中文   |   English   |   Email Alert
Fig/Tab
Advanced Search
image

Gas-Sensing Technology for Human Breath Detection

Jinglong Zhao , Wenfeng Shen , Dawu Lv , Jiaqi Yin , Tongxiang Liang , Weijie Song

Progress in Chemistry ›› 2023, Vol. 35 ›› Issue (2) : 302 -317.

PDF
Progress in Chemistry ›› 2023, Vol. 35 ›› Issue (2) : 302-317 DOI: 10.7536/PC220712
research-article
Scan to read & Share
Gas-Sensing Technology for Human Breath Detection
Author information +
History +
PDF

Abstract

Changes in volatile organic compounds concentrations in human breath are closely related to certain diseases, and the diagnosis of diseases by analyzing volatile organic compounds in human breath is a non-invasive, and easy-to-use tool that has received increasing attention in recent years for applications in disease diagnosis and early screening. There are currently two main types of equipment for detecting volatile organic compounds in exhaled breath: mass spectrometry-based analytical instruments and gas-sensitive sensors. With easy integration, miniaturization, low cost, and simple operation, gas-sensitive sensors have broad application prospects in the future diagnosis and early screening of large-scale population diseases. This review systematically describes the working mechanism of gas sensors, sensor performance, the current status of application of different sensitive materials and the application of different gas sensor types in human breath detection, the types of volatile organic compounds in human breath that are associated with some diseases are also introduced, which is followed by a brief introduction to the means of breath sampling and the data processing methods currently in use. Finally, the problems of current gas sensor technology in breath detection are pointed out, and the prospects of gas sensor technology in human breath detection are foreseen.

Contents

1 Introduction

2 Related Research on Sensor Technology

2.1 Sensor Detection Technology

2.2 Sensor Sensitive Materials

2.3 Types of Gas Sensors

3 VOCs in different diseases and the current development status

3.1 Exhaled breath biomarkers in different diseases

3.2 Electronic nose technology in disease diagnosis

4 Sensor Response Data Acquisition and Processing

4.1 Breath sampling

4.2 Eigenvalue Extraction

4.3 Pattern recognition algorithm

5 Conclusion and Outlook

Graphical abstract

Keywords

breath analysis / gas sensors / volatile organic compounds / non-invasive diagnostics

Cite this article

Download citation ▾
Jinglong Zhao, Wenfeng Shen, Dawu Lv, Jiaqi Yin, Tongxiang Liang, Weijie Song. Gas-Sensing Technology for Human Breath Detection. Progress in Chemistry, 2023, 35(2): 302-317 DOI:10.7536/PC220712

登录浏览全文

4963

注册一个新账户 忘记密码

1 Introduction

As early as ancient Greece, doctors have used human breath to diagnose diseases. For example, the smell of rotten apples in the breath can diagnose patients with diabetes, and the smell of fish in the breath can diagnose liver disease and urine in the breath may diagnose kidney disease[1]. In recent years, with the development of detection technology and in-depth study of breath, it has been found that there are more than 3000 Volatile Organic Compounds (VOCs) in human breath, some of which can be used as breath biomarkers for disease diagnosis, which indicates that breath analysis has great potential in non-invasive clinical medical detection[2][3].
Traditional medical testing methods are usually invasive testing or body fluid testing, which may cause trauma to patients or inconvenience in sampling. Breath detection has the advantages of non-invasive and convenient sampling, which is suitable for large-scale disease screening. At present, there are four main detection technologies for VOCs in human exhalation: Gas Chromatography-Mass Spectrometry (GC-MS), Proton Transfer Reaction-Mass Spectrometry (PTR-MS), Selected Ion Flow Tube-Mass Spectroscopy (SIFT-MS) and Gas sensor[4][5][6][7]. Among them, GC-MS, PTR-MS and SIFT-MS have the advantages of high specificity, high sensitivity and low detection limit. However, they all have the disadvantages of complex equipment operation, expensive detection cost and huge equipment volume, which make them unable to enter the general clinical stage on a large scale, and then difficult to be applied to ordinary communities and families. In contrast, gas sensors have outstanding advantages such as miniaturization, low cost, real-time detection and trace-level detection accuracy, and are expected to become an important technical means of disease detection in the future[8].
In this paper, the working principle and material preparation of gas sensors are introduced, and the application progress of sensors in human exhalation is also introduced. The types of VOCs in human exhalation, exhalation sampling methods and data processing methods related to some diseases were described. Finally, the current research of sensor technology in breath detection is summarized.

2 Research on Sensor Technology

2.1 Sensor Detection Technology

In recent years, sensors have gradually become an important diagnostic tool in exhalation analysis, and have gradually shown their strong application potential[9]. Gas sensors have the advantages of low cost, simple operation and small device size, and have great potential in the development of portable devices. In addition, short response time, direct access to results and short recovery time make gas sensors more advantageous in clinical monitoring and personalized screening, diagnosis and disease prognosis monitoring[10]. Exhalation analysis of the sensor can be achieved in two different ways. The first method is to use a single sensor material to identify a single gas, but other gas interference has a greater impact on the detection results, and there are some limitations. The other is to use an array of sensors made of different materials to simultaneously respond to all or most of the components of a complex mixture in a complementary manner, and to detect complex compounds rather than a single exhaled compound. This method achieves the purpose of cross detection, with more accurate detection results and smaller errors[11]. In order to make the sensor array have a good effect in disease diagnosis, pattern recognition algorithms are usually used to analyze the exhalation analysis data, such as principal component analysis, support vector machine, k-nearest neighbor and artificial neural network[12,13]. Through the combination of sensor test data and intelligent pattern recognition algorithm, the database model is established, and then the data is further verified to achieve the purpose of medical diagnosis.

2.2 Sensor Sensitive Material

2.2.1 Metal oxide semiconductor material

Metal oxide semiconductor (MOS) materials are compound semiconductor materials formed by Metal and oxygen. These materials exhibit wide band gap and full-spectrum electronic properties[14].
In recent years, many scholars have developed many gas sensors with metal oxides as sensitive materials, which also show excellent performance in breath detection. For example, the gas sensor with ternary metal oxide Zn2SnO4 developed by Hanh et al. As a sensitive material has a good response to 125 ppm acetone at 450 ℃, and the detection limit can reach 0.67 ppb by theoretical calculation, which is far lower than the acetone concentration in the breath of diabetic patients, and has great potential in diabetes monitoring[15]. Choi et al. Synthesized a Nb-doped WO3 acetone sensor by ultrasonic spray pyrolysis method, and the detection limit of the sensor can reach 8.9 ppb, which can effectively monitor the acetone concentration in the breath of ketogenic diet patients, and is of great significance to the development of portable respiratory sensors[16]. Alev et al. Synthesized WO3 nanosheets on a Al2O3 substrate by hydrothermal method, and the sensor prepared by this material showed excellent sensitivity to ammonia of 1 – 80 ppm, providing an important means for non-invasive detection of kidney diseases[17]. Selvaraj et al. Developed an acetone sensor with ZnO/NiO as the sensitive material, which can detect diabetes. The sensor has a maximum response of 6.06 to 100 ppm acetone at room temperature. It is one of the more excellent sensors at present and has a good application prospect in diabetes detection[18].
Although the sensors prepared by metal oxides have many advantages, they also have some disadvantages. Metal oxide semiconductor sensors have high operating temperature, so their power consumption is relatively high. In addition, its selectivity is poor, and it may respond to the same kind of gas, so it can not accurately distinguish the target gas.

2.2.2 Conductive polymer material

Conducting polymer is one of the important materials for the preparation of different types of VOCs sensors. In the early 1980s, polypyrrole (PPy), polyaniline (PANI), polythiophene (PTh) and their derivatives were used as sensitive materials for gas sensors[19]. When conductive polymers are used for VOCs sensing materials, the target gas interacts with them through weak intermolecular forces (such as π-π interactions, hydrogen bonds, and dipole-dipole interactions), as shown in Figure 1, and they behave as electron donors or electron acceptors, increasing or decreasing the carrier concentration, thereby changing the conductivity of the conductive polymer[20]. Due to the strong affinity of most VOCs to the polymer through various intermolecular interactions and the abundant adsorption sites on the polymer chain, the internal structure of the polymer film changes. In addition to charge/proton doping, it may also lead to changes in its electrical properties. For example, the swelling of the polymer induced by gas adsorption can expand its interchain distance, thereby limiting the electron hopping between chains and reducing the conductivity of the sensing film (Figure 1). In some cases, certain alcohols and ketones can interact strongly with nitrogen atoms on conducting polymers such as PANI, causing their molecular chains to expand to a more stretched structure, thereby improving electron transfer.
The molecular skeleton and side chains/groups of the polymer can be easily modified, resulting in changes in electrical properties and surface chemistry, and the sensor prepared from the sensitive material has a high response to specific VOCs. Conducting polymers are also easy to synthesize and have good mechanical properties. Zhang et al. Prepared a room temperature H2S sensor based on SnO2/RGO/PANI on PET substrate for early diagnosis of halitosis in human breath. The detection limit of the sensor reached 50 ppb, which was lower than the concentration of H2S in the breath of halitosis patients[21]. Shahmoradi et al. Used PPy/SRGO gas sensor prepared by in-situ polymerization to analyze the change of ammonia concentration in the breath of dialysis patients before and after dialysis.The results showed that the concentration of ammonia in the breath of patients after dialysis was significantly reduced, which was consistent with the results of blood urea nitrogen, indicating that the sensor technology has good feasibility in breath detection[22]. Reiner-Rozman et al. prepared a polyaniline-based sensor. The sensor prepared by F4TCNQ-doped polyaniline film can identify five different breath biomarkers of cancer, which is one of the excellent polymer sensors in the field of breath detection[23].

2.2.3 Carbon-based material

Carbon-based materials include carbon black, carbon nanotubes, graphene, etc. They have a high specific surface area, which is beneficial to improve the sensitivity of sensitive materials, resulting in better gas sensing properties. The sensitivity and selectivity of carbon-based materials can be engineered by different techniques, both to create defects and to graft functional groups on their surface in a controlled manner. Due to their unique physical structure and excellent electrical properties, they have attracted wide attention and are used as high-performance VOCs sensing materials. Moreover, the tunability of the chemical and surface characteristics of carbon-based materials increases their potential for building selective sensors against VOCs gases.
Carbon black is one of the earliest sensitive materials used in VOCs sensors. It is usually compounded with an insulating organic polymer to make a sensitive film material. Carbon black makes the film conductive, while the organic polymer provides chemical diversity between the sensor elements. The percolation theory well describes the relationship between the resistivity and the carbon black content in the carbon black-organic polymer composite[24]. At low carbon black content, the composite is an insulator. With the increase of carbon black content, the resistivity of the composite decreased sharply. When this composite membrane swells due to the absorption of gas, the conduction pathway is disrupted, causing a change in resistance[25].
Carbon nanotubes, another promising nanomaterial, are widely used in the detection of VOCs. According to their structure, they can be divided into two categories: single-walled carbon nanotubes and multi-walled carbon nanotubes. The diameters of SWNTs are in the range of 0.8 – 2 nm, and those of MWNTs are in the range of 5 – 100 nm[26]. Multi-walled carbon nanotubes show metallic character, while single-walled carbon nanotubes show metallic or semiconducting character depending on the chiral angle between the hexagon and the tube axis. Carbon nanotubes have excellent electrical properties, such as field effect transistors or chemiresistive sensors for the detection of VOCs in exhaled air[80]. Carbon nanotubes also have excellent mechanical properties, making them ideal materials for flexible gas sensors[27].
Graphene (G) is a single-atom carbon layer structure with hexagonal honeycomb lattice arrangement, which shows excellent gas sensing properties due to its high electron mobility and high specific surface area[28]. Chen et al. Discovered the gas-sensing behavior of graphene and pointed out that the local carrier concentration of graphene can be adjusted by gas molecules, which leads to the change of sensor conductivity[29]. However, the value of the carrier change due to the interaction with gas molecules is smaller than the number of background carriers of the metal oxide, thereby making the change in resistance very small. Graphene Oxide (GO) is a derivative of Graphene, which has abundant functional groups (carboxyl, carbonyl, hydroxyl and epoxy) on its surface. These groups can form covalent bonds or hydrogen bonds with gas molecules, thus enhancing the adsorption of gas and accelerating the charge transfer between GO and gas. Choi et al. Demonstrated that the sensitivity of GO-based gas sensors depends on the functional group of GO[30]. reduced Graphene Oxide (reduced Graphene Oxide, rGO) is another graphene derivative prepared by further reduction of GO. Compared with GO, there are still some oxygen functional groups on the surface of rGO, and some new vacancies and defects are produced due to the reduction, which can also serve as active sites for gas adsorption[31]. In addition, the conductivity of rGO is higher than that of GO, which makes the sensor have excellent sensitivity to VOCs gas at lower power consumption[32]. In addition, gas sensing materials based on graphene and its derivatives usually require additional modifications to improve the adsorption and electronic properties, and the general methods include doping with various heteroatoms (nitrogen, sulfur, boron, silicon, etc.) and precious metals[33,34][35]. Shanmugasundaram et al. Developed a formaldehyde sensor with low detection limit, which was prepared by 3D rGO-doped SnO2 materials, and can accurately identify lung cancer patients and healthy individuals, and the detection limit of the sensor is 10 ppb, which is one of the more excellent sensors for human breath detection at present[36].

2.3 Type of gas sensor

According to the different sensing mechanism, gas sensors can be divided into four main types of sensors: resistive, capacitive, electrochemical and piezoelectric.

2.3.1 Resistance gas sensor

Resistive sensors are one of the most common types of gas sensors. They are characterized by easy preparation, simple working principle and small size, so they are suitable for large-scale production to prepare sensor arrays[37]. Resistive sensors are based on the change in conductivity caused by the presence of adsorbed gas molecules. Therefore, the response of the sensor depends on the interaction between the target gas and the surface, as well as the area available for adsorption and desorption. MOS-based resistive gas sensor is the most common in breath detection, which can be divided into n-type semiconductor and p-type semiconductor according to the semiconductor type of MOS material. The gas sensor with metal oxide semiconductor as the sensitive material has the advantages of small size, low cost and high responsiveness. As shown in Figure 2, the general response mechanism of MOS-based resistive acetone gas sensor is briefly introduced by taking the reducing gas acetone as an example. In air, due to the high electron affinity of oxygen molecules, they will be adsorbed on the surface of the sensor and get electrons from the sensing layer. Therefore, depending on the operating temperature, the surface of the gas sensor will have different oxygen ion species, as shown in the following equation
O2(g)→O2(ads)
O2(ads)+e-→ O 2 -(ads) (<150 ℃)
O 2 -(ads)+e-→2O- (150~400 ℃)
2O-+e-→ O 2 - (>150 ℃)
In general, molecules and ions are stable below 150 ° C, while other forms are stable at higher temperatures[38]. Depending on the n-type or p-type nature of the gas sensing material, an electron depletion layer or a hole accumulation layer will be created on the surface of the gas sensor, respectively. Since most of the charge carriers in the n-type sensor are electrons and the charge carriers in the p-type gas sensor are holes, the resistance of the n-type sensor will decrease relative to an air environment, while the resistance of the p-type gas sensor will increase relative to an air environment. When acetone gas is injected into the gas test chamber, it will react with the oxygen ions that have been adsorbed, and electrons will be released as follows:
(CH3COCH3)g→(CH3COCH3)ads
(CH3COCH3)ads+8O-→3CO2+3H2O+8e-
In addition to the common MOS-based resistive gas sensors, gas sensors based on MOS and graphene and their derivatives have also been developed for exhalation. S Sánchez-Vicente et al. Developed an electronic nose based on graphene-doped TiO2, which detects ethanol, acetone, carbon monoxide and nitric oxide, and detects diseases such as Staphylococcus aureus infection, diabetes, asthma, chronic obstructive pulmonary disease and cystic fibrosis through breath[39].

2.3.2 Capacitive gas sensor

Capacitive gas sensor is composed of sensing layer, substrate, heater and electrode, and its working principle is the change of dielectric properties of sensing layer under different environmental conditions. The simplest capacitive structure consists of a dielectric material between parallel plates. The capacitance represents the ability of these two parallel plates to hold this charge. The working mechanism of a capacitive sensor is shown in Figure 3. The sensing surface of the probe is usually a charged plate, and the target is another plate. The front end of the sensor continuously changes the voltage on the sensing surface and determines the capacitance between the probe and the target[40]. Capacitive sensors also have some applications in breath detection. For example, Bahoumina et al. Used carbon composite polymer as a sensitive material to ink-jet print capacitive microwave sensors on paper[41]. The sensor was characterized for ethanol vapor and showed a sensitivity of about -2.5 kHz/ppm from 0 to 2000 ppm, which can be used to detect the amount of ethanol in human breath.

2.3.3 Electrochemical sensor

Electrochemical sensors typically consist of a sensing electrode and a counter electrode separated by a thin layer of electrolyte[37]. The detection of the analyte takes place at an appropriate electrode where a chemical reaction (oxidation or reduction) has taken place. The reacting sensing electrode is characterized by a high surface to volume ratio (for signal enhancement) and consists of a catalytic material, such as platinum, palladium, or carbon-coated metal, specific for the desired analyte[42]. The analyte-electrode reaction generates a sufficient electrical signal, which is measured relative to the counter electrode.
Cross-reacting electrochemical sensor based on polymer and nanomaterials (E. G., MNP, CNT, graphene derivatives, and combinations thereof) can be use for that diagnosis of diabetes, lung cancer[43]. Maier et al. Used Prussian Blue electrocatalyst to modify carbon electrode to develop a wearable paper-based sensor for H2O2 detection of lung disease biomarkers[44]. Nonanal detection was achieved using sensors fabricated from materials decorated with SnO2 nanoparticles, SnO2 nanosheets, as well as noble metal catalysts (Pt, Au, and Pd), aiming to achieve early diagnosis of lung cancer[45]. Au nanoparticle-modified MoS2 nanosheets were used for oxygen-based detection of VOCs, such as the diabetes biomarker acetone, and the responsiveness and selectivity of the sensor were improved due to the electronic contribution from the Au nanoparticles to the MoS2, as shown in Figure 4 (a) using MoS2 hydrocarbons, resulting in a small signal alternation due to dipole scattering through the electron cloud of the -CH2- group[46]; (B) Oxygen-group VOCs using MoS2, which donate electrons to MoS2 (left), are oxidized by adsorbed oxygen species, releasing electrons to MoS2 (right). (C) oxy-VOCs using MoS2 decorated with Au NPs, which increases the electron concentration and thus the adsorption of oxygen species on MoS2; (D) returns more electrons upon interaction with oxy-VOCs.

2.3.4 Piezoelectric transducer

Piezoelectric sensors are sensitive to mechanical stress and are often used as mass-sensitive sensors. Acoustic wave devices are used for piezoelectric sensors, also known as mass, gravity, or microbalance sensors. An oscillating circuit is used to generate sound waves to cause the piezoelectric crystal to resonate. The most important piezoelectric gas sensor categories are Quartz Crystal Microbalance (QCM) and Surface Acoustic Wave (SAW).
QCM sensors possess quartz crystal resonators functionalized with different appropriate sensing elements such as conducting polymers, MOS or nanomaterials[47]. The acoustic wave propagates in the crystal either perpendicular or parallel to the surface. As the gas is absorbed at the sensitive surface of the crystal, the mass changes, alternating resonant frequencies[48]. Typically, an increase in mass results in a decrease in the oscillation frequency of the resonator[42]. Fig. 5A shows the sensing mechanism of the QCM sensor, the oscillation of the quartz crystal substrate and the propagation of a transverse acoustic wave on the quartz substrate caused by an alternating electric field applied to the electrodes[49]; The interaction of the sensing layer with the gas changes the mass on the substrate, which changes the wave amplitude and velocity, resulting in a shift of the resonance frequency. QCM sensors are widely used for selective gas detection and may be suitable for breath analysis. The sensor array composed of three sensors prepared by Tao et al. Using MWCNTs sensitive materials modified by TiO2 nanoparticles has excellent selectivity for acetone, which can be used for screening and monitoring diabetes mellitus[50].
In SAW gas sensors, the sound wave propagates only parallel to the surface of the piezoelectric crystal, penetrating about one sound wave length at the depth of the crystal[37]. The crystal surface is decorated with a chemically selective layer, and a change in the mass or conductivity of the chemical interface affects the propagation frequency in the analyte[51]. As a result, the propagation frequency of the SAW changes, which can then be measured. The sensing mechanism is introduced in Figure 5B, where the surface wave is confined to one acoustic wavelength on the surface of the piezoelectric material and is caused by an input RF voltage applied to the plug-in transmitter. When the SAW reaches the receiving plug-in transmitter, the mechanical energy is converted to radio frequency as an output[49]. The adsorption of the analyte on the piezoelectric coating causes a mass change, which eventually leads to a change in frequency.

3 Application of Sensors in Different Diseases

Mass spectrometry has obvious advantages in quantitative and qualitative analysis, but its shortcomings are also obvious, such as large equipment volume, complex operation technology, time-consuming analysis, and unable to provide real-time results[52]. Therefore, mass spectrometry-based detection instruments are not yet available for large-scale clinical applications and personalized medical monitoring. Gas sensors for exhaled breath analysis show potential as clinical diagnostic tools because such sensors are low cost, small in size, simple to operate, and capable of providing detection results in real time. The detection principle of the exhalation sensor is divided into two types: (1) non-target strategy and (2) target strategy. The former is designed to assess respiratory patterns without identifying potential respiratory biomarkers. The latter aims to identify different biomarkers of exhalation using highly sensitive gas sensors or sensor arrays. In addition, the gas sensor used for breath analysis must be able to have the basic working conditions of high sensitivity, high selectivity and detection of trace-level VOCs concentration.
Changes in the concentration of one or more VOCs in human exhaled breath are associated with specific diseases, and these VOCs are called exhaled biomarkers of disease[53]. Although more than 3000 kinds of VOCs have been found in human exhalation, only dozens of VOCs can be used as biomarkers of exhalation, as shown in Figure 6. This article briefly introduces several common diseases and their associated VOCs gases: increased concentrations of carbon monoxide and nitric oxide are related to asthma[54]; The increase of acetaldehyde and hydrocarbon concentration is related to lung disease. The increase of ketones (acetone as the main component) is related to ketonemia. Increased ammonia concentration is associated with gastric disease and renal dysfunction. The increase of hydrogen concentration is related to digestive system diseases. Elevated methane concentrations may be associated with intestinal disease.

3.1 Expiratory biomarkers in different diseases

3.1.1 Chronic kidney disease

Chronic kidney disease (CKD) has become a worldwide public health and health problem. According to statistical data, CKD has become the third disease with increasing mortality in the world, which seriously endangers human health[55]. At present, there is no specific drug or medical treatment to effectively cure chronic kidney disease. With the development of the disease, the renal function of patients will gradually decrease, and even develop into renal failure, uremia and so on, which will bring a heavy burden to patients and their families. It is worth noting that there are no obvious symptoms in the early stage of chronic kidney disease, and the symptoms are not obvious until the middle and late stages, when the treatment effect is far less than that in the early stage[56]. Therefore, it is necessary to carry out early screening and early intervention treatment for chronic kidney disease, so as to reduce the burden of patients and prolong their life. In recent years, by exploring the relationship between expiratory biomarkers and chronic kidney disease, researchers have devoted themselves to applying this non-invasive, cost-effective method to the early screening of chronic kidney disease. The diagnosis of chronic kidney disease based on breath is based on the fact that the ammonia concentration in the breath of healthy people ranges from 200 to 1750 ppb, and the ammonia concentration in the breath of people with kidney disease ranges from 820 to 14,700 ppb[57][58]. Narasimhan et al. tested the changes of ammonia in the breath of 7 patients with end-stage renal disease during hemodialysis. The ammonia content in the breath of the patients at the beginning of dialysis was between 1500 and 2000 ppb[59]. After 15 min and 30 min of dialysis, the ammonia content in the breath decreased sharply. The ammonia content in the breath was slowly reduced to 150-200 ppb until the end of dialysis. The blood urea nitrogen and creatinine in the blood of patients during dialysis were analyzed, and the data comparison showed that the blood indicators of chronic kidney disease were highly correlated with the ammonia concentration in the breath, which indicated that the ammonia concentration in the breath could effectively reflect the health status of the human kidney.

3.1.2 Helicobacter pylori

Gastric cancer and gastric ulcer are common diseases at present, among which gastric cancer is an important public health problem facing mankind, listed as the fifth most common malignant tumor and the fourth leading cause of cancer-related death in the world, posing a serious threat to human health[60]. Helicobacter pylori is an important cause of gastric cancer and gastric ulcer, with a worldwide infection rate of about 50%. The World Health Organization classifies Helicobacter pylori as the first carcinogen, so it is very important to screen for Helicobacter pylori[61][62]. The methods of detecting Helicobacter pylori include endoscopy, urease rapid test, stool antigen test and serum test. The first two methods are invasive and may cause the risk of cross-infection. The latter two methods are non-invasive but inconvenient to sample[63][64][65][66]. Therefore, simple, rapid, risk-free and non-invasive testing methods have become the preferred choice for people. After continuous exploration and research, breath detection technology has become an important means of early screening of Helicobacter pylori. At present, the breath test of Helicobacter pylori requires the infected person to take urea containing 13C/14C calibration elements, as shown in Figure 7. Helicobacter pylori produces highly active urease in the stomach to decompose urea into carbon dioxide and ammonia, which are absorbed through the mucus layer of the stomach, transported to the lungs with the blood, and excreted in the form of exhalation through the blood-lung barrier. However, the detection of 13C is relatively expensive, and the 14C is radioactive, which will cause potential harm to the human body. Therefore, ammonia, another VOCs associated with Helicobacter pylori, has become the preferred choice for the detection of Helicobacter pylori. The concentration of ammonia in the breath of Helicobacter pylori infected patients before urea intake (without calibration elements) was significantly lower than that of healthy people, and the concentration of ammonia in the breath of Helicobacter pylori infected patients after urea intake was 8 times higher than that of healthy people[67]. Kearney et al. Used a gas sensor to test the breath of 13 subjects. After the subjects took the urea capsule, the gas sensor was used to test the breath of the subjects. The results showed that the ammonia content in the breath of Helicobacter pylori infected persons after taking urea was significantly higher than that of healthy people[67].

3.1.3 Diabetes

Diabetes mellitus is a chronic disease caused by metabolic disorders that lead to excessive glucose levels in the body[68]. Imbalanced blood glucose levels may trigger various health complications such as kidney failure, heart disease, retinopathy, neuropathic neuropathy, and pressure imbalance in the lower extremities, and even premature death[69]. The total number of people with diabetes worldwide has reached 463 million, with nearly 4.2 million deaths[70]. Early screening is essential in order to effectively treat diabetes. The breath analysis technology developed in recent years has the advantages of non-invasive, convenient sampling and high accuracy, which provides the possibility for early screening of diabetes mellitus. Deng et al. Used gas chromatography-mass spectrometry to analyze the breath of diabetic patients and healthy people, and the results showed that the concentration of acetone in the breath of healthy people was less than 0.76 ppm, while the concentration of acetone in the breath of diabetic patients was higher than 1.71 ppm, which indicated that the high concentration of aceton in human breath could be used as a basis for the diagnosis of diabetes[71]. Righettoni et al. Developed a portable acetone sensor and analyzed the breath test of five different diabetic patients. The result of the sensor test showed that the acetone content in the breath of diabetic patients was higher than that of normal people, which further verified that the high concentration of acetone in human breath was closely related to diabetes[72].

3.1.4 Respiratory disease

a。 Asthma
Asthma, also known as bronchial Asthma, is a chronic airway inflammatory disease involving a variety of cells and their components. This inflammation is usually accompanied by increased airway reactivity, resulting in recurrent symptoms such as wheezing, chest tightness, and coughing[73]. The incidence of asthma in children is also high, and the prevalence rate of asthma in children in the world is as high as 11.6% to 13.7%, which is seriously endangering the health of infants and young children. Early treatment can effectively improve symptoms and reduce the risk of deterioration, so early screening for asthma is very necessary[74]. Bronchoalveolar lavage and bronchial biopsy are the traditional methods for the diagnosis of asthma, but these methods are invasive and not suitable for children, especially those under 4 years old, and are also difficult to meet the purpose of early screening[75]. Breath detection is suitable for any age group and is also a new non-invasive detection method. Fractional exhaled nitric oxide (FeNO) is the only exhaled biomarker used in the clinical routine of asthma at present. The concentration of NO in the breath of healthy individuals is 6.7 ~ 51.1 ppb, and the concentration of NO in the breath of asthmatic patients can increase to hundreds of ppb. Therefore, the increase of NO concentration in the breath is closely related to asthma, which can be used as the basis for the diagnosis of asthma[76][77]. For example, Abdel-Aziz et al. Used sensors to conduct breath tests on healthy children and children with asthma, healthy adults and adults with asthma (a total of 655 people), and established two different training sets and validation sets, respectively[78]. The machine learning model utilizing the VOC curve distinguished between specific and non-specific participants, and the areas under the receiver operating characteristic curves were 0.84 and 0.72 in the training and validation sets, respectively, which also indicated that the sensor technology could be used to discriminate and accurately classify asthma patients of different ages.
b。 Chronic obstructive pulmonary disease
Chronic obstructive pulmonary disease (COPD) is a common disease characterized by persistent airflow limitation. Its main clinical manifestations are Chronic cough, shortness of breath, chest tightness or wheezing[79]. The World Health Organization's latest projections of mortality and causes of death show that the prevalence of COPD will continue to rise in the next 40 years as smoking rates rise in developing countries and population aging intensifies in high-income countries.It is predicted that by 2060, more than 5.4 million people will die of COPD and related diseases every year, so early screening and intervention for COPD is essential. Pulmonary function test, imaging examination, blood gas analysis, radionuclide pulmonary ventilation imaging and other commonly used examination methods in clinical medical treatment can only confirm whether there is COPD, but not suitable for early screening of COPD[80][81][82][83]. Breath test is a non-invasive, simple and low-cost prospective test, which provides the possibility of early screening for COPD. Jare Jareño-Esteban et al analyzed the breath of 57 COPD patients and 100 healthy subjects by gas chromatography-mass spectrometry for five VOCs, namely hexanal, heptanal, nonanal, propionic acid and nonanoic acid.The results showed that there was a significant difference in the concentration of hexanal in the breath between COPD patients and healthy subjects (no smoking history and smoking cessation, excluding smokers), so hexanal can be used as one of the important breath biomarkers of COPD[84]. Rodr Rodríguez-Aguilar et al. Developed an electronic nose system to detect VOCs in the exhaled breath of patients with COPD[85]; The breath of 23 COPD patients and 33 healthy subjects was detected, and the breath of each subject was analyzed by electronic nose, and the data was processed by intelligent algorithm. The results showed that the concentration of VOCs in the breath of COPD patients was significantly higher than that of healthy subjects. Similarly, they used an electronic nose test system composed of gas sensors to test the breath of 50 healthy subjects, 50 COPD patients and 30 lung cancer patients, and also used a variety of VOCs as biomarkers to distinguish healthy subjects from COPD patients with 100% accuracy[86]. It can be seen that the detection of COPD by exhalation is a non-invasive detection method with broad application prospects, but a single exhalation biomarker can not be used as the detection object of COPD, and a variety of VOCs gases need to be cross-validated to achieve better results. Table 1 shows the analysis methods and results of researchers using different expiratory biomarkers to detect COPD.
c。 Lung cancer
According to the Global Cancer Report 2021, lung cancer is the cancer with the highest incidence and mortality in the world, with 2.1 million new cases and 1.8 million deaths[90]. The traditional detection method is to diagnose lung cancer by Computer Tomography (CT) or lung biopsy with blood auxiliary examination[91][92][93]. However, there are two difficulties in the medical diagnosis of lung cancer, one is that there are no obvious symptoms in the early stage of the disease, people often ignore the condition and miss the best treatment period; On the other hand, the symptoms of the body are in the middle and late stages, and the choice of treatment methods is limited. Therefore, traditional detection techniques are not suitable for early screening of lung cancer. Breath detection, developed in recent years, can achieve early screening of lung cancer by analyzing lung cancer biomarkers in human breath. Binson et al. Developed an electronic nose system containing five metal oxide semiconductor gas sensors to analyze the breath of 93 healthy subjects and 51 patients with lung cancer, with a classification accuracy of 79.31%, an overall sensitivity of 70.00%, and a specificity of 84.21%[94]. Increased concentrations of VOCs in human breath, such as 1-butanol, octahydrofuran, 2-pentanone, and hexane, have been found to be highly associated with lung cancer[7][95][96]. Therefore, non-invasive breath analysis technology can be used to achieve early screening and early intervention of patients, which is expected to provide more treatment options and prolong the life of patients.

3.2 Application of electronic nose technology in disease diagnosis

In recent years, with the rapid development of computer technology and the increasing research on sensor array technology and information fusion technology, the research on Electronic nose (E-nose) technology has gradually become one of the research hotspots in the field of engineering, and many countries have listed the research on this kind of biological olfactory mechanism and its functional imitation as a priority development topic[97]. E-nose can identify both simple and complex odors, and has the characteristics of low cost, high sensitivity, high accuracy, fast detection speed, small data error and good repeatability[98]. This technology has great potential for breath detection and could be used in the future as a real-time non-invasive diagnostic tool for clinical practice and personalized medical monitoring[99]. At present, a variety of E-noses for breath detection have been developed in clinical studies, which have achieved good results in distinguishing patients with diseases from healthy subjects.For example, Marom et al. Used a sensor array composed of three sensors to analyze the breath of patients with chronic kidney disease in different stages, and recruited 20 healthy subjects, 6 patients in stage 2, 16 patients in stage 3, 12 patients in stage 4 and 8 patients in stage 5[100]. In the breath test of stage 2 and stage 3 patients and healthy subjects, the discrimination accuracy was 79%. In the breath test of stage 4 and stage 5 patients, the accuracy of discrimination was 85%. The accuracy of breath test in stage 2, 3, 4 and 5 patients was 76%. By distinguishing patients in different stages, a better classification result is obtained, which provides an important technical guarantee for the application of breath detection and a solid foundation for the future application of breath detection in medical treatment. Shnayder et al. Used the electronic nose system to detect the breath of 26 patients with Helicobacter pylori and several healthy people, and collected the breath of volunteers after oral administration of 500 mg urea for 5 minutes.The test results showed that the sensitivity of the electronic nose to the signal generated by the breath of healthy people and patients infected with Helicobacter pylori was nearly 10 times different, which indicated that the electronic nose could effectively distinguish Helicobacter pylori infection[101]. Sarno et al. Proposed an electronic nose composed of an electrochemical sensor array to detect the breath of 20 diabetic patients and 20 healthy people. The collected breath samples were tested and classified by K-nearest neighbor algorithm. The classification accuracy, sensitivity and specificity of diabetes and healthy people were 95%, 91.3% and 94.12%, respectively[102]. The concepts of accuracy, sensitivity and specificity are calculated according to the following formulas:
A c c u r a c y = T P + T N T P + F P + T N + F N
S e n s i t i v i t y = T P T P + F N
S p e c i f i c i t y = T N T N + F P
TP, FP, TN, and FN are true positive, false positive, true negative, and false negative, respectively.
Tenero et al. Used the electronic nose system to detect the breath of 28 asthmatic children and 10 healthy children, and analyzed the breath data by combining the principal component analysis algorithm and the penalized logistic regression model. The results showed that the area under the curve of the electronic nose was 0.85, the sensitivity was 0.79, and the specificity was 0.84, indicating that the electronic nose system could effectively distinguish asthmatic patients[103]. van Bragt et al. Used the electronic nose system to analyze the breath of 364 patients with chronic obstructive pulmonary disease. 254 breath samples were collected as training data, and the remaining 110 samples were used as verification data[104]. The pattern recognition algorithm combined with principal component analysis was used to distinguish the patients whose condition was aggravated in the past three months from those whose condition was not aggravated, and the accuracy rate reached more than 90% in both data training and data verification stages. Chen et al. Proposed a self-developed electronic nose system for non-invasive diagnosis of lung cancer. 134 healthy people and 101 lung cancer patients were tested for breath, and 235 breath samples were collected for testing and data analysis combined with principal component analysis and extreme gradient lifting algorithm[105]. The complete test process is shown in Figure 8. First, the breath sample is collected (Figure 8 A), then the breath sample is tested in the sensor array module and the data is collected (Figure 8 B), and finally the pattern recognition algorithm is used for data analysis. The red and green dots represent cancer patients and healthy people (Figure 8 C). The accuracy, sensitivity and specificity were 93.59%, 95.60% and 91.09%, respectively. The electronic nose system based on metal oxide semiconductor designed by Binson et al. Breath tests were performed on 32 patients with lung cancer, 38 patients with chronic obstructive pulmonary disease and 72 healthy subjects.In the detection of lung cancer patients, the accuracy, sensitivity and specificity of the pattern recognition algorithm were 91. 3%, 84.4% and 94. 4%, respectively, while in the discrimination of chronic obstructive pulmonary disease patients, the accuracy, sensitivity and specificity of the support vector machine algorithm were 90. 9%, 81.6% and 95. 8%, respectively[7].
As a low-cost and non-invasive diagnostic tool, the analysis of human exhalation by electronic nose system is a promising method, but there are some problems. Although the electronic nose system has achieved good results in breathing tests for some diseases, it is also essential to improve the characteristics of the electronic nose system, such as stability, repeatability, and to expand its relevance.

4 Acquisition and processing of sensor response data

4.1 Exhalation sampling

Breath sampling is also very important in breath detection, which is generally divided into direct breath sampling and indirect breath sampling. Direct breath sampling is to blow the breath of the subject directly to the breath detection device[106]; Indirect breath sampling is to collect the breath into the air bag for testing, which is also a common collection method in the field of breath detection[107,108]. The air bags used are made of specific materials, such as polyvinyl fluoride, fluorinated ethylene propylene, Teflon, etc. Teflon is the most commonly used air bag material in breath sampling studies. In order to eliminate the interference of impurity gas, the gas bag should be flushed with N2 before gas collection[109]. Because of the complex composition of human exhalation, the collection of exhalation should be standardized as far as possible, otherwise it will bring greater errors to the test. The concentration of endogenous VOCs is higher at the end of expiration, and the concentration of exogenous VOCs is higher at the beginning of expiration. Therefore, when the indirect method is used to collect exhalation, the exhalation should be collected within the appropriate expiration time to avoid large deviation[110]. In addition, due to the high humidity in human exhalation, water vapor may condense on the inner wall of the airbag to absorb the target gas during indirect breath sampling, resulting in poor results, so the airbag is usually heated at 40 ℃ for 1 H to eliminate the problem of water vapor condensation as far as possible.
It should be noted that the breath sample also has a certain shelf life, so the collected breath should be tested as soon as possible, and it should not be stored for a long time to avoid deterioration. Beauchamp et al. Compared the stability of VOCs after storing breath samples in Tedlar bags for 10 and 70 H[111]. The results showed that 80% of the compounds in the sample remained stable within 10 H, but within 70 H, the compounds in the sample changed greatly, resulting in a large deviation in the breath test. Try not to eat food with pungent odor a few hours before collecting exhaled breath, and rinse your mouth with purified water before collecting exhaled breath[112]. In addition, try not to do strenuous exercise before gas collection, and try to ensure smooth exhalation to avoid excessive errors in data collection[113].

4.2 Eigenvalue extraction

A commonly used feature extraction method is to use the raw data obtained from the sensors to extract the feature values. To improve the classification accuracy of an exhalation analysis system, feature extraction algorithms are often used before training a selected classification or regression model[98,114,115]. Common methods for extracting features from raw sensor data include maximum response, area under the sensor curve, maximum derivative, and extreme response. A more advanced feature extraction and reduction technique is the singular value decomposition algorithm, which decomposes the input matrix into orthogonal eigenvectors and eigenvalues.
In order to remove redundant features and retain the most useful information, it is also a good choice to apply feature selection algorithms, such as principal component analysis[94]. The sensor array contains a data set of high-dimensional responses of multiple different gas sensors, and the algorithm projects the high-dimensional data to a low-dimensional data subspace, so it is simple and reliable to use the algorithm to reduce the dimensionality and extract low-dimensional features[116,117]. Another way to reduce the number of unimportant features is statistical methods, which can eliminate the effect of human test error by applying manual prescreening methods. Reducing the number of redundant features can improve model accuracy and training speed and prevent overfitting by reducing the computational complexity of the algorithm[88].

4.3 Pattern recognition algorithm

4.3.1 Principal component analysis

Principle component analysis (PCA) is another unsupervised linear technique that transforms multidimensional reduced data into a lower number of orthogonal variables called principal components, which can be represented by a limited number of variables (e.g. two or three). PCA has also proven to be a powerful and effective algorithm for clearly visualizing the most important information contained in large data sets. Saidi et al. Analyzed breath samples from 16 chronic kidney disease (CKD) patients, 6 diabetes mellitus (DM) patients and 22 healthy subjects by PCA algorithm, and the identification results are shown in Fig. 9a. The three-dimensional PCA plot shows a good classification of exhalations corresponding to CKD, DM and healthy subjects[117]. The first and third discriminant components (PC1, PC3) clearly distinguished various respiratory States, while the second discriminant component (PC2) clearly distinguished exhalation in healthy subjects and patients with CKD and DM; After one month, 8 new volunteers were recruited again, and 48 breath samples were collected for classification and identification. The analysis results are shown in Figure 9b, and all the breath samples were correctly identified.

4.3.2 Artificial neural network

Artificial Neural Networks (ANN) is a machine learning algorithm, which is similar to the biological nervous system and consists of hierarchically arranged nodes (neurons), through which the information vector is converted into some output signals. Its simple structure is shown in Figure 10, which consists of an input layer, a hidden layer and an output layer[98]. It is inspired by the structure of the biological nervous system[113,118]. Multilayer feedforward neural network is a commonly used algorithm in ANN, which consists of a group of interconnected neurons. The connections between units move forward from the input layer to the output layer only through the hidden layer. The number of hidden layers of ANN changes according to the amount of tasks to be completed[118,119]. ANNs constitute a "machine learning framework" in order to process data and come up with highly accurate results. Marzorati et al. Used the electronic nose system to identify patients with early lung cancer and analyzed the breath of 10 healthy people and 6 patients with lung cancer, and used the ANN algorithm to classify them. After cross-validation, the identification sensitivity, specificity and accuracy of ANN were 85.7%, 100% and 93.8%, respectively[120].

4.3.3 Support vector machine

Support vector machines (SVM) is one of the most commonly used classification algorithms and a supervised learning model for small but high-dimensional data sets[121]. The algorithm uses a hyperplane to divide an m-dimensional input data set into an l-dimensional feature space, which can be used to solve classification and regression problems, and plays an important role in respiratory analysis[122,123]. One of the main features of SVM is the need to select a suitable kernel that will fit the input data well and reduce the training time[109]. The use of SVM algorithm in the field of breath detection has achieved good classification results, such as high sensitivity and specificity of more than 90% in distinguishing healthy and diabetic VOCs samples. Yan et al. Used an SVM algorithm to distinguish the exhalation samples of healthy and diabetic patients, which was trained on 140 samples randomly selected from patients and the same number of healthy samples[123]. The remaining samples (139 per class) were used to validate the model; The SVM algorithm was used for 50 times of training, and the results showed that the average sensitivity of diabetes screening was 91. 51%, and the specificity was 90. 77%.

4.3.4 K-nearest neighbor algorithm

The k-nearest neighbors (k-NN) algorithm is a supervised learning algorithm that classifies a given unknown sample into a class based on the distance between the K closest examples in the training dataset. In the k-NN classification algorithm, a new observation is assigned to a class in the following way: a new observation X is placed in the data space, a particular k-NN is determined according to the selected distance measure, and then the data is predicted based on the majority of classes represented by the selected k-NN. The class of new observations can be selected not only based on the majority class, but also by a weighting method[124]. Gou et al. Used the breath samples of 57 diabetic patients and 48 healthy people, and used the K-NN algorithm to distinguish the breath samples of healthy people and patients[125]. The sensitivity and specificity of the analysis were 87.67% and 86.87%, respectively. Sarno et al. Have shown that the results of K-NN algorithm are more accurate and easier to implement than SVM and neural network models[102]. The k-NN classifier performed the fastest classification in the network system designed by the researchers.

5 Conclusion and prospect

Breath analysis is a non-invasive, unlimited sampling analysis technique, and it does not burden the subject, so it is a very promising clinical medical testing method. In recent years, sensor technology based on breath analysis has made great progress, and although the technology continues to improve, it is still a long way from becoming the mainstream of diagnosing diseases. Most of today's applications are proof-of-principle and case-control studies for a small number of patients. At present, there are still many shortcomings: (1) lack of test standardization; (2) The identification of expiratory biomarkers is complicated; (3) Some diseases cannot be diagnosed by a single biomarker; (4) Lack of reliable data analysis methods. To overcome these limitations, breath analysis studies should focus on standardization of breath sampling and detection and systematic improvement of technical/physiological/pathophysiological hybrids to identify endogenous VOCs and determine effective exhalation patterns of biomarkers and establish standard correlations between VOCs concentrations in blood and breath. In addition, the development of portable, low-cost nanomaterial sensors that meet clinical needs and the optimization of sensor training and validation by using different subject groups are the key to the development of sensors suitable for clinical diagnosis. To achieve all of the above, interdisciplinary research and collaboration are essential prerequisites.

References

Publishing order | Descend order by publishing year | Descend order by cited within
[1]

Francesco F D, Fuoco R, Trivella M G, Ceccarini A. Microchem. J., 2005, 79(1/2): 405.
[2]

de Lacy Costello B, Amann A, Al-Kateb H, Flynn C, Filipiak W, Khalid T, Osborne D, Ratcliffe N M. J. Breath Res., 2014, 8(1): 014001.
[3]

Nakhleh M K, Amal H, Jeries R, Broza Y Y, Aboud M, Gharra A, Ivgi H, Khatib S, Badarneh S, Har-Shai L. ACS nano, 2017, 11: 112.
[4]

Wilkinson M, White I, Hamshere K, Holz O, Schuchardt S, Bellagambi F G, Lomonaco T, Biagini D, Di F F, Fowler S J. J. Breath Res., 2021, 15(2): 026006.
[5]

Li B Z, Zou X, Wang H M, Lu Y, Shen C Y, Chu Y N. Anal. Biochem., 2019, 581: 113344.
[6]

Stefanuto P H, Zanella D, Vercammen J, Henket M, Schleich F, Louis R, Focant J F. Sci. Rep., 2020, 10: 16159.
[7]

Song G, Qin T, Liu H, Xu G B, Pan Y Y, Xiong F X, Gu K S, Sun G P, Chen Z D. Lung Cancer, 2010, 67(2): 227.
[8]

Tai H L, Wang S, Duan Z H, Jiang Y D. Sens. Actuat. B Chem., 2020, 318: 128104.
[9]

Ratiu I A, Ligor T, Bocos-Bintintan V, Mayhew C A, Buszewski B. J. Clin. Med., 2020, 10(1): 32.
[10]

Neerincx A H, Vijverberg S J H, Bos L D J, Brinkman P, van der Schee M P, de Vries R, Sterk P J, Maitland-van der Zee A H M V. Pediatr. Pulmonol., 2017, 52(12): 1616.
[11]

van der Schee M P, Fens N, Brinkman P, Bos L J, Angelo M D, Nijsen T E, Raabe R, Knobel H H, Vink T J, Sterk P J. J. Breath Res., 2012, 7(1): 016002.
[12]

Behera B, Joshi R, Anil Vishnu G K, Bhalerao S, Pandya H J. J. Breath Res., 2019, 13(2): 024001.
[13]

Shehada N, Cancilla J C, Torrecilla J S, Pariente E S, Brönstrup G, Christiansen S, Johnson D W, Leja M, Davies M P A, Liran O, Peled N, Haick H. ACS Nano, 2016, 10(7): 7047.
[14]

Yang D, Gopal R A, Lkhagvaa T, Choi D. Meas. Sci. Technol., 2021, 32(10): 102004.
[15]

Hanh N H, Ngoc T M, van Duy L, Hung C M, van Duy N, Hoa N D. Sens. Actuat. B Chem., 2021, 343: 130147.
[16]

Choi H J, Chung J H, Yoon J W, Lee J H. Sens. Actuat. B Chem., 2021, 338: 129823.
[17]

Alev O, Arslan L, Buyukkose S, Ozturk Z. SENSORS AND ELECTRONIC INSTRUMENTATION ADVANCES ( SEIA' 2018), Turkey, INT FREQUENCY SENSOR ASSOC-IFSA, 2018, 77.
[18]

Selvaraj B, Rayappan J B B, Babu K J. J. Alloys Compd., 2022, 914: 165224.
[19]

Dai J, Ogbeide O, MacAdam N, Sun Q, Yu W B, Li Y, Su B L, Hasan T, Huang X, Huang W. Chem. Soc. Rev., 2020, 49(6): 1756.
[20]

Khatib M, Haick H. ACS Nano, 2022, 16(5): 7080.
[21]

Zhang D Z, Wu Z L, Zong X Q. Sens. Actuat. B Chem., 2019, 289: 32.
[22]

Shahmoradi A, Hosseini A, Akbarinejad A, Alizadeh N. Anal. Chem., 2021, 93(17): 6706.
[23]

Reiner-Rozman C, Pichler B, Madi V, Weißenböck P, Hegedüs T, Aspermair P, Bintinger J. Sensors, 2022, 22(14): 5379.
[24]

Kirkpatrick S. Rev. Mod. Phys., 1973, 45(4): 574.
[25]

Llobet E. Sens. Actuat. B Chem., 2013, 179: 32.
[26]

Rana M M, Ibrahim D S, Mohd Asyraf M R, Jarin S, Tomal A. Sens. Rev., 2017, 37(2): 127.
[27]

Yu M F, Lourie O, Dyer M J, Moloni K, Kelly T F, Ruoff R S. Science, 2000, 287(5453): 637.
[28]

Guo J, Ruan B X, Zhu J Q, Dai X Y, Xiang Y J, Zhang H. J. Phys. D: Appl. Phys., 2017, 50(43): 434003.
[29]

Chen X, Wang Y, Xiang Y J, Jiang G B, Wang L L, Bao Q L, Zhang H, Liu Y, Wen S C, Fan D Y. J. Light. Technol., 2016, 34(21): 4948.
[30]

Choi Y R, Yoon Y G, Choi K S, Kang J H, Shim Y S, Kim Y H, Chang H J, Lee J H, Park C R, Kim S Y, Jang H W. Carbon, 2015, 91: 178.
[31]

Dua V, Surwade S, Ammu S, Agnihotra S, Jain S, Roberts K, Park S, Ruoff R, Manohar S. Angew. Chem. Int. Ed., 2010, 49(12): 2154.
[32]

Deng S Z, Tjoa V, Fan H M, Tan H R, Sayle D C, Olivo M, Mhaisalkar S, Wei J, Sow C H. J. Am. Chem. Soc., 2012, 134(10): 4905.
[33]

Yan W Y, Zhou Q, Chen X, Huang X J, Wu Y C. Sens. Actuat. B Chem., 2016, 230: 761.
[34]

Chen Y, Gao B, Zhao J X, Cai Q H, Fu H G. J. Mol. Modeling, 2012, 18(5): 2043.
[35]

GutÉs A, Hsia B, Sussman A, Mickelson W, Zettl A, Carraro C, Maboudian R. Nanoscale, 2012, 4(2): 438.
[36]

Shanmugasundaram A, V Manorama S, Kim D S, Jeong Y J, Weon Lee D. Chem. Eng. J., 2022, 448: 137736.
[37]

Haick H. J. Phys. D: Appl. Phys., 2007, 40(23): 7173.
[38]

Kim H J, Lee J H. Sens. Actuat. B Chem., 2014, 192: 607.
[39]

Sánchez-Vicente C, Santos J P, Lozano J, Sayago I, Sanjurjo J L, Azabal A, Ruiz-Valdepeñas S. Sensors, 2020, 20(24): 7223.
[40]

Rivadeneyra A, LÓpez-Villanueva J A. Micromachines, 2020, 11(4): 367.
[41]

Bahoumina P, Hallil H, Lachaud J L, AbdelghaniA, Frigui K, Bila S, Baillargeat D, Zhang Q, Coquet P, Paragua C. Proceedings, 2017, 1: 439.
[42]

Fraden J.. Handbook of Modern Sensors. New York, NY: Springer New York, 2010. 569.
[43]

Obermeier J, Trefz P, Wex K, Sabel B, Schubert J K, Miekisch W. J. Breath Res., 2015, 9(1): 016008.
[44]

Maier D, Laubender E, Basavanna A, Schumann S, Güder F, Urban G A, Dincer C. ACS Sens., 2019, 4(11): 2945.
[45]

Masuda Y, Itoh T, Shin W, Kato K. Sci. Rep., 2015, 5: 10122.
[46]

Chen W Y, Yen C C, Xue S C, Wang H Y, Stanciu L A. ACS Appl. Mater. Interfaces, 2019, 11(37): 34135.
[47]

Behera B, Joshi R, Anil Vishnu G K, Bhalerao S, Pandya H J. J. Breath Res., 2019, 13(2): 024001.
[48]

Queralto N, Berliner A N, Goldsmith B, Martino R, Rhodes P, Lim S H. J. Breath Res., 2014, 8(2): 027112.
[49]

Lukman Hekiem N L, Md Ralib A A, Mat Hattar M A B, B Ahmad F, Nordin A N, Rahim R A, Za’bah N F. Sens. Actuat. A Phys., 2021, 329: 112792.
[50]

Tao W Y, Lin P, Ai Y Q, Wang H R, Ke S M, Zeng X R. Anal. Biochem., 2016, 494: 85.
[51]

Broza Y Y, Haick H. Nanomedicine, 2013, 8(5): 785.
[52]

Zhou X Y, Xue Z J, Chen X Y, Huang C H, Bai W Q, Lu Z L, Wang T. J. Mater. Chem. B, 2020, 8(16): 3231.
[53]

Haick H, Broza Y Y, Mochalski P, Ruzsanyi V, Amann A. Chem. Soc. Rev., 2014, 43(5): 1423.
[54]

Hakim M, Broza Y Y, Barash O, Peled N, Phillips M, Amann A, Haick H. Chem. Rev., 2012, 112(11): 5949.
[55]

Bharati J, Jha V. Kidney Res. Clin. Pract., 2022, 41(1): 22.
[56]

Bikbov B, Purcell C A, Levey A S, Smith M, Abdoli A, Abebe M, Adebayo O M, Afarideh M, Agarwal S K, Agudelo-Botero M, Ahmadian E, Al-Aly Z, Alipour V, Almasi-Hashiani A, Al-Raddadi R M, Alvis-Guzman N, Amini S, Andrei T, Andrei C L, Andualem Z, Anjomshoa M, Arabloo J, Ashagre A F, Asmelash D, Ataro Z, Atout M M W, Ayanore M A, Badawi A, Bakhtiari A, Ballew S H, Balouchi A, Banach M, Barquera S, Basu S, Bayih M T, Bedi N, Bello A K, Bensenor I M, Bijani A, Boloor A, Borzì A M, Cámera L A, Carrero J J, Carvalho F, Castro F, Catalá-LÓpez F, Chang A R, Chin K L, Chung S C, Cirillo M, Cousin E, Dandona L, Dandona R, Daryani A, Das Gupta R, Demeke F M, Demoz G T, Desta D M, Do H P, Duncan B B, Eftekhari A, Esteghamati A, Fatima S S, Fernandes J C, Fernandes E, Fischer F, Freitas M, Gad M M, Gebremeskel G G, Gebresillassie B M, Geta B, Ghafourifard M, Ghajar A, Ghith N, Gill P S, Ginawi I A, Gupta R, Hafezi-Nejad N, Haj-Mirzaian A, Haj-Mirzaian A, Hariyani N, Hasan M, Hasankhani M, Hasanzadeh A, Hassen H Y, Hay S I, Heidari B, Herteliu C, Hoang C L, Hosseini M, Hostiuc M, Irvani S S N, Islam S M S, Jafari Balalami N, James S L, Jassal S K, Jha V, Jonas J B, Joukar F, Jozwiak J J, Kabir A, Kahsay A, Kasaeian A, Kassa T D, Kassaye H G, Khader Y S, Khalilov R, Khan E A, Khan M S, Khang Y H, Kisa A, Kovesdy C P, Kuate Defo B, Kumar G A, Larsson A O, Lim L L, Lopez A D, Lotufo P A, Majeed A, Malekzadeh R, März W, Masaka A, Meheretu H A A, Miazgowski T, Mirica A, Mirrakhimov E M, Mithra P, Moazen B, Mohammad D K, Mohammadpourhodki R, Mohammed S, Mokdad A H, Morales L, Moreno Velasquez I, Mousavi S M, Mukhopadhyay S, Nachega J B, Nadkarni G N, Nansseu J R, Natarajan G, Nazari J, Neal B, Negoi R I, Nguyen C T, Nikbakhsh R, Noubiap J J, Nowak C, Olagunju A T, Ortiz A, Owolabi M O, Palladino R, Pathak M, Poustchi H, Prakash S, Prasad N, Rafiei A, Raju S B, Ramezanzadeh K, Rawaf S, Rawaf D L, Rawal L, Reiner R C, Rezapour A, Ribeiro D C, Roever L, Rothenbacher D, Rwegerera G M, Saadatagah S, Safari S, Sahle B W, Salem H, Sanabria J, Santos I S, Sarveazad A, Sawhney M, Schaeffner E, Schmidt M I, Schutte A E, Sepanlou S G, Shaikh M A, Sharafi Z, Sharif M, Sharifi A, Silva D A S, Singh J A, Singh N P, Sisay M M M, Soheili A, Sutradhar I, Teklehaimanot B F, Tesfay B E, Teshome G F, Thakur J S, Tonelli M, Tran K B, Tran B X, Tran Ngoc C, Ullah I, Valdez P R, Varughese S, Vos T, Vu L G, Waheed Y, Werdecker A, Wolde H F, Wondmieneh A B, Wulf Hanson S, Yamada T, Yeshaw Y, Yonemoto N, Yusefzadeh H, Zaidi Z, Zaki L, Zaman S B, Zamora N, Zarghi A, Zewdie K A, Ärnlöv J, Coresh J, Perico N, Remuzzi G, Murray C J L, Vos T. Lancet, 2020, 395(10225): 709.
[57]

Španěl P, Davies S, Smith D. Rapid Commun. Mass Spectrom., 1998, 12(12): 763.
[58]

Davies S, Spanel P, Smith D. Kidney Int., 1997, 52(1): 223.
[59]

Narasimhan L R, Goodman W, Patel C K N. Proc. Natl. Acad. Sci. Belarus-Agrar. Ser., 2001, 98: 4617.
[60]

Pang M J, Burclaff J R, Jin R, Adkins-Threats M, Osaki L H, Han Y N, Mills J C, Miao Z F, Wang Z N. Cell. Mol. Gastroenterol. Hepatol., 2022, 13(1): 19.
[61]

Sokolova O, Naumann M. Cell. Mol. Gastroenterol. Hepatol., 2022, 23(3):1883.
[62]

Hooi J K Y, Lai W Y, Ng W K, Suen M M Y, Underwood F E, Tanyingoh D, Malfertheiner P, Graham D Y, Wong V W S, Wu J C Y, Chan F K L, Sung J J Y, Kaplan G G, Ng S C. Gastroenterology, 2017, 153(2): 420.
[63]

Uematsu J, Sugimoto M, Hamada M, Iwata E, Niikura R, Nagata N, Fukuzawa M, Itoi T, Kawai T. J. Clin. Med., 2022, 11(8): 2198.
[64]

Cho J H, Jin S Y, Park S. Expert Rev. Anti-Infect. Ther., 2022, 20: 923.
[65]

Almashhadany D A, Zefenkey Z F, Zaki A M. J. Infect. Dev. Countr., 2022, 16(3): 516.
[66]

Han Y M, Chung S J, Yoo S, Yang J I, Choi J M, Lee J, Kim J S. World J. Gastroenterol., 2022, 28(8): 853.
[67]

Kearney D J, Hubbard T, Putnam D. Dig. Dis. Sci., 2002, 47(11): 2523.
[68]

Turner C, Walton C, Hoashi S, Evans M. J. Breath Res., 2009, 3(4): 046004.
[69]

Thati A, Biswas A, Chowdhury S R, Sau T K. Int. J. Smart Sens. Intell. Syst., 2015, 8(2): 1244.
[70]

Sun H, Saeedi P, Karuranga S, Pinkepank M, Ogurtsova K, Duncan B B, Stein C, Basit A, Chan J C, Mbanya J C. Diabetes Res. Clin. Pract., 2022, 183: 109119.
[71]

Deng C H, Zhang J, Yu X F, Zhang W, Zhang X M. J. Chromatogr. B, 2004, 810(2): 269.
[72]

Righettoni M, Tricoli A, Gass S, Schmid A, Amann A, Pratsinis S E. Anal. Chimica Acta, 2012, 738: 69.
[73]

Chen T, Liu T N, Li T, Zhao H, Chen Q M. Clin. Chimica Acta, 2021, 515: 61.
[74]

Ferrante G, La Grutta S. Front. Pediatr., 2018, 6: 186.
[75]

Pedersen S E, Hurd S S, Lemanske R F, Becker A, Zar H J, Sly P D, Soto-Quiroz M, Wong G, Bateman E D. Pediatr. Pulmonol., 2011, 46(1): 1.
[76]

Lane C, Knight D, Burgess S, Franklin P, Horak F, Legg J, Moeller A, Stick S. Thorax., 2004, 59: 757.
[77]

Horváth I, Barnes P J, Loukides S, Sterk P J, Högman M, Olin A C, Amann A, Antus B, Baraldi E, Bikov A. Eur. Resp. J., 2017, 49(4): 1600965.
[78]

Abdel-Aziz M I, Brinkman P, Vijverberg S J H, Neerincx A H, de Vries R, Dagelet Y W F, Riley J H, Hashimoto S, Montuschi P, Chung K F, Djukanovic R, Fleming L J, Murray C S, Frey U, Bush A, Singer F, Hedlin G, Roberts G, DahlÉn S E, Adcock I M, Fowler S J, Knipping K, Sterk P J, Kraneveld A D, Maitland-van der Zee A H M V. J. Allergy Clin. Immunol., 2020, 146(5): 1045.
[79]

Zhai P, Li H, Zhao R, Li X, Wang H. Chin. J. Ind. Hyg. Occup. Dis., 2019, 37: 899.
[80]

Kubota M. Nihon Rinsho. J. Clin. Med., 2016, 74: 763.
[81]

Li Z L, Liu L G, Zhang Z Q, Yang X H, Li X Y, Gao Y L, Huang K W. Acad. Radiol., 2022, 29(5): 663.
[82]

McKeever T M, Hearson G, Housley G, Reynolds C, Kinnear W, Harrison T W, Kelly A M, Shaw D E. Thorax, 2016, 71(3): 210.
[83]

Jögi J, Ekberg M, Jonson B, Bozovic G, Bajc M. Eur. J. Nucl. Med. Mol. Imaging, 2011, 38(7): 1344.
[84]

Jareño-Esteban J J,Muñoz-Lucas M Á, GÓmez-Martín Ó, Utrilla-Trigo S, GutiÉrrez-Ortega C, Aguilar-Ros A, Collado-Yurrita L, Callol-Sánchez L M. Arch. De Bronconeumología Engl. Ed., 2017, 53(5): 251.
[85]

Rodríguez-Aguilar M, Ramírez-García S, Ilizaliturri-Hernández C, GÓmez-GÓmez A, Van-Brussel E, Díaz-Barriga F, Medellín-Garibay S, Flores-Ramírez R. Biomed. Chromatogr., 2019, 33: e4684.
[86]

Rodríguez-Aguilar M, Díaz de LeÓn-Martínez L, Gorocica-Rosete P, PÉrez-Padilla R, Domínguez-Reyes C A, Tenorio-Torres J A, Ornelas-Rebolledo O, Mehta G, Zamora-Mendoza B N, Flores-Ramírez R. Clin. Chimica Acta, 2021, 518: 83.
[87]

van Berkel J J B N, Dallinga J W, Möller G M, Godschalk R W L, Moonen E J, Wouters E F M, van Schooten F J. Respir. Med., 2010, 104(4): 557.
[88]

Phillips C O, Syed Y, Parthaláin N M, Zwiggelaar R, Claypole T C, Lewis K E. J. Breath Res., 2012, 6(3): 036003.
[89]

Bregy L, Nussbaumer-Ochsner Y, Martinez-Lozano Sinues P, García-GÓmez D, Suter Y, Gaisl T, Stebler N, Gaugg M T, Kohler M, Zenobi R. Clin. Mass Spectrom., 2018, 7: 29.
[90]

Siegel R L, Miller K D, Fuchs H E, Jemal A. CA A Cancer J. Clin., 2021, 71(1): 7.
[91]

Shiinoki T, Fujimoto K, Kawazoe Y, Yuasa Y, Kajima M, Manabe Y, Ono T, Hirano T, Matsunaga K, Tanaka H. Biomed. Phys. Eng. Express, 2022, 8(2): 025008.
[92]

Campos-Silva C, Cáceres-Martell Y, Sánchez-Herrero E, Sandúa A, Beneitez-Martínez A, González Á, Provencio M, Romero A, Jara-Acevedo R, Yáñez-MÓ M, ValÉs-GÓmez M. J. Nanobiotechnology, 2022, 20: 72.
[93]

Wu Y H, Zeng Z, Zhang T, Liu C X, Xu J G, Deng R J, Wang Q, Lin F. Sens. Actuat. B Chem., 2022, 354: 131244.
[94]

Binson V A, Subramoniam M, Mathew L. Clin. Chimica Acta, 2021, 523: 231.
[95]

Jouyban A, Djozan D, Mohammadandashti P, Alizadeh-Nabil A, Ghorbanpour H, Khoubnasabjafari M, Mohammadzadeh M. BioImpacts, 2017, 7(2): 99.
[96]

Ligor T, Pater Ł, Buszewski B. J. Breath Res., 2015, 9(2): 027106.
[97]

Licht J C, Grasemann H. Int. J. Mol. Sci., 2020, 21(24): 9416.
[98]

Karakaya D, Ulucan O, Turkan M. Int. J. Autom. Comput., 2020, 17(2): 179.
[99]

van der Sar I G, Wijbenga N, Nakshbandi G, Aerts J G J V, Manintveld O C, Wijsenbeek M S, Hellemons M E, Moor C C. Respir. Res., 2021, 22: 246.
[100]

Marom O, Nakhoul F, Tisch U, Shiban A L, Abassi Z, Haick H. Nanomedicine, 2012, 7(5): 639.
[101]

Shnayder E, Moshkin M, Petrovskii D, Shevela A, Babko A, Kulikov V. Secondary,Olfaction and Electronic Nose:Proceedings of the 13th International Symposium on Olfaction and Electronic Nose, Brescia: American Institute of Physics, 2009, 1137.
[102]

Sarno R, Wijaya D R. 2017 11th ICTS. IEEE, 2017: 241.
[103]

Tenero L, Sandri M, Piazza M, Paiola G, Zaffanello M, Piacentini G. J. Breath Res., 2020, 14(4): 046003.
[104]

van Bragt J J, Brinkman P, de Vries R, Vijverberg S J, Weersink E J, Haarman E G, de Jongh F H, Kester S, Lucas A, Sterk P J. ERJ Open Research, 2020, 6(4): 00307.
[105]

Chen K, Liu L, Nie B, Lu B C, Fu L D, He Z C, Li W, Pi X T, Liu H Y. Comput. Biol. Med., 2021, 131: 104294.
[106]

Chen K, Liu L, Nie B, Lu B C, Fu L D, He Z C, Li W, Pi X T, Liu H Y. Comput. Biol. Med., 2021, 131: 104294.
[107]

Lekha S, Suchetha M. Secondary, IEEE Sens. J., 2015: 0955.
[108]

Siegel A P, Daneshkhah A, Hardin D S, Shrestha S, Varahramyan K, Agarwal M. J. Breath Res., 2017, 11(2): 026007.
[109]

Jalal A H, Umasankar Y, Chowdhury M, Bhansali S. Meet. Abstr., 2017, MA2017- 02(50): 2130.
[110]

Yu J B, Byun H G, So M S, Huh J S. Sens. Actuat. B Chem., 2005, 108(1/2): 305.
[111]

Li J, Peng Y L, Liu Y, Li W W, Jin Y, Tang Z T, Duan Y X. Clin. Chimica Acta, 2014, 436: 59.
[112]

Beauchamp J, Herbig J, Gutmann R, Hansel A. J. Breath Res., 2008, 2(4): 046001.
[113]

Yan K, Zhang D. 36th Annual International Conference of the IEEE Engineering in Medicine and Biology Society. Chicago, IL, USA. IEEE, 2014, 6406.
[114]

Asadi H, Dowling R, Yan B, Mitchell P. PLoS One, 2014, 9(2): e88225.
[115]

Paulsson N, Larsson E, Winquist F. Sens. Actuat. A Phys., 2000, 84(3): 187.
[116]

Breath Signal Analysis for Diabetics.Eds.: Zhang D, Guo D, Yan K. Springer. 2017.
[117]

Marzorati D, Mainardi L, Sedda G, Gasparri R, Spaggiari L, Cerveri P. Chemosensors, 2021, 9(8): 209.
[118]

Saidi T, Zaim O, Moufid M, El Bari N, Ionescu R, Bouchikhi B. Sens. Actuat. B Chem., 2018, 257: 178.
[119]

Alaqtash M, Sarkodie-Gyan T, Yu H Y, Fuentes O, Brower R, Abdelgawad A. Annual International Conference of the IEEE Engineering in Medicine and Biology Society. Boston, MA, USA. IEEE, 2011, 453.
[120]

Chau T. Gait Posture, 2001, 13(2): 102.
[121]

Marzorati D, Mainardi L, Sedda G, Gasparri R, Spaggiari L, Cerveri P. 41st Annual International Conference of the IEEE Engineering in Medicine and Biology Society. Berlin, Germany. IEEE, 2019, 1584.
[122]

Feng S B, Farha F, Li Q J, Wan Y L, Xu Y, Zhang T, Ning H S. Sensors, 2019, 19(17): 3760.
[123]

Lekha S, Suchetha M. International Conference on Communications and Signal Processing (ICCSP). Melmaruvathur, India. IEEE, 2015, 955.
[124]

Yan K, Zhang D, Wu D R, Wei H, Lu G M. IEEE Trans. Biomed. Eng., 2014, 61(11): 2787.
[125]

Arowolo M O, Adebiyi M O, Adebiyi A A, Olugbara O. J. Big Data, 2021, 8: 29.
[126]

Gou J P, Ma H X, Ou W H, Zeng S N, Rao Y B, Yang H B. Expert Syst. Appl., 2019, 115: 356.

Funding

Zhejiang Natural Science Foundation(LGF22F010008)

Ningbo Natural Science Foundation(202003N4362)

Ningbo Natural Science Foundation(202003N4332)

PDF (13287KB)

2944

Accesses

0

Citation

Detail
Sections
Recommended

/

〈 〉

Address of Editorial Office: No. 33 Beisihuan Xilu, Zhongguancun , Beijing 100190, China
Tel: +86 10 8262 7757
Fax: +86 10 8262 7757
E-mail: progchem@mail.las.ac.cn
http://www.progchem.ac.cn
    • 　
    Copyright © Editorial Office of Progress in Chemistry