We avoided excessive effort or repeated sampling in each breath collection as previous studies reported that it might alter the quality of collected VOCs104. The disturbance from other factors to breath test results is minimal. However, such confounding factors are most likely present in the real implementation and can affect at least breath-prints to a certain degree. Further study is now being conducted to reveal the effects of various confounders.
Our study using GeNose C19 did not evaluate the distinctive concentration of each VOC found in breath samples of patients with positive or negative COVID-19. However, to investigate the types of VOCs produced in exhaled breaths of the positive and negative COVID-19 patients, we conducted another characterization based on GCMS for several exhaled breaths of patients (see Supplementary Table 3). In the extracted results, there was no significant difference in the composition of VOCs between patients with positive and negative COVID-19, suggesting that the difference in the breath-print pattern may be contributed by the variation in the concentration or proportion of several VOCs rather than the presence of one or two signature VOCs. For example, acetone was reported to be one of the VOCs with the highest concentration emitted by healthy humans104. However, in COVID-19-positive patients, acetone was reported to be in a lower proportion, compared to the healthcare worker or healthy control group72. Meanwhile, another VOC (i.e., ethyl butanoate) has been reported as one of the signature VOCs in COVID-19 patients, whose concentration is slightly higher compared to the healthy control72.
Anosmia (i.e., the olfactory system cannot accurately detect or correctly identify odors) is one of the most frequently identified COVID-19 symptoms45,105. CO has been linked with this issue because it is an olfactory transduction byproduct related to the reduction of cyclic nucleotide-gated channel activity that causes a loss of olfactory receptor neurons45,106. In our GCMS results (Supplementary Table 3), six sensors in GeNose C19 (i.e., S1, S3, S4, S5, S6, and S8) could detect CO. Aside from CO, the GCIMS studies in Dortmund, Germany, and Edinburgh, UK indicated that aldehydes (ethanol and octanal), ketones (acetone and butanone), and methanol are biomarkers for COVID-19 discrimination24. This result is however different from the finding from another research group in Garches, France, using the proton transfer reaction quadrupole time-of-flight MS, where four types of VOCs (i.e., 2,4-octadiene, methylpent-2-enal, 1-chloroheptane, and nonanal) could discriminate between COVID-19 and non-COVID-19 acute respiratory distress syndrome74. Studies conducted in two cities in the USA (Detroit, Michigan and Janesville, Wisconsin) by Liangou et al. reported another set of eight compounds (i.e., nitrogen oxide, acetaldehyde, butene, methanethiol, heptanal, ethanol, methanol-water cluster, and propionic acid) as key biomarkers for the COVID-19 identification in human breath. Moreover, in Leicester, UK, seven exhaled breath features (i.e., benzaldehyde, 1-propanol, 3,6-methylundecane, camphene, beta-cubebene, iodobenzene, and an unidentified compound) measured by the desorption coupled GCMS were employed to separate RT-qPCR-positive COVID-19 patients from healthy ones107. In our measurement, camphene was detected only in the negative COVID-19 breath sample by S10.
Furthermore, Chen et al. reported two sequential GCMS studies in Beijing, China, that resulted in totally different breath-borne biomarkers for COVID-19 screening, despite using the same measurement approach72,108. Their first measurement reported in 2020 indicated that COVID-19 and non-COVID-19 patients could be differentiated by solely monitoring three types of VOCs (i.e., ethyl butanoate, butyraldehyde, and isopropanol)72. Nonetheless, in their second report in 2021, acetone was detected as the biomarker among many VOC species because its levels were substantially lower for COVID-19-positive patients than those of other conditions73. In our GeNose C19 sensor array, acetone can be detected in S8109. Recently, ammonia has also been proposed as another biomarker for COVID-19, whose relation to complications stemming from the liver and kidneys was affected by the SARS-CoV-2 infection110.
In all the already described examples of MS studies worldwide, the identification and determination of specific COVID-19 biomarkers in breath clearly remain challenging. Here, different discriminant compounds can be yielded depending on several parameters (e.g., measurement technique, filtering approach, location, and breath sample type). Nonetheless, we can still extract some information from our GCMS results (Supplementary Table 3). A hydrocarbon of ethylene was sensed by S10 in the positive COVID-19 breath sample. Meanwhile, for the negative COVID-19 breath samples, other hydrocarbons (i.e., butyl aldoxime, decane, and benzene) were detected by S10. Furthermore, S2 and S9 could measure a few specific esters (i.e., benzoic acid, 3-hydroxymandelic acid, and acetic acid) in the negative COVID-19 samples. Generally, the appearances of the three sensors (S2, S9, and S10) were dominant as compared to those of the others. For instance, S2 and S9 were highly sensitive toward aldehydes and esters, whereas S10 was likely to be reactive toward hydrocarbons.
Regardless of the successful compound extraction and its association with GeNose C19 sensor array, our GCMS characterization was only performed in a low number of samples. Therefore, a further investigation with a larger number of breath samples still requires to be carried out in the near future to correlate the measurement results of GeNose C19 and GCMS methods in a more thorough way, especially in Indonesia. This method also includes more investigations on the possible influence from other respiratory-related viruses (e.g., influenza, respiratory syncytial virus, and rhinovirus). The presence of viruses other than SARS-CoV2 will affect the VOC profile in breaths to a certain degree. However, in our current setup, it will be mostly recognized by the AI algorithm in GeNose C19 as non-reactive, which means that it contains VOC-based breath-prints not typical to a SARS-CoV2 infection. Influenza and rhinovirus infections manifest a high amount of heptane, nitric oxide, and isoprene111. Consistently, our preliminary study on breath samples from a few patients confirmed to have rhinovirus based on RT-qPCR and showed a high response on S8, suggesting a high amount of isoprene or isopropanol. However, further comparison analysis with more numbers of validated breath samples data will be definitely necessary to obtain a solid conclusion on this matter.
In terms of the enhanced sensing technology, once the VOC biomarkers can be clearly determined, a molecular imprinting method could be applied to generate highly selective sensors that target these specific VOC markers. Hence, the sensitivity and specificity of GeNose C19 and its overall accuracy can be further improved. Another critical step for the system development is to conduct a diagnostic test with a large cohort to strongly elucidate its potential as a diagnostic tool in the near future.
Other limitations of our study are that a direct correlation between the level of the virus gained from the swab and the amount of VOC concentration could not be drawn. These conditions are partly caused by the fact that VOCs were not directly produced by the virus, but rather by host cells infected by the virus as a part of their metabolic response to the infection. GeNose C19 could only predict the presence of the virus based on the resulting VOCs in the breath produced by respiratory tract epithelial cells and immune cells that were infected by the SARS-CoV2 virus. Nevertheless, a study on the correlation between the positivity rate of breath results and level of the cycle threshold value (Ct value) gained from RT-qPCR examination has been of interest for the next research. Here, more insights into the performance of GeNose C19 will be gained in terms of sensitivity, specificity, and accuracy levels correlated with the level of Ct value of RT-qPCR. The Ct value itself is currently accepted as an alternative parameter to determine the level of the viral load in each individual on the basis of the minimum cycle threshold necessary to duplicate the viral component to be read. Nonetheless, GeNose C19 combined with RT-qPCR using the Ct value has a limitation for estimating the exact number of the viral load. It was also a question of whether a person with a positive PCR test result for SARS-CoV-2 is automatically infectious or infectious only if the Ct value is below a certain limit (e.g., Ct value of <35)112,113. In another study, knowing the typical viral load of SARS-CoV-2 in bodily fluids and host tissues, the total number and mass of SARS-CoV-2 virions in an infected person could be estimated114. Each infected person carries the total number and mass of SARS-CoV-2 virions of 1091011 virions and 1100g, respectively, during the peak infection114.
Again, this study was meant to demonstrate a proof of concept that breath sampling and detection can be used to predict COVID-19 infection. Essentially, the calculated performance values in our study show the reliability of the DNN algorithm in predicting the training and testing data of breath samples, suggesting the great potential of the GeNose technology, fortified by the DNN algorithm to be used as a COVID-19 screening tool. Here, we performed the study using a so-called open-label design, where we already knew the COVID-19 status of the subjects before conducting sampling and classifying the sampled data into case and control groups. Using this method, we read, found, and compared the breath sample pattern profiles in each respected group and employed them as training data to build our first AI database, in which all data were validated by the test results of RT-qPCR supported with clinical data. A combined measurement of GeNose C19 with GCMS will be conducted in the near future to answer questions related to distinctive VOCs for COVID-19.
Lastly, another critical step for the system development is to confirm the usability and performance in the clinical setting, where a study on the clinical diagnosis of COVID-19 with a larger number of exhaled breath samples is currently performed to prove the potential of GeNose C19 as a rapid COVID-19 screening tool using a cross-sectional design and double-blind randomized sampling. Here, breath samples and nasopharyngeal swab specimens are taken in the situation where the operator or sampler does not know the true condition of patients. A double-blind fashion means that neither the sampler nor subjects know their true condition during the sampling process. The breath samples were analyzed by GeNose C19 without knowing the result of RT-qPCR, and swab samples were examined by RT-qPCR without prior knowledge of the GeNose C19 result. Both results were then compared to each other to draw a conclusion. In this approach, RT-qPCR will still be used as the reference standard.
Read more from the original source:
Fast and noninvasive electronic nose for sniffing out COVID-19 based on exhaled breath-print recognition | npj Digital Medicine - Nature.com