Early Disease Diagnosis System Based on Sebum Gas Analysis

Abstract

In view of the problems of difficult early diagnosis of current diseases and low diagnostic accuracy, the present invention proposes an early disease diagnosis system based on sebum gas analysis. A sebum gas collection and infrared spectrum enhancement device prepared by combining a microelectromechanical system, a new two-dimensional material and a metal surface plasmon resonance technology is used to collect trace sebum gas from the human body and enhance its infrared spectrum by metal plasmon. Then, the metal plasmon-enhanced infrared spectrum of the sebum gas is input into the early disease diagnosis model, and finally, the early diagnosis of whether the diagnosed person is ill is realized. The early disease diagnosis system based on sebum gas analysis disclosed by the present invention provides a non-invasive diagnosis method for the diagnosis of diseases, and has the advantages of being non-invasive, simple, efficient, and having no pollution to the environment and no ecological damage.

Description

Technical Field

[0001] The present invention belongs to the technical field of early disease diagnosis, and particularly relates to an early disease diagnosis system based on sebum gas analysis. Background Art

[0002] Substance and energy metabolism is one of the characteristics of living organisms, and the products of human metabolism can reflect abnormal changes in the human body. When human sweat and sebum metabolites are generated, they will emit specific odors to express certain diseases. For example, the fat of diabetic patients is oxidized in the liver to produce ketone bodies, emitting the smell of rotten apples; patients with chronic nephritis or liver diseases emit ammonia smell due to the retention of urea nitrogen, creatinine, etc. in the blood; the sebum gas of patients with Parkinson's Disease (PD) contains organic gases such as perillaldehyde, showing a body odor similar to "musk"; patients with Alzheimer's disease emit formaldehyde gas from their bodies, etc.

[0003] For many chronic diseases suffered by the human body, such as cancer, Parkinson's disease, Alzheimer's disease, etc., their onset is a slow process, and some diseases do not show symptoms until the course of the disease is as long as 10 years or even longer. Taking Parkinson's disease as an example, in 2015, the British Broadcasting Corporation reported that Joy Milne, a nurse, smelled a special smell from her husband with PD 12 years before he showed relevant signs and symptoms. In 2019, Dr. Tilo Kunath, a famous Parkinson's disease expert at the University of Edinburgh in the UK, also confirmed through research that four organic components in human sebum, such as hippuric acid (C9H9NO3), perillaldehyde (C 10 H 14 O), eicosane (C 20 H 42 ), and octadecane (C 18 H 38 ), are closely related to PD. If these specific gases can be detected in time at the early stage of the disease, it will provide a basis for diagnosing and treating the disease before it occurs, and then realizing early diagnosis and early treatment of the disease.

[0004] However, the gases released by the human body are difficult to detect. First of all, the gases that form body odor are a mixture of gases released from human sweat, sebum, etc., containing organic volatile components, with very low content and difficult to collect. Existing methods for collecting human sebum gases all use gauze to wipe the skin of a person to collect sebum, and then the gases are released through thermal desorption attachment. The collection process is cumbersome and complex, with very high requirements for equipment, and impurities that cannot be removed will be mixed in during the gas release process, seriously affecting the detection accuracy. Secondly, for the analysis of sebum gases, it is impossible to avoid using a gas chromatograph, which is expensive and complex to operate, greatly increasing the threshold and difficulty of sebum gas analysis. Thirdly, the components of human sebum gases are complex and belong to volatile organic compounds (VOCs). Existing detection methods for VOCs can only measure the total concentration of VOCs, but cannot determine their component content.

[0005] Currently, scholars have begun to pay attention to the relationship between human sebum gases and diseases, and attempt to diagnose diseases by analyzing sebum gases. The research group of Chen Xing developed a portable "electronic nose" combining gas chromatography and surface acoustic wave sensing to analyze human sebum gases for the diagnosis of PD, obtaining a specificity of 91.6% and a sensitivity of 91.7%. The team of Kinji OHNO in Japan obtained an analysis accuracy rate of 90.2% and a specificity of 85.2% by analyzing the relationship between the chemical substances in skin gases and PD, and concluded that the gases on the skin may be used as biomarkers for PD. The team of Nobutaka Hattori identified PD by analyzing sebum RNA, age, and gender information. The team of Professor Hosam Haick used a chemiresistor sensor, a silicon nanowire sensor, and gas chromatography-mass spectrometry to analyze the breath samples of PD patients, obtaining an accuracy rate of 76%, a sensitivity of 77%, and a specificity of 73%.

[0006] To sum up, the analysis of human sebum gases at home and abroad is still restricted by sebum sampling methods, detection instruments, and analysis methods, and cannot achieve good results. Summary of the Invention

[0007] In view of the problems of difficult early diagnosis of current diseases and low diagnosis accuracy rate, the present invention proposes an early disease diagnosis system based on sebum gas analysis, which uses a sebum gas collection and infrared spectrum enhancement device prepared by combining a micro-electro-mechanical system (MEMS), a new two-dimensional material, and metal surface plasmon resonance technology to collect trace human sebum gases and enhance their infrared spectra by metal plasmon, and then inputs the metal plasmon-enhanced infrared spectrum of the sebum gas into an early disease diagnosis model. After analysis by the early disease diagnosis model, the early diagnosis of whether the diagnosed person is ill is finally realized.

[0008] To achieve the above object, the technical solution of the present invention is as follows:

[0009] A disease early diagnosis system based on sebum gas analysis includes a sebum gas collection and its infrared spectrum enhancement device, a microscopic infrared spectrometer, a disease early diagnosis model, and a computer. The sebum gas collection and its infrared spectrum enhancement device is attached to the forehead or back of the human body for 3 - 5 minutes to collect human sebum gas. Then, the microscopic infrared spectrometer is used to scan the plasma-enhanced infrared spectrum of the sebum gas. Next, the plasma-enhanced infrared spectrum of the sebum gas is input into the disease early diagnosis model installed in the computer, and after analysis by the disease early diagnosis model, a diagnosis result of whether the person is ill is output.

[0010] The sebum gas collection and its infrared spectrum enhancement device includes a fixing band and a fixing frame provided on the fixing band. A nano gas sensor core is provided inside the fixing frame.

[0011] The nano gas sensor core includes a substrate. The material of the substrate is an optical window material with an infrared transmittance of more than 90%, preferably CaF2 crystal, to ensure that the nano gas sensor core can transmit infrared light and maintain an infrared transmittance of more than 85%. A gas-sensitive two-dimensional material is attached to the upper surface of the substrate to adsorb sebum gas and allow infrared light to pass through. At the same time, a nano metal array is also provided on the upper surface of the substrate, which generates a surface plasmon enhancement effect under the irradiation of infrared light to enhance the infrared spectrum of the sebum gas adsorbed on the two-dimensional material. The upper surface of the nano gas sensor core is adhesively bonded 1 mm - 2 mm below the lower surface inside the fixing frame.

[0012] The nano metal array is a tiny metal array formed by one unit structure or a tiny metal array formed by a combination of multiple different unit structures. The metal is preferably gold or silver. The unit structure is a bowtie structure composed of a pair of isosceles triangles, an elongated rectangular structure, or other structures with a surface plasmon resonance peak in the range of 4.7 μm - 10.5 μm. The height of the unit structure is 80 nm - 120 nm. The above different structures correspond to different surface plasmon resonance peaks, and the surface plasmon resonance of different resonance peaks realizes the enhancement of the infrared spectra of different sebum gases. The tiny metal array formed by a combination of multiple different unit structures has a wider range of surface plasmon resonance peaks than the tiny metal array formed by one unit structure, realizing the enhancement of multiple characteristic peaks of the infrared spectrum of sebum gas.

[0013] The fixing frame is also provided with a breathable protective layer, preferably a breathable protective gauze adhesively bonded to the outer edge of the fixing frame, to prevent the skin from coming into contact with the nano gas sensor core during human testing and prevent the infiltration of impurities.

[0014] On the upper surface of the nano gas-sensing core substrate, a single layer or multiple layers of two-dimensional materials are attached. The two-dimensional materials are molybdenum disulfide, graphene, carbon nanotubes, or other two-dimensional materials with gas-sensing properties, enabling the nano gas-sensing core to adsorb sebum gas and increase the sensitivity and stability of gas adsorption.

[0015] The fixing band is preferably a long strip of tape, making the shape and size of the sebum gas collection and infrared spectrum enhancement device similar to a band-aid for convenient use.

[0016] The early disease diagnosis model uses neural network, principal component regression, partial least squares regression, kernel method, random forest, deep learning, or other effective spectral analysis methods and is established through the following steps:

[0017] 1) Use the sebum gas collection and infrared spectrum enhancement device to collect more than 5000 sebum gas samples including patients with a certain disease and those without the disease to establish a sample database;

[0018] 2) When the number of gas samples of patients in step 1) is less than 2000, use the method of making extended samples to extend the collected sebum gas samples of patients;

[0019] 3) Select 80% of the samples in the sample database for training and the remaining 20% of the samples for testing, optimize the model parameters, and complete the construction of the early disease diagnosis model.

[0020] The specific method for extending samples in step 2) is as follows: Remove the nano gas-sensing core of the sebum gas collection and infrared spectrum enhancement device that has adsorbed the sebum gas of the patient, place its lower surface on the heating plate in the airtight heatable gas chamber, and make the nano metal array above the central hole of the heating plate. Place the airtight heatable gas chamber on the sample stage of the microscopic infrared spectrometer, and make the infrared light of the microscopic infrared spectrometer penetrate the upper and lower infrared-transmitting window plates of the heatable gas chamber and the central hole of the heating plate. Open the exhaust valve of the heatable gas chamber, introduce nitrogen with a purity of 99.99% into the heatable gas chamber for 2 - 3 minutes of cleaning purge, then control the temperature of the heating plate in the heatable gas chamber to rapidly rise linearly to 100 °C through the heating controller, and then stepwise rise at a step of 10 °C to 110 °C, 120 °C, 130 °C, 140 °C, 150 °C, 160 °C, 170 °C, 180 °C, 190 °C. While starting the heating controller, introduce nitrogen with a purity of 99.99% at 150 sccm into the heatable gas chamber. Obtain extended samples of the original samples at different stepwise temperatures and directly scan the metal plasma-enhanced infrared spectra of the extended samples.

[0021] The present invention has the following beneficial effects:

[0022] 1) The sebum gas collection and its infrared spectrum enhancement device is used to collect sebum gas, and then the gas sample is taken to the laboratory for analysis to obtain a diagnostic result, providing a truly non-invasive diagnostic method for disease diagnosis;

[0023] 2) The early disease diagnosis system based on sebum gas analysis can diagnose whether a person has a disease in the pre-symptomatic stage, providing a basis for the diagnosis and treatment of diseases in the early stage and precise medicine;

[0024] 3) The key component of the early disease diagnosis system based on sebum gas analysis is the sebum gas collection and its infrared spectrum enhancement device. Its shape and size are similar to those of a conventional band-aid. The process of collecting human sebum gas is as simple and efficient as operating a band-aid, and it causes no pollution to the environment and ecological damage. Brief Description of the Drawings

[0025] Figure 1 It is a schematic diagram of the composition of the early disease diagnosis system based on sebum gas analysis of the present invention;

[0026] Figure 2 It is an exploded view of the sebum gas collection and its infrared spectrum enhancement device of the present invention;

[0027] Figure 3 It is a three-dimensional structure schematic diagram of the nano gas sensor core of the sebum gas collection and its infrared spectrum enhancement device of the present invention;

[0028] Figure 4 It is a schematic diagram of the nano metal array of Embodiment 1 of the present invention;

[0029] Figure 5 It is a comparison diagram of the metal plasma enhanced infrared spectra of sebum gas of PD patients and healthy young people without PD in Embodiment 1 of the present invention;

[0030] Figure 6 It is a framework diagram of the GADF coupled CBAM-CNN deep learning network of Embodiment 1 of the present invention;

[0031] Figure 7 It is a two-dimensional color picture of the infrared spectrum of sebum gas of PD patients of the present invention after GADF transformation;

[0032] Figure 8 It is an iterative process diagram of training the CBAM-CNN model in Embodiment 1 of the present invention.

[0033] In the figure, 1 - sebum gas collection and its infrared spectrum enhancement device, 2 - microscopic infrared spectrometer, 3 - early disease diagnosis model, 4 - computer, 5 - breathable protective layer, 6 - fixed frame, 7 - nano gas sensor core, 8 - fixing band, 9 - nano metal array, 10 - two-dimensional material, 11 - substrate. Detailed Embodiments

[0034] The detailed technical solution of the present invention will be introduced below in conjunction with the accompanying drawings:

[0035] As Figure 1 shown, the early disease diagnosis system based on sebum gas analysis includes a sebum gas collection and its infrared spectrum enhancement device 1, a microscopic infrared spectrometer 2, an early disease diagnosis model 3, and a computer 4; the sebum gas collection and its infrared spectrum enhancement device 1 is attached to the forehead or back of the human body for 3 - 5 minutes, and then the microscopic infrared spectrometer 2 is used to scan the plasma-enhanced infrared spectrum of the sebum gas, and then the plasma-enhanced infrared spectrum of the sebum gas is input into the early disease diagnosis model 3 installed in the computer 4, and the diagnosis result of whether the person is ill is output after analysis by the early disease diagnosis model.

[0036] As Figure 2 shown, the sebum gas collection and its infrared spectrum enhancement device 1 includes a fixing band 8 and a fixing frame 6 provided on the fixing band, and a nano gas sensor core 7 is provided inside the fixing frame 6; a breathable protective layer 5 is further provided on the upper surface of the fixing frame 6, specifically a breathable protective gauze bonded to the outer edge of the fixing frame.

[0037] As Figure 3 shown, the nano gas sensor core includes a substrate 11, the material of the substrate 11 is an optical window material with an infrared transmittance of more than 90%, a single-layer or multi-layer gas-sensitive two-dimensional material 10 is attached to the upper surface of the substrate 11, and at the same time, a nano metal array 9 is further provided on the upper surface of the substrate 11; the nano metal array 9 is a tiny metal array formed by one unit structure or a tiny metal array formed by a combination of multiple different unit structures, the metal is preferably gold or silver, the unit structure is a bow-tie structure composed of a pair of isosceles triangles, a slender rectangular structure or other structures with a plasma resonance peak in the range of 4.7μm - 10.5μm, and the height of the unit structure is 80nm - 120nm.

[0038] The early disease diagnosis model is established by using neural network, principal component regression, partial least squares regression, kernel method, random forest, deep learning or other effective spectral analysis methods.

[0039] Example 1

[0040] As Figure 1 shown, in this example, taking the diagnosis of Parkinson's disease (PD) as an example, an early disease diagnosis system based on sebum gas analysis is provided, including a sebum gas collection and its infrared spectrum enhancement device 1, a microscopic infrared spectrometer 2, an early disease diagnosis model 3, and a computer 4;

[0041] As Figure 2As shown, the sebum gas collection and infrared spectrum enhancement device 1 includes a fixing band 8 and a fixing frame 6 provided on the fixing band. A nano gas sensor core 7 is provided inside the fixing frame 6. Among them, the fixing frame 6 is a square hollow structure with a height of 6 mm and a wall thickness of 2 mm, and its internal size is 5.5 mm × 5.5 mm. The fixing frame 6 is also provided with a breathable protective layer, and the breathable protective layer is a square sterile breathable gauze of 9.5 mm × 9.5 mm, and its edge is aligned with the outer edge of the fixing frame 6 and adhesively bonded to the upper surface of the fixing frame 6. The fixing band 8 is a long strip-shaped adhesive tape with glue on both ends of the upper surface and no glue on the middle part of the upper surface and the lower surface. The part without glue in the middle of the upper surface is a square of 10 mm × 10 mm. The nano gas sensor core 7 is adhesively fixed to the middle part of the upper surface of the fixing band without glue on the lower surface of the fixing frame 8.

[0042] As Figure 3 shown, the nano gas sensor core 7 includes a substrate 11. The material of the substrate 11 is an optical window material with an infrared transmittance of more than 90%. A single layer or multiple layers of gas-sensitive two-dimensional materials 10 are attached to the upper surface of the substrate 11. At the same time, a nano metal array 9 is also provided on the upper surface of the substrate 11.

[0043] In this embodiment, the substrate 11 of the nano gas sensor core 7 is a square CaF2 crystal with a length, width and height of 5 mm, 5 mm and 0.5 mm respectively and double-sided polished. The gas-sensitive two-dimensional material is monolayer molybdenum disulfide. The nano metal array 9 is processed on CaF2 by electron beam lithography. The upper surface of the nano gas sensor core 7 is adhesively fixed upward inside the fixing frame 6 at a position 1.5 mm away from the lower surface of the fixing frame 6.

[0044] In this embodiment, according to the infrared spectra of 4 organic gases (hippuric acid, perillaldehyde, eicosane and octadecane) closely related to PD reported by Dr. Tilo Kunath, it is determined that the main infrared spectrum peak of the mixed gas formed by the mixture of hippuric acid, perillaldehyde, eicosane and octadecane appears between 5.85 μm - 6.18 μm. Accordingly, the corresponding nano metal array 9 is constructed. As Figure 4 shown, the nano metal array 9 is composed of 4 different bow-shaped unit structures. The metal is gold, and the thickness of the gold is 80 nm. The bottom length L of the triangle in the bow is 80 nm, 90 nm, 100 nm and 116 nm respectively, the height H is 100 nm, 120 nm, 135 nm and 150 nm respectively, the distance b between the vertices of the two triangles is 46 nm, 75 nm, 95 nm and 120 nm respectively, the internal horizontal spacing T of the 4 bow arrays is 60 nm, 100 nm, 300 nm and 500 nm respectively, the internal vertical spacing S of the 4 bow arrays is 1 μm, and the horizontal spacing T between the 4 bow arrays h is 600 nm, and the vertical spacing S between the 4 bow arraysL is 2 μm.

[0045] Different from the narrow-peak spectrum enhanced by conventional single-unit structure metal plasma, the nano-metal array adopted in this embodiment obtains a wide-peak metal plasma enhanced infrared spectrum of 5.88 μm - 6.09 μm. As Figure 5 shown, after adsorbing the gas of sebum from PD patients, spectral enhancement appears between 5.95 μm - 6.08 μm. At the same time, slight secondary peak enhancement also appears between 5.85 μm - 5.88 μm. After adsorbing the gas of sebum from healthy young people without PD, only slight spectral enhancement appears between 5.86 μm - 5.89 μm, and no spectral enhancement appears between 5.95 μm - 6.08 μm.

[0046] The early disease diagnosis model is constructed for the early diagnosis of PD to achieve the early diagnosis of PD. Since the components of sebum gas are complex and there is cross-sensitivity between the infrared spectra of PD patients and those of people without PD ( Figure 5 in the range of 5.86 μm - 5.88 μm). Therefore, in order to eliminate the cross-sensitivity of the spectra and further extract the characteristic information of the spectra, the early disease diagnosis model of this embodiment is based on 12,188 sebum gas samples and adopts Figure 6Construction of a convolutional neural network (CBAM-CNN) with attention mechanism introduced by the Gram angle difference field (GADF) as shown. First, spectral data is transformed into a two-dimensional color image through GADF, and then through convolution (7×7, Conv), max pooling (3×3, Maxpool), CBAM attention module, residual block (Residualblock), CBAM attention module, average pooling (1×1, Avgpool), and fully connected layer (512, FC) to finally achieve the classification of spectral data. 12,188 sebum gas samples were obtained by collecting and expanding the sebum gas samples of 1,862 people using a sebum gas collection and infrared spectrum enhancement device for sebum gas. Among them, there were 6,088 sebum gas samples of PD patients who had been diagnosed, and 6,100 sebum gas samples of people without PD; among the 1,862 people, there were 526 PD patients who had been diagnosed and 1,336 people without PD. Among them, the 526 PD patients included 393 male patients and 133 female patients respectively, with the patient age ranging from 56 years old to 83 years old and the disease duration ranging from 0.5 year to 8 years; the 1,336 people without PD included 720 male and 616 female respectively, with the age covering 7 years old to 35 years old; each of the 526 PD patients provided at least 20 different sebum gas samples, and the 20 different samples were collected before and 1 hour after breakfast, lunch, and dinner on 2 days, 1 hour after taking morning and evening medications, before and after skin cleaning, and also included samples collected after aerobic exercise for individual patients and extended samples of some samples; each of the 1,336 people without PD provided at least 6 different sebum gas samples, which were collected before and after breakfast and dinner on 2 days and 1 hour after meals, before and after skin cleaning; the metal plasma enhanced infrared spectra of all 12,188 sebum gas samples were scanned with a microscopic infrared spectrometer and transformed into two-dimensional color images using the GADF method; 4,870 and 4,880 images were randomly selected from the GADF-transformed two-dimensional color images of 6,088 PD patients and 6,100 people without PD respectively to train the CBAM-CNN deep learning network, and the remaining 1,218 and 1,220 two-dimensional color images transformed by the GADF method were used for testing. Finally, the construction of the PD early diagnosis model was completed through parameter optimization.

[0047] As an example, Figure 7 The two-dimensional color image transformed by the GADF method for the sebum gas sample of a PD patient is given. When the relevant parameters for training the CBAM-CNN deep learning network are batch size = 16, optimizer is adam, learning rate = 0.0001, and the number of training epochs = 25, the model accuracy reaches 99.1%, as Figure 8 shown; the test accuracy for 2,438 test samples is 98.2%.

[0048] The method for preparing the extended sample of the sample is as follows: Remove the nano-gas sensor core on the sebum gas collection and infrared spectrum enhancement device that has adsorbed the sebum gas of the PD patient, place its lower surface on the heating plate in the sealed heatable gas chamber, and make the nano-metal array located above the central hole of the heating plate. Place the sealed heatable gas chamber on the sample stage of the microscopic infrared spectrometer, and make the infrared light of the microscopic infrared spectrometer penetrate the upper and lower infrared-transmitting window plates of the heatable gas chamber and the central hole of the heating plate. Open the exhaust valve of the heatable gas chamber, introduce nitrogen with a purity of 99.99% into the heatable gas chamber to clean and purge the gas chamber for 2 - 3 minutes. Then, control the temperature of the heating plate in the heatable gas chamber to linearly rise to 100 °C through the heating controller, and then stepwise rise by 10 °C steps to 110 °C, 120 °C, 130 °C, 140 °C, 150 °C, 160 °C, 170 °C, 180 °C, 190 °C. While starting the heating controller, introduce nitrogen with a purity of 99.99% at 150 sccm into the heatable gas chamber. Obtain the extended sample of the original sample at the corresponding temperature under different stepwise temperatures, and directly scan the metal plasma-enhanced infrared spectrum of the extended sample.

[0049] When collecting sebum gas, attach the air-permeable protective layer of the sebum gas collection and infrared spectrum enhancement device to the back of the person to be diagnosed, and fix the sebum gas collection and infrared spectrum enhancement device on the human body through the fixing strap, and keep it for 5 minutes to complete the collection of the sebum gas of the person to be diagnosed. Then, take out the nano-gas sensor core on the sebum gas collection and infrared spectrum enhancement device, turn its upper surface upward, scan the infrared spectrum of the nano-gas sensor core that has adsorbed the sebum gas of the person to be diagnosed using a microscopic infrared spectrometer, and input the scanned spectrum into the early disease diagnosis model for early PD diagnosis. The early disease diagnosis model for early PD diagnosis outputs the classification result of whether the person has PD, and finally realizes the early diagnosis of whether the person to be diagnosed has PD.

Claims

1. A disease early diagnosis system based on sebum gas analysis, characterized in that, It includes a sebum gas collection and its infrared spectrum enhancement device, a microscopic infrared spectrometer, and a computer installed with an early disease diagnosis model; the sebum gas collection and its infrared spectrum enhancement device includes a fixed frame, and a nano gas sensor core is arranged inside the fixed frame. The nano gas sensor core includes a substrate, and a single-layer or multi-layer gas-sensitive two-dimensional material is attached to the upper surface of the substrate. At the same time, a nano metal array is also arranged on the upper surface of the substrate; the sebum gas collection and its infrared spectrum enhancement device is attached to the forehead or back of the human body for 3 - 5 minutes to collect the human sebum gas, and then the microscopic infrared spectrometer is used to scan the plasma-enhanced infrared spectrum of the sebum gas. Then, the plasma-enhanced infrared spectrum of the sebum gas is input into the early disease diagnosis model installed in the computer, and the diagnosis result of whether the person is ill is output after being analyzed by the early disease diagnosis model.

2. The early disease diagnosis system based on sebum gas analysis according to claim 1, characterized in that, The sebum gas collection and its infrared spectrum enhancement device includes a fixing band, and the fixed frame is arranged on the fixing band.

3. The early disease diagnosis system based on sebum gas analysis according to claim 2, wherein The material of the substrate is an optical window material with an infrared transmittance of more than 90%.

4. The early disease diagnosis system based on sebum gas analysis according to claim 3, characterized in that, The material of the substrate is CaF2 crystal.

5. The early disease diagnosis system based on sebum gas analysis according to claim 3, characterized in that, The nano metal array is a tiny metal array formed by one unit structure, or a tiny metal array formed by a combination of multiple different unit structures. The metal is gold or silver. The unit structure is a structure with a plasma resonance peak at 4.7 μm - 10.5 μm, and the height of the unit structure is 80 nm - 120 nm.

6. The early disease diagnosis system based on sebum gas analysis according to claim 5, wherein The unit structure is a bow-tie structure composed of a pair of isosceles triangles or an elongated rectangular structure.

7. The early disease diagnosis system based on sebum gas analysis according to claim 2, characterized in that A breathable protective layer is also arranged on the fixed frame, specifically, breathable protective gauze adhered to the outer edge of the fixed frame.

8. The early disease diagnosis system based on sebum gas analysis according to claim 3, characterized in that, The two-dimensional material attached to the upper surface of the nano gas sensor core substrate is molybdenum disulfide, graphene, or carbon nanotubes.

9. The early disease diagnosis system based on sebum gas analysis according to claim 2, characterized in that, The fixing band is a long strip of adhesive tape.

10. The early disease diagnosis system based on sebum gas analysis according to claim 1, characterized in that, The early disease diagnosis model adopts neural network, principal component regression, partial least squares regression, kernel method, random forest, or deep learning, and is established by the following steps: 1) Use the sebum gas collection and its infrared spectrum enhancement device to collect more than 5000 sebum gas samples including patients with a certain disease and those without the disease, and establish a sample database; 2) When the number of gas samples of patients in step 1) is less than 2000, use the method of making extended samples to extend the collected sebum gas samples of patients; 3) Select 80% of the samples in the sample database for training, and the remaining 20% of the samples for testing, optimize the model parameters, and complete the construction of the early disease diagnosis model.

11. The early disease diagnosis system based on sebum gas analysis according to claim 10, characterized in that, The sample expansion method in step 2) is specifically as follows: Remove the nano-gas sensor core of the sebum gas collection and infrared spectrum enhancement device adsorbed with the patient's sebum gas, place its lower surface on the heating plate in the sealed and heatable gas chamber, and make the nano-metal array above the central hole of the heating plate. Place the sealed and heatable gas chamber on the sample stage of the microscopic infrared spectrometer, and make the infrared light of the microscopic infrared spectrometer penetrate the upper and lower infrared transmission window plates of the heatable gas chamber and the central hole of the heating plate. Open the exhaust valve of the heatable gas chamber, introduce nitrogen with a purity of 99.99% into the heatable gas chamber to clean and purge the gas chamber for 2 - 3 minutes. Then, control the temperature of the heating plate in the heatable gas chamber to rapidly rise linearly to 100 °C through the heating controller, and then stepwise increase the temperature in 10 °C steps to 110 °C, 120 °C, 130 °C, 140 °C, 150 °C, 160 °C, 170 °C, 180 °C, 190 °C. While starting the heating controller, introduce 150 sccm of nitrogen with a purity of 99.99% into the heatable gas chamber. Obtain the expanded samples of the original samples at the corresponding temperatures at different stepwise temperatures, and directly scan the metal plasma enhanced infrared spectra of the expanded samples.

Images