1 Introduction
diabetes mellitus,known as Diabetes in traditional Chinese medicine,is a group of metabolic diseases characterized by chronic hyperglycemia caused by multiple causes.Diabetes mellitus,as the third major disease endangering human health,has attracted worldwide attention.According to the Guidelines for the Prevention and Treatment of Type 2 Diabetes in China published in 2020,the prevalence rate of Diabetes is 11.2%,and there are about 114 million diabetic patients in China
[1⇓~3]。 At present,the detection methods of diabetes mellitus are invasive,and they also require certain laboratory conditions and professional technicians to operate.In addition,traditional detection methods also have some limitations,for example,glycosylated hemoglobin can only reflect the recent blood sugar control,can not reflect the fluctuation of blood sugar
[4]; blood glucose measurement can only provide instantaneous blood glucose value,but can not fully reflect the situation of blood glucose metabolism
[5]。 Therefore,how to find a more convenient,fast,accurate and non-invasive method for diabetes detection has always been the goal pursued by the medical community。
In recent years,exhaled breath detection,as a new way of disease detection,is developing rapidly.Compared with traditional detection methods,exhaled gas detection has many advantages,such as non-invasive,convenient and safe.Therefore,this technology has been applied to the detection and screening of lung cancer,chronic kidney disease,diabetes and other diseases
[6⇓~8]。 in ancient Greece,doctors discovered the smell of"rotten apples"in the exhaled breath of diabetic patients,but did not know the composition of the smell.Until 1971,Pauling et al.Used gas chromatography to discover that there were hundreds of trace Volatile Organic Compounds(VOCs)in human exhaled air
[9]。 Subsequently,in the 1990s,Phillips et al.Used a similar gas chromatography(GC)technique to find 22 VOCs in the exhaled gas of lung cancer patients,which were mainly composed of alkanes and benzene derivatives,laying the foundation for the development of exhaled gas detection technology
[10]。
At present,many studies on the detection and analysis of exhaled breath in diabetes mellitus have been reported
[11,12]。 in this paper,search engines such as CNKI,VIP,Wanfang,PubMed and Web of Science were used to search the Chinese and English literatures related to this topic by using the keywords"exhaled breath detection","diabetes","acetone"and"volatile organic compounds".the research progress of exhaled breath detection and analysis in diabetes mellitus was reviewed from the following three aspects:the composition and source of exhaled breath in diabetes mellitus,the detection method of exhalated breath in diabetes mellitus,and the algorithm of exhalated breath detection in diabetes mellitus。
2 Composition and source of exhaled gas in diabetes mellitus
2.1 Exhaled gas composition
in addition to the well-known gases such as nitrogen,carbon dioxide,oxygen and other inert gases,There are also Trace organic compounds in human exhaled gas.trace organic compounds can be divided into Non-volatile organic compounds(NOCs)and volatile organic compounds(VOCs),of which VOCs account for less than 1%of the total volume of exhaled gas.there are about 1488 kinds of VOCs in human breath
[13]。 These VOCs are mainly ketones,aldehydes,aromatics and alcohols.From the source,VOCs can be divided into endogenous and exogenous
[14]。 Exogenous components are organic compounds that are absorbed by the human body and excreted again,while endogenous components are considered to be trace organic compounds produced by the metabolism of the body(including tumor cells and microorganisms).the composition and content changes of these compounds reflect the changes of homeostasis system,so they are used in non-invasive detection of lung cancer,diabetes,chronic kidney disease,cirrhosis,breast cancer and so on
[8,15⇓⇓⇓~19]。
2.2 Causes and physiological sources of changes in exhaled breath composition in diabetic patients
Diabetes mellitus is a metabolic disease,which is caused by chronic hyperglycemia due to many factors,and the defect of insulin secretion and(or)utilization is the main cause.Long-term carbohydrate,fat and protein metabolic disorders can cause damage to multiple systems,resulting in chronic progressive damage,dysfunction and failure of tissues and organs such as eyes,kidneys,nerves,heart and blood vessels.Acute severe metabolic disorders,such as diabetic ketoacidosis(DKA)and hyperosmolar hyperglycemia syndrome,may also occur in severe situations or under stress。
Usually,the human body can convert food into energy,and the metabolites are excreted through breathing,sweat,urine and so on.However,in diabetic patients,due to the lack of insulin or the obstruction of insulin action,blood sugar levels rise,resulting in the imbalance of metabolites in the body。
For example,acetone is known as a biomarker of diabetes,and acetone is one of the three ketones metabolized by the liver.When the body's carbohydrate energy supply is insufficient,fat will be broken down into fatty acids and metabolized in the liver to produce ketone bodies.For patients with uncontrolled diabetes,fat metabolism is accelerated,leading to excessive accumulation of ketones in the blood and symptoms of DKA
[20]。 the main source of acetone is acetoacetate decarboxylation reaction,and acetone is very volatile,in the human body for gas exchange or blood circulation,excessive acetone will appear in the human exhaled gas.However,it is worth noting that acetone is not the only biomarker of exhaled gas in diabetes,and studies have found that acetone can also exist in lung cancer,esophageal cancer and other diseases
[21][22]。 in addition,acetone exhaled by healthy people is usually less than 0.8 ppm,while it is more than 1.8 ppm In diabetics
[23]。
Isopropanol is a new potential biomarker,which plays an important role in the detection of exhaled breath in diabetes
[24]。 in the human body,Isopropanol is a product produced by the reverse catalytic reduction of acetone by alcohol dehydrogenase.Abnormal fat metabolism in diabetes leads toβ-oxidation of fatty acids to produce reduced nicotinamide adenine dinucleotide(NADH),which stimulates excessive acetone reduction in the liver,resulting in the production of isopropanol.Isopropanol concentration is positively correlated with acetone concentration because acetone concentration is increased in diabetic patients.that is to say,the concentration of isopropyl alcohol in diabetic patients is higher than That in healthy people
[25,26]。 It has been found that the concentration of isopropyl alcohol in the exhaled breath of healthy people is usually less than 0.026 ppm,and the concentration of isopropyl alcohol in the exhaled breath of diabetic patients is usually greater than 0.043 ppm
[27]。
Studies have shown that abnormal levels of acetylcholinesterase(AChE)are associated with diabetes and its complications.As a neurotransmitter and enzyme,AChE can degrade acetylcholine in human body,and its activity is related to the apoptosis of islet beta cells,which may be one of the factors leading to insulin-dependent diabetes mellitus
[28]。
methyl nitrate is an endogenous molecule that may reflect the dynamic changes of blood glucose concentration in diabetic patients.Studies have shown that long-term hyperglycemia can lead to enhanced mitochondrial metabolism in cells,which can lead to hyperoxidation and accelerate the synthesis of Methyl nitrate
[29]。 in addition,the increase in ketone bodies also acidifies the blood,further accelerating the synthesis of methyl nitrate.At the same time,long-term hyperglycemia also inhibits the normal metabolism of aromatic compounds such as ethylbenzene,toluene and xylene in the liver,resulting in an increase in their concentration in the blood.These aromatic compounds are highly volatile,so their content in the exhaled breath will also increase
[30]。
According to the general understanding of clinical research,oxidative stress is an important mechanism leading to the occurrence and development of diabetes
[31]。 oxidative stress is the production of highly reactive molecules,such as reactive oxygen species(ROS)and reactive nitrogen species(RNS),which are overproduced or reduced in clearance when the body is exposed to harmful stimuli.These highly active molecules can directly destroy tissues and cells,and can also act as cell messengers to activate a variety of signal transduction pathways,thus indirectly causing tissue and cell damage.Patients with T2DM are more likely to suffer from oxidative stress because of their reduced immune function,and the end products of oxidative stress are hydrocarbons,which are more volatile in the blood and more easily detected in human exhaled breath
[32]。 Normally,the concentration of isoprene in the exhaled breath of healthy people is less than 3.5 nmol/L
[33]。
it has been found that ROS is closely related to the pathogenesis of inflammatory diseases,and It plays a direct or indirect role in tissue damage caused by oxidative stress.ROS are also involved in the impairment ofβ-cell function during the development of diabetes,leading to the genetic ablation of KATP channels and triggering the upregulation of antioxidant enzymes.Researchers can assess ROS-induced lipid peroxidation,protein oxidation,and DNA damage by measuring several markers.8-isoprostane has long been considered the best marker of lipid peroxidation
[34,35]。 It is now found that the concentration of 8-isoprostane in the exhaled breath of diabetic patients increases with the severity of the disease
[36]。
Nowadays,no unique biomarker of diabetes has been found in the study of exhaled gas detection of diabetes at home and abroad,and acetone is mostly used as a single VOCs to detect diabetes.Therefore,the accuracy and sensitivity of combining multiple VOCs to diagnose diabetes will be higher.Now,the combination of digital system and exhaled breath detection technology can improve the accuracy of screening by combining and analyzing the detection results of exhaled breath of subjects
[37⇓~39]。 Therefore,the best joint detection model is obtained
[40]。 It can be used for non-invasive screening of pre-diabetes or high-risk groups to reduce the incidence of diabetes。
3 Diabetes mellitus exhaled gas detection method
At present,many detection technologies have been applied to the detection of exhaled gas in diabetes mellitus,such as GC,gas chromatography-mass spectrometry(GC-MS),selected ion flow tube mass spectrometry(SIFT-MS),proton transfer reaction mass spectrometry(PTR-MS),laser spectroscopy,electronic nose,sensors and so on
[41][42][8][43][44][45][46]。 Each method has its advantages and disadvantages.for example,GC and GC-MS are the most commonly used methods For exhaled gas detection.SIFT-MS,PTR-MS,laser spectroscopy,etc.have high sensitivity.Sensors and electronic noses Have the advantages of small size,convenience and low price。
3.1 Gas chromatography detection method
the detection of exhaled breath in diabetes mellitus by gas chromatography can be divided into two steps,one is exhaled breath collection,The other is exhaled gas pre-enrichment and analysis。
3.1.1 Exhaled breath sample collection
the total amount of exhaled air in a human body is about 500 mL,which can be divided into two parts.The first 150 mL of exhaled air is upper airway gas,also known as"dead space gas.".The remaining 350 mL of exhaled air comes from The alveoli and is referred to as"alveolar air."
[47,48]。 alveolar gas is the product of exchange between pulmonary circulation and blood,and is considered to be the headspace gas in blood.Because of its special properties,Alveolar gas is regarded as a target component of exhaled breath for disease diagnosis。
At present,the main devices for exhaled gas storage are gas bags,exhaled gas condensers,adsorption tubes,bronchoscope-guided collection equipment,etc
[49,50]。 When the air bag collects exhaled air,it must ensure that it is isolated from the outside air and that no absorption or release of VOCs occurs.this can be achieved through the use of sealed gas pockets and the selection of materials that are chemically stable and have low adsorption.TedlarR air bag is made of polyvinyl fluoride,which is the most widely used air bag in exhaled gas collection at home and abroad.the material of This air bag is chemically inert to most compounds,and has the characteristics of good corrosion resistance,low adsorption and stable gas storage for up to 10 hours
[51]。 Exhaled breath condensate(EBC)refers to the liquid that is condensed and compressed into 1-3 mL by a condenser and stored At-70℃.at present,the collection and storage methods of EBC are not standardized,and most studies use self-made equipment to collect EBC,but the lumen and storage chamber of the equipment are made of glass,aluminum,polystyrene and other materials,which may adsorb the substances to be detected and affect the detection results of EBC
[52]。
In the process of human exhaled gas collection,alveolar gas collection will be affected by"dead space gas"from the upper respiratory tract such as mouth,throat and nasal cavity,resulting in dilution of alveolar gas,which will lead to the concentration of trace volatile organic metabolites in the mixture lower than that in alveolar gas.Dead space gas interferes with the analysis of exhaled gas,so alveolar gas should be collected when exhaled gas is collected.At present,the alveolar gas collection device mainly adopts an automatic alveolar gas collection device which collects exhaled gas based on the change of CO
2concentration.Schubert et al first developed a CO
2monitoring and collection device,which starts to collect alveolar gas when the CO
2content is about 4%,thus eliminating the influence of dead space gas
[53]。 In the first stage,the human body is in the plateau stage at the end of inspiration and the beginning of expiration,and the exhaled gas at this stage is dead space gas.During this period,the exhaled gas was mainly O
2,and the content of CO
2was almost zero.The content of CO
2in segmentⅡincreased sharply to about 4%,and the exhaled gas was mainly the mixture of dead space gas and alveolar gas.Stage III refers to the gradual stabilization of CO
2content after reaching 4%-5%,which is alveolar gas,and exhaled gas can be collected
[54]。
However,it should be noted that exhaled gas collection is affected by many factors,including environmental factors,respiratory rate,respiratory depth and respiratory rhythm,which may affect the final test results.Therefore,when collecting exhaled gas,we should try to maintain a natural,stable and regular breathing pattern,and avoid a fast,deep and uneven breathing pattern,so as to reduce the occurrence of errors.in addition,due to the existence of environmental factors,In order to reduce their interference on the results,exhaled gas samples and environmental samples can be collected at the same time,and the measured environmental concentration can be removed as the background.This method can effectively reduce the influence of environmental background on the results of exhaled breath collection
[55,56]。 Therefore,during exhaled gas collection,attention should be paid to maintaining the natural breathing pattern and adopting the way of environmental background removal to reduce the influence of environmental factors and breathing patterns on the results。
3.1.2 Preconcentration of exhaled breath sample
because of the low concentration of VOCs in the exhaled gas,VOCs are generally captured in the enrichment adsorption material by enrichment after the gas is collected.Therefore,trace organic compounds can be accurately and efficiently identified by enrichment.Adsorbent enrichment is widely used Because it is easy to store and can enrich a variety of compounds,but it requires additional steps to collect and treat the Adsorbent.Adsorbent materials can be classified as single or composite materials such as Tenax-TA,Carbotrap Y,UiO-66,Tenax-TA/Carbopack B,TENTAX-TA/Sulficarb
[57⇓~59]。
solid-phase microextraction(SPME)is a solvent-free or solvent-less enrichment technique.This technique utilizes a solid-phase needle coated with an adsorbed phase,such as a polymer or fibrous material,to adsorb gas-phase analytes by insertion into the injection port.Subsequently,the needle was removed and inserted directly into the gas chromatograph for further analysis
[60]。 SPME has several advantages,including simplicity,rapid operation,no solvent requirement,and high sensitivity
[61]。 For example,in the study of organic matter in patients with Gestational Diabetes Mellitus(GDM),Sana et al.Used solid-phase microextraction gas chromatography/mass spectrometry to detect plasma and urine samples
[62]。 This approach aims to analyze metabolomic changes in patients with GDM and investigate The mechanisms associated with cognitive decline.the results suggest that 2-propanol may be used as a potential volatile marker to assess cognitive impairment in pregnant women with GDM。
3.1.3 GC, GC-MS method
exhaled breath detection appeared in the 1970s.Pauling et al.used GC to separate more than 200 VOCs in human Exhaled breath.However,GC can be Used for quantitative analysis and qualitative analysis of known substances,but not for qualitative analysis of unknown substances
[63]。 the emergence of GC-MS technology just makes up for The deficiency of GC
[64]。
Nowadays,GC and GC-MS are still the classical methods in the detection of exhaled gas in diabetes mellitus
[41,65]。 Principle of GC technology:GC system includes carrier gas part,sample injection part,separation part,detection part and data detection and processing part.the sample is injected at one end of the column and subsequently brought into the column by the gas mobile phase.the adsorption or solubility of each component in the sample between the stationary phase and the mobile phase is different,that is,the partition coefficient of each component is different.When these components are partitioned between the two phases for many times and move with the mobile phase,their movement speed in the chromatographic column will also be different.A component with a smaller partition coefficient will remain in the stationary phase for a shorter time and will therefore be able to flow out of the end of the column faster.chromatographic column and detector are the core of GC system.different chromatographic columns and detectors are used to separate different gases,and the separation and detection effects will be different.Nowadays,flame ionization detector(FID)and mass spectrometry detector(MS)are widely used
[66][67]。 DB-624,DB-5MS,HP-INNOwax,etc.Are generally used as chromatographic columns
[68][69][70]。
Liu et al.Measured the concentration of acetone in breath andβ-hydroxybutyrate in fingertip blood of 99 diabetic patients by GC-MS
[24]。 According to urine ketone concentration,patients were divided into five groups:1(-),2(±),3(+),4(++),5(+++).the results showed that the sensitivity and specificity of exhaled acetone were 90.9%and 77.1%,respectively,when the concentration ofβ-hydroxybutyrate in Blood was used as the standard to evaluate the sensitivity and specificity of acetone in respiratory gas and ketone in urine.the sensitivity and specificity of urine ketone were 63.6%and 85.7%,respectively.These findings suggest that the specificity of acetone in exhaled breath is similar to that of urine ketone,but the sensitivity is higher.in addition,even in the group with a negative urine ketone body test,the blood tests forβ-hydroxybutyric acid and exhaled acetone were positive,6.7%and 18.8%,respectively.Therefore,urine ketone concentration may not be a reliable predictor of early diabetic ketosis in a timely manner.blood and breath gas tests for ketones help to rule out false negative results。
Yan et al.Used GC-MS and metabolomics technology to analyze multiple data of human exhaled gas to identify the differences of respiratory metabolites between type 2 diabetes mellitus(T2DM)and healthy people,and to find unique biomarkers
[40]。 A combination of eight potential biomarkers was successfully identified.In addition,the combination of isopropanol,2,3,4-trimethylhexane,2,6,8-trimethyldecane,tridecane,and undecane was found to have a specificity of 100%and a sensitivity of 97.7%for identifying T2DM.these results suggest that These compounds may be the best biomarkers for clinical diagnosis of T2DM。
Grabowska-Polanowska et al.Used GC-MS technology to analyze the expired gas of patients with chronic kidney disease(CKD)and T2DM,and found that trimethylamine(TMA)was only present in CKD patients
[71]。 in addition,the concentration of expired methyl mercaptan(MeSH)was higher in CKD patients with diabetes and lower in patients with only renal dysfunction or in the healthy group.These detected VOCs can be used for the diagnosis of CKD and T2DM。
At present,GC and GC-MS methods need sample preparation and pre-concentration in the detection of exhaled gas,which is highly complex
[72][73]。
3.2 Mass spectrometric direct detection method
SIFT-MS and PTR-MS are new methods for rapid on-line detection of exhaled breath with high sensitivity,and both methods can continuously monitor the concentration changes of compounds,such as diabetes mellitus,which can play a long-term monitoring role。
Malina et al.Used SIFT-MS to study the concentration of acetone in the exhaled breath of 38 T2DM patients with long-term dietary modification
[74]。 Anthropomorphic measurements,dietary intake,and medication use were recorded.bloodβ-hydroxybutyric acid(ketone body),glycosylated hemoglobin(HbA1c),and glucose were analyzed by point-of-care capillary(finger prick)testing.exhaled acetone was found to vary between 160 and 862 ppb(median 337 ppb)and was significantly higher in men(median 480 ppb versus 296 ppb,p=0.01).Although no association was observed between acetone in Exhaled breath and dietary macronutrients or immediate capillary Blood tests.But SIFT-MS expiratory analysis provides a rapid,reproducible,and easy-to-perform measure of acetone concentration in patients with T2DM。
Siegmund et al.Collected and analyzed exhaled breath samples from 21 patients with T2DM and 26 healthy controls
[75]。 PTR-MS was used to analyze VOCs in the range of 20~200 atomic mass units.the study findings identified eight mass characteristics of endogenous VOCs(Table 2)that were significantly different in the gas distribution of T2DM patients and healthy controls.Linear discriminant analysis of These VOCs yielded a sensitivity of 90%and a specificity of 92%.these results suggest that exhaled endogenous VOCs can distinguish T2DM patients from healthy controls by multivariate analysis。
However,SIFT-MS and PTR-MS have some limitations.For example,SIFT-MS and PTR-MS can not identify compounds,and the detection range of PTR-MS is narrow
[73]。
3.3 Pectroscopy detection method
GC,GC-MS,SIFT-MS and other equipment used for exhaled gas detection have some problems,such as large equipment volume,complex sample collection and pre-enrichment,which are difficult to promote.in contrast,laser spectroscopy has the advantages of good robustness,short response time and high sensitivity,so it has broad application prospects in routine use in hospitals
[76][77]。 At present,the spectroscopy methods that can be applied to exhaled breath detection mainly include tunable diode laser absorption spectroscopy,cavity ring-down spectroscopy,integrated cavity output spectroscopy,photoacoustic spectroscopy,external cavity quantum cascade laser and vacuum ultraviolet spectroscopy system
[44,78][78,79][80][81][82][83]。
Fufurin et al.Introduced a method based on infrared laser spectroscopy for diagnosis of type 1 diabetes mellitus(T1DM)
[84]。 A quantum cascade laser was used to emit in the spectral range of 5.3-12.8 microns in combination with a Herriot multi-vent cell with a 76 meter optical path.This method,used to collect and dry exhaled human air samples,measured 1200 infrared exhaled gas spectra from 60 healthy volunteers(control group)and 60 volunteers with confirmed T1DM(target group).A one-dimensional convolutional neural network was used to classify healthy and T1DM volunteers with an accuracy of 99.7%,a recall of 99.6%,and an AUC score of 99.9%。
acetone is one of the important indicators for the diagnosis of diabetes at this stage,so the study of acetone has attracted much attention.exhaled acetone in normal subjects ranges from 300 to 1000 ppbV,while Exhaled acetone in diabetics ranges from 1500 to 2500 ppbV
[85]。 Sun et al.Detected the concentration of acetone in the exhaled breath of healthy people and T2DM patients by the ring breath acetone analyzer based on cavity ring spectroscopy
[76]。 the results of the experiment showed that the average acetone concentration of all T2DM patients was higher than that of healthy subjects on fasting,2 hours after breakfast,2 hours after lunch and 2 hours after dinner.Nadeem et al.Used an external cavity quantum cascade laser to study acetone spectroscopy
[86]。 Using an external cavity quantum cascade laser for broadband direct absorption spectroscopy and wavelength modulation spectroscopy,the research team was able to measure the entire molecular absorption band of acetone.In the second harmonic wavelength modulation spectroscopy(WMS-2 f),a modulation amplitude of 10 GHz was used to maximize the WMS-2 f signal from the q-branch peak of acetone,further improving the noise equivalent absorption sensitivity(NEAS)and the minimum detectable absorption(MDA).A NEAS of 1.9×10
-8cm
-1·Hz
-1/2and an MDA of 15 ppbv were achieved in less than 10 seconds,resulting in a significant increase in sensitivity and response speed when detecting acetone.Kudo et al.Proposed a vacuum ultraviolet(VUV)spectroscopy system to measure acetone content in human exhaled breath
[87]。 the spectroscopy system consists of a deuterium light source,a hollow fiber gas cell,and a fiber-coupled compact spectrometer suitable for the VUV region,and the measurement is performed by detecting the strongly absorbed acetone peak at 195 nm.the standard addition method is used to improve the measurement accuracy,and the test is based on human respiration.the results show that the standard deviation is 0.074 ppm and the precision is 0.026 ppm for a healthy person at an acetone concentration of about 0.8 ppm.This new spectral system is used to detect exhaled gas in diabetes mellitus,which improves the accuracy of detection and avoids misdiagnosis and missed diagnosis。
Currently,spectroscopy techniques are characterized by high sensitivity,ranging from ppm to ppt levels
[88]。 the advent of spectroscopy for The detection of exhaled breath has made it possible for exhaled breath analysis to move from mass spectrometry-based,time-consuming laboratory studies to spectrometry-based,real-time clinical testing
[89]。 However,spectroscopy detection methods have the problem of insufficient ability to detect multiple compounds at the same time
[90]。
3.4 Sensor detection method
in recent years,sensors In the field of medical applications have been developed rapidly,and have made great contributions to disease detection,treatment and monitoring.the main advantage of the sensor is its small size,which meets the needs of diabetic patients to monitor the progress of the disease every day.At present,the sensors commonly used to detect the exhaled gas of diabetic patients include electronic nose,metal oxide(MOx)gas sensor,chemical sensor and biological sensor
[45,91][92⇓⇓⇓~96][46][97]。
the electronic nose is a potential tool for screening and analysis of various respiratory diseases,which integrates a sensor array and an artificial neural network capable of detecting and identifying specific VOCs patterns.Due to its non-invasive nature,The electronic nose can be used as an effective disease monitoring technique
[98]。 Bahos et al.Designed a novel electronic nose consisting of a surface acoustic wave(SAW)sensor array based on a zeolitic imidazolate framework and ZIF-8 and ZIF-67 nanocrystals(pure and combined with gold nanoparticles)as sensitive layers
[99]。 It was found that the sensor had high sensitivity,good reproducibility,short response time and rapid signal recovery for low concentrations of 5 ppm,10 ppm and 25 ppm of the label.Weng and other researchers have independently developed a vehicle-mounted electronic nose sensor array,which can detect biomarkers such as ethanol,acetone,alkanes,carbon monoxide and methyl nitrate in the exhaled gas of diabetic patients
[100]。 the gradient marching method is used to select the feature subset,and then combined with the particle swarm optimization algorithm,the 24 most effective features are extracted.This optimization not only reduces the number of sensors by 56%,but also reduces the system cost.Experiments show that the accuracy of vehicle electronic nose in detecting diabetes is 93.33%.in addition,the system is low in cost,small in size,and easy to install in the vehicle.the optimized vehicle electronic nose system provides a more feasible method for the initial screening of diabetes in vehicles,and can be used as an aid to existing detection methods。
Prasanth et al.Developed an optical fiber sensor based on standing wave field,which was coated with SnO
2/MoS
2double-layer film to detect different concentrations of acetone
[101]。 The sensor response is enhanced by 23.5%.When 250 ppm acetone concentration was used for analysis,the response time and recovery time were about 14 s and 17 s.Response time and recovery time are greatly reduced.This study shows that the SnO
2/MoS
2coated sensor has the potential to create a handheld sensor system for monitoring diabetes 。
Ramji et al.Developed a chemiresistive sensor based on graphene for the detection of exhaled gas in diabetes
[102]。 the sensor has a high sensitivity for the selective detection of acetone(5.66 for 1 ppm acetone vapor),and the response time and recovery time are 10 s and 12 s,respectively,at low concentrations.the mean sensor response in diabetic patients was 1.1 times higher than that in healthy subjects when analyzed after exhaled breath collection in 13 healthy subjects and 17 diabetic patients,respectively.the results show that the sensor is sensitive to acetone and small in size,so it can be used for the detection of exhaled gas in diabetes mellitus.But at the same time,the existing electronic nose technology generally can not screen for multiple diseases,the detection is easily affected by the water vapor in the analyte,and can not identify a single compound in a complex gas mixture
[103]。
4 Diabetes exhaled breath detection algorithm
Medical integration,as a product of the era of big data,reflects the exhaled gas of the human body truly,objectively and accurately through computer technology.Algorithms can be used to process exhaled gas data.and can improve accuracy,sensitivity and specificity
[105⇓~107]。 Comprehensive literature research shows that although there are still many limitations in the detection of diabetes mellitus by multi-function respiratory analyzer,such as high metabolic variability between patients.However,the results obtained using the algorithm are very encouraging.Most of these algorithms can accurately detect diabetes in exhaled breath with an accuracy of more than 90%
[108]。 At present,the commonly used algorithms for exhaled breath detection mainly include Support Vector Machine(SVM),Decision Tree(DT),K-Nearest Neighbor(KNN),Principal Component Analysis(PCA),Convolutional Neural Network(CNN),etc。
SVM is a machine learning method,whose core idea is to solve the optimal classification hyperplane based on the structural risk minimization criterion.By finding the optimal hyperplane,SVM can classify the data set into different classes,thus achieving the classification task
[109]。 Yan et al.Used SVM algorithm to classify the breath samples of healthy people and diabetic patients
[110]。 the study used 140 randomly selected samples of patients and the same number of healthy people for training,and the remaining samples(139 per category)were used for model validation.the authors performed 50 times of model training,and the results showed that the average sensitivity was 91.51%and the specificity was 90.77%,indicating that the algorithm can be used for diabetes screening。
CNN uses convolution operation to extract features from input data,reduces the size and number of feature maps through pooling operation,extracts important features,and maps features to output results through fully connected layers
[111]。 Lekha et al.found that the average execution time of CNN algorithm to extract the original signal features is 0.1203 seconds
[112]。 this computation time is significantly reduced compared to The time required for feature extraction by the SVD technique(0.4803 seconds)and PCA(0.5226 seconds).Therefore,the CNN algorithm is very suitable for real-time non-invasive detection and classification of diabetic patients.the results of This study provide new ideas and methods for medical diagnosis using deep learning technology。
5 Summary and Prospect
the detection and Analysis of exhaled breath in diabetes mellitus is a potential and cutting-edge research field.analysis of VOCs in exhaled breath of diabetic patients can yield important information about metabolic status and disease process.Existing studies have already shown the presence of disease-associated unique compounds in the exhaled breath of diabetic patients,which provides a potential new way to achieve non-invasive,convenient and fast diagnosis and monitoring of diabetes
[40,113]。
Compared with traditional laboratory examination,exhaled gas detection has the advantages of non-invasive,convenient and easy to achieve.This technology can realize real-time monitoring of diabetic patients and early warning of potential diabetic patients。
However,the current study faces several challenges and limitations.First,the concentration of volatile organic compounds in exhaled breath is usually low,so the sensitivity and accuracy of detection methods need to be improved.Secondly,there is interference with other physiological and environmental factors,and there is a need to better distinguish specific signals associated with diabetes.In addition,standardized methods and processes need to be further developed to ensure the reproducibility and comparability of results.In the study,the number and diversity of samples,as well as the association with other clinical indicators,also need to be considered to improve the value of the method in clinical application。
Key points for the development of exhaled breath detection and analysis in diabetes:(1)by characterizing VOCs,the study can further identify specific compound combinations for different types and stages of diabetes.this analysis can help establish the fingerprint of exhaled breath in diabetes and provide more accurate guidance for early diagnosis of the disease and individualized treatment.(2)the complex mechanism of diabetes and the dynamic process of disease development can be further revealed By integrating the exhaled gas analysis of diabetes with other biomarkers,imaging and genetic data.the integration of This multimodal analysis can help to provide more comprehensive and accurate information and provide a deeper understanding of the research and treatment of diabetes.(3)Large-scale equipment needs to be miniaturized to achieve the purpose of home health monitoring
[114]。
diabetes mellitus,as a kind of disease that endangers human health,can be controlled in the early screening of the disease,so as to cut off the occurrence and development of the disease from the root.Future studies are needed to overcome the current challenges and further explore the mechanisms and biological significance of exhaled gas analysis in diabetes.At the same time,it is applied to clinical practice and interdisciplinary collaboration with other scientific fields to achieve the goal of individualized treatment and diabetes management.in order to achieve clinical translation,the following key steps are needed:(1)clinical validation:large-scale clinical studies are carried out to verify the accuracy and reliability of exhaled breath analysis technology in diabetes diagnosis,treatment monitoring,etc.(2)Multi-disciplinary cooperation:promote the integration of medical industry,strengthen the cooperation and exchange between exhaled breath analysis technology and clinical medicine,biomedical engineering,data science and other fields,and jointly solve the challenges and obstacles faced by technology in the process of clinical transformation.(3)Standardization and normalization:to formulate standardized operation procedures and quality control standards for exhaled breath analysis technology,ensure the repeatability and comparability of the technology in different clinical environments,and improve the credibility of the technology in clinical practice。
National Key Research and Development Program of China(2018YFA0208504)
Shanghai "Science and Technology Innovation Action Plan" Medical Innovation Research Special Program(22Y11900600)
General project of the National Natural Science Foundation of China(8217142522)
PDF
