MOF (Metal Organic Framework)-derived heterojunction as well as preparation method and application thereof in metabolism detection

Abstract

The invention discloses an MOF-derived heterojunction, a preparation method thereof and application of the MOF-derived heterojunction in metabolism detection, and relates to the technical field of heterojunction material preparation. The preparation method comprises the following steps: performing first-stage heat treatment on an MOF precursor to prepare a metal oxide, and reacting the metal oxide with a titanium source to obtain a product; or directly reacting the MOF precursor with a titanium source to obtain a product; and carrying out second-stage heat treatment on the product in an inert atmosphere or air to obtain the MOF-derived heterojunction. The invention relates to application of MOF-derived heterojunction as a matrix in detection of serum metabolic fingerprints by laser desorption / ionization mass spectrometry and application in construction of a disease diagnosis model. The heterojunction material prepared by the invention has the advantages of high ionization efficiency, good photothermal effect and strong biological matrix interference resistance, and can be used as an LDI MS nano matrix to be applied to respiratory disease serum metabolism fingerprint diagnosis and disease diagnosis models so as to distinguish respiratory diseases.

Description

Technical Field

[0001] This invention relates to the field of heterojunction material preparation technology, specifically to MOF-derived heterojunctions, their preparation methods, and their applications in metabolic detection. Background Technology

[0002] Laser desorption / ionization mass spectrometry (LDI MS) has shown great potential in metabolomics analysis due to its high throughput, high sensitivity, and ease of operation. The performance of this technique critically depends on the matrix material used. Inorganic nanomaterials, such as titanium dioxide (TiO2), have become ideal candidates to replace traditional organic matrices due to their high stability, low background interference, and strong ultraviolet absorption. However, existing TiO2 nanomatrices still have significant technical shortcomings and limitations in LDI MS applications: ① High charge recombination rate: As a semiconductor material, TiO2's photogenerated electron-hole pairs recombine very easily and rapidly, which severely limits its photoinduced charge transfer efficiency and final ionization efficiency when used as a matrix, resulting in weak metabolite detection signals.

[0003] ②Low photothermal conversion efficiency: The photothermal effect of TiO2 material itself is weak, which is not conducive to the effective thermal desorption of analytes under laser irradiation, further reducing the detection sensitivity.

[0004] ③ Poor resistance to biological matrix interference: Serum and other complex biological samples contain high concentrations of salt and protein, which can severely inhibit the ionization process of analytes and produce matrix effects. Conventional TiO2 matrices lack effective resistance to this interference and are difficult to use directly for high-throughput detection of clinical serum samples.

[0005] Therefore, developing an LDI MS nanomatrix that can simultaneously possess efficient charge separation, strong photothermal effect, and excellent salt / protein resistance is crucial for promoting the application of metabolomics in clinical precision diagnosis. Summary of the Invention

[0006] The purpose of this invention is to at least solve one of the technical problems existing in the prior art, and to provide a MOF-derived heterojunction, its preparation method and its application in metabolic detection. Specifically, it provides a metal-organic framework (MOF)-derived metal oxide / TiO2 heterojunction material, its preparation method and its application in metabolic detection. The MOF-derived metal oxide / TiO2 heterojunction material of this invention has high ionization efficiency, good photothermal effect and strong resistance to biological matrix interference, and can be used as an LDI MS nanomatrix and applied in the diagnosis of serum metabolic fingerprints for respiratory diseases.

[0007] This invention uses MOF as a precursor. A first-stage calcination or hydrothermal treatment of the MOF generates MOF-corresponding metal oxide nanoparticles. These MOF-corresponding metal oxide nanoparticles are then reacted with a titanium source, followed by a second-stage calcination to prepare MOF-derived metal oxide / TiO2 heterojunction materials, such as hollow-structured Co3O4 / TiO2 heterojunction nanomaterials, ZnO / TiO2 heterojunction nanomaterials, Fe2O3 / TiO2 heterojunction nanomaterials, Cr2O3 / TiO2 heterojunction nanomaterials, and CuO / TiO2 heterojunction nanomaterials. In these materials, a tight heterojunction interface is formed between the MOF-derived metal oxides (Co3O4, ZnO, Fe2O3, Cr2O3, CuO) and TiO2.

[0008] The MOF-derived metal oxide / TiO2 heterojunction material prepared by this invention has high ionization efficiency, good photothermal effect, and strong resistance to biological matrix interference. It can be used as an LDI MS nanomatrix and applied in the diagnosis of respiratory diseases using serum metabolic fingerprints. Furthermore, this invention combines the serum metabolic fingerprint of respiratory diseases with machine learning algorithms to construct a diagnostic model for respiratory diseases and / or a lung cancer staging model. This model is used to distinguish healthy controls (HC) from patients with bronchial asthma (BA), chronic obstructive pulmonary disease (COPD), interstitial lung disease (ILD), and lung cancer (LCa), and further staging lung cancer.

[0009] The technical solution of the present invention is as follows: In a first aspect, the present invention provides a method for preparing a MOF-derived heterojunction, the method comprising the following steps: S1. The MOF precursor is prepared into a metal oxide by a first-stage heat treatment, and then the metal oxide is reacted with a titanium source to obtain the product; or, the MOF precursor is reacted directly with a titanium source without the first-stage heat treatment to obtain the product. S2. The product is subjected to a second-stage heat treatment in an inert atmosphere or air to obtain a MOF-derived heterojunction, wherein the MOF-derived heterojunction is a MOF-derived metal oxide / TiO2 heterojunction.

[0010] Optionally, the MOF precursor includes at least one of ZIF-67, ZIF-8, MIL-101(Fe), MIL-101(Cr), and HKUST-1; The MOF-derived metal oxide / TiO2 heterojunction includes at least one of Co3O4 / TiO2 heterojunction materials, ZnO / TiO2 heterojunction materials, Fe2O3 / TiO2 heterojunction materials, Cr2O3 / TiO2 heterojunction materials, and CuO / TiO2 heterojunction materials. Among these, the Co3O4 / TiO2 heterojunction material has a hollow structure.

[0011] Optionally, when the MOF precursor is ZIF-67 or MIL-101(Cr), the preparation method includes the following steps: S1. ZIF-67 or MIL-101(Cr) is subjected to a first-stage heat treatment in air to obtain metal oxide nanoparticles; the metal oxide nanoparticles are dispersed in a dilute acid aqueous solution, a solution containing a titanium source is added, and the reaction is stirred to obtain the product. S2. The product is subjected to a second-stage heat treatment in an inert atmosphere to obtain a MOF-derived heterojunction; When the MOF precursor is MIL-101(Fe), the preparation method includes the following steps: S1. Disperse MIL-101(Fe) in a dilute acid aqueous solution, add a solution containing a titanium source, stir the reaction, and obtain the product; S2. The product is subjected to a second-stage heat treatment in an inert atmosphere to obtain a MOF-derived heterojunction.

[0012] Optionally, the first stage of heat treatment includes: heating from room temperature to 350℃-550℃ at a heating rate of 1℃ / min-5℃ / min, and holding at 350℃-550℃ for 1h-5h. The second stage of heat treatment includes: heating from room temperature to 350℃-550℃ at a heating rate of 1℃ / min-5℃ / min, and holding at 350℃-550℃ for 2h-5h. The concentration of the solution containing the titanium source is 45 wt%-55 wt%, wherein the titanium source includes bis(ammonium lactic acid) dihydroxytide (TiBALDH). The concentration of the dilute acid aqueous solution is 0.05 M - 0.2 M; The ratio of the metal oxide nanoparticles, dilute acid aqueous solution, and titanium-containing solution is 180 mg-220 mg: 92 mL-100 mL: 3 mL-5 mL.

[0013] Optionally, when the MOF precursor is ZIF-8 or HKUST-1, the preparation method includes the following steps: ZIF-8 or HKUST-1 was dispersed in an organic solvent, a titanium source was added, and a first-stage hydrothermal reaction was carried out to obtain the product. The product was subjected to a second-stage heat treatment in air to obtain a MOF-derived heterojunction.

[0014] Optionally, the organic solvent is anhydrous ethanol; the titanium source is tetrabutyl titanate; the addition ratio of the MOF precursor, organic solvent and titanium source is 80 mg-120 mg: 30 mL-50 mL: 3 mL-5 mL; the first stage hydrothermal reaction includes reacting at 140℃-160℃ for 14h-18h; the second stage heat treatment includes: heating from room temperature to 350℃-550℃ at a heating rate of 1℃ / min-5℃ / min, and calcining at 350℃-550℃ for 2h-5h.

[0015] Secondly, the present invention provides a MOF-derived heterojunction obtained by the preparation method described above, wherein the MOF-derived heterojunction comprises a MOF-derived metal oxide and TiO2, and a tight heterojunction interface is formed between the MOF-derived metal oxide and TiO2.

[0016] Thirdly, the present invention provides the application of the MOF-derived heterojunction as a matrix in the detection of serum metabolic fingerprints by laser desorption / ionization mass spectrometry.

[0017] Optionally, the application includes the use of the MOF-derived heterojunction as a matrix in the diagnosis of respiratory diseases using laser desorption / ionization mass spectrometry detection of serum metabolic fingerprints.

[0018] Optionally, the application specifically includes the following steps: Step 1: Disperse the heterojunction nanomaterial in deionized water to prepare a matrix suspension with a concentration of 1-10 mg / mL.

[0019] Step 2: Mix the clinically collected serum sample with an organic solvent (such as acetonitrile) at a volume ratio of 1:1 to 1:4, vortex and centrifuge to collect the supernatant to precipitate proteins.

[0020] Step 3: Mix the serum supernatant from Step 2 with the matrix suspension from Step 1 at a volume ratio of 1:1 and vortex to mix.

[0021] Step 4: Spot 1-2 μL of the mixture onto the mass spectrometer target plate and dry at room temperature to form co-crystallized spots.

[0022] Step 5: Use LDI MS to collect data from the spots and obtain serum metabolic fingerprint profiles.

[0023] Furthermore, the LDI MS mentioned in step 5 is preferably a MALDI-TOF MS with a laser wavelength of 355 nm and a positive ion reflection mode for acquisition.

[0024] Fourthly, the present invention provides an application of the MOF-derived heterojunction described above in constructing a disease diagnostic model.

[0025] The obtained metabolic fingerprint profiles are combined with machine learning algorithms to construct a disease diagnosis model, which is used to distinguish healthy controls (HC) from patients with bronchial asthma (BA), chronic obstructive pulmonary disease (COPD), interstitial lung disease (ILD), and lung cancer (LCa), and further to stage lung cancer.

[0026] Optionally, the application method includes: The MOF-derived heterojunction was dispersed in deionized water to prepare a matrix suspension. Mix the serum sample with an organic solvent, vortex, and centrifuge to collect the supernatant. The supernatant is mixed with the matrix suspension and vortexed to obtain a mixture; The mixture was spotted onto a mass spectrometry target plate and dried at room temperature to form co-crystallized spots. LDI MS was used to collect data from the spots to obtain serum metabolic fingerprint profiles; The serum metabolic fingerprint was combined with machine learning algorithms to construct a disease diagnosis model.

[0027] Optionally, the disease diagnosis model includes a respiratory disease diagnosis model and / or a lung cancer staging diagnosis model, wherein the respiratory disease includes at least one of bronchial asthma, chronic obstructive pulmonary disease, interstitial lung disease and lung cancer, and the lung cancer staging includes early lung cancer, intermediate lung cancer and late lung cancer; The machine learning algorithms include one or more combinations of neural networks (NN), support vector machines (SVM), K-nearest neighbors (KNN), and Naive Bayes (NB). A diagnostic panel of metabolite features is constructed through feature selection for classification of the five cases mentioned above.

[0028] This invention has at least one of the following beneficial effects: 1. Material Innovation and Performance Enhancement: Through MOF-guided structural engineering, a metal oxide / TiO2 heterojunction was successfully constructed. This metal oxide / TiO2 heterojunction not only provides a large specific surface area to enhance laser absorption and analyte adsorption, but its built-in electric field also effectively suppresses the recombination of photogenerated charges, thereby significantly improving ionization efficiency. Simultaneously, the synergistic effect of the heterojunction enhances the photothermal conversion performance of the material, promoting analyte desorption.

[0029] 2. Excellent anti-interference ability: The heterojunction nanomaterials, especially the hollow CoTi, have a unique molecular sieve effect and pore structure, which can effectively resist the interference of high concentrations of salt and protein in serum, and realize direct, high-throughput analysis of complex biological samples.

[0030] 3. Highly Effective Clinical Diagnostic Value: This invention is the first to apply the CoTi-LDI MS platform to the metabolic fingerprint analysis of serum samples from respiratory diseases. Combined with machine learning, it not only achieves accurate identification of BA, COPD, ILD, and LCa (with a five-category AUC as high as 0.956), but also successfully performs lung cancer staging, providing a powerful tool for early screening and non-invasive diagnosis of respiratory diseases.

[0031] 4. Cost-effectiveness and universality: The preparation method is simple and low-cost. The developed metabolite diagnostic panel greatly improves diagnostic efficiency while ensuring high accuracy, and has good prospects for clinical translation. Attached Figure Description

[0032] Figure 1 The images shown are transmission electron microscope (TEM) images of the CoTi heterojunction prepared in this invention; where A in the image is a TEM image at 200 nm; and B in the image is a TEM image at 10 nm.

[0033] Figure 2 The X-ray diffraction (XRD) patterns of the CoTi heterojunction and its precursor prepared in this invention.

[0034] Figure 3 The bar chart shows the signal intensity comparison of CoTi prepared in this invention with other control matrices for the detection of standard metabolites. In the figure, A is a bar chart comparing the signal intensity of the five matrices prepared in Examples 1-5 for the detection of standard metabolites; B is a bar chart comparing the signal intensity of ZIF-67, Co3O4, TiO2 and CoTi prepared in Example 1 for the detection of standard metabolites.

[0035] Figure 4 The graph shows the signal stability of the CoTi matrix prepared in this invention and other control matrices in detecting metabolite standards under high salt (1 M NaCl) and high protein (5 mg / mL BSA) conditions; where A in the graph represents the high salt environment and B in the graph represents the high protein environment.

[0036] Figure 5 The images show representative spectra of five serum samples (HC, BA, COPD, ILD, LCa) obtained using the CoTi heterojunction matrix in Example 7.

[0037] Figure 6This is a t-SNE dimensionality reduction visualization of five groups of samples (HC, BA, COPD, ILD, LCa) based on the diagnostic feature metabolite feature panel in Example 8.

[0038] Figure 7 The present figure shows the receiver operating characteristic (ROC) curves of the neural network (NN) model in Example 8, which is based on a diagnostic feature metabolite panel and performs five-class classification on an independent validation set. Detailed Implementation

[0039] To make the technical problems solved, the technical solutions, and the beneficial effects of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0040] Example 1: Preparation of Co3O4 / TiO2 heterojunction nanomaterials This embodiment provides a method for preparing a hollow Co3O4 / TiO2 heterojunction nanomaterial (denoted as CoTi), including the following steps: (1) Synthesis of ZIF-67 3.0 mmol of cobalt nitrate hexahydrate was dissolved in 40 mL of methanol, and this solution is denoted as solution A. 12.0 mmol of 2-methylimidazole was dissolved in 40 mL of methanol, and this solution is denoted as solution B. Under room temperature and magnetic stirring, solution B was rapidly poured into solution A, and the mixture immediately turned purple. After stirring continuously for 10 minutes, the mixture was allowed to stand for 24 hours. After the reaction was complete, the purple precipitate was collected by centrifugation (8000 rpm, 5 min), washed three times with methanol, and then dried in a vacuum oven at 60 °C for 12 hours to obtain ZIF-67 crystals.

[0041] (2) Preparation of MOF-derived Co3O4 Take 500 mg of the ZIF-67 powder prepared in step (1), spread it evenly in a porcelain boat, and place it in a tube furnace. Under air atmosphere, heat from room temperature to 450℃ at a heating rate of 2℃ / min, and calcine at 450℃ for 2 hours. After calcination, allow it to cool naturally to room temperature to obtain black powdered Co3O4.

[0042] (3) Preparation of Co3O4 / TiO2 (CoTi) heterojunction nanomaterials Weigh 200 mg of the Co3O4 powder prepared in step (2) and disperse it in 96 mL of 0.1 M dilute hydrochloric acid solution. Sonicate the solution at room temperature for 30 minutes to form a uniform suspension. Then, under vigorous magnetic stirring, slowly add 4 mL of an aqueous solution containing 50 wt% titanate bis(ammonium lactic acid) dihydroxyl (TiBALDH) to the suspension. After the addition is complete, continue stirring at room temperature for 3 hours. After the reaction is complete, centrifuge the mixture (10000 rpm, 10 min) to collect the solid product, and wash it three times each with deionized water and anhydrous ethanol. Place the washed product in a vacuum drying oven at 60 °C and dry overnight. Finally, place the dried precursor powder in a tube furnace and heat it to 400 °C at a rate of 2 °C / min under a nitrogen atmosphere, and heat-treat at this temperature for 5 hours. After heat treatment, the material is cooled to room temperature in the furnace to obtain a hollow Co3O4 / TiO2 heterojunction nanomaterial (denoted as CoTi).

[0043] Figure 1 The image shown is a transmission electron microscope (TEM) image of the CoTi heterojunction prepared in this invention, which shows its unique hollow structure and heterojunction interface.

[0044] Figure 2 The X-ray diffraction (XRD) patterns of the CoTi heterojunction and its precursor prepared in this invention are shown, proving the successful composite of the Co3O4 and TiO2 phases.

[0045] Example 2: Preparation of ZnO / TiO2 heterojunction nanomaterials This embodiment provides a method for preparing ZnO / TiO2 heterojunction nanomaterials (denoted as ZnTi), including the following steps: 100 mg ZIF-8 was dispersed in 40 mL of anhydrous ethanol, 4 mL of tetrabutyl titanate was added, and the mixture was transferred to a 100 mL high-pressure reactor lined with polytetrafluoroethylene. The reaction was carried out at 150 °C for 16 hours.

[0046] After the reaction was completed, the product (ZIF-8@TiO2) was collected by centrifugation. After washing and drying, the product was calcined in air at 500℃ for 3 hours at a rate of 5℃ / min to obtain ZnTi.

[0047] Example 3: Preparation of Fe2O3 / TiO2 heterojunction nanomaterials This embodiment provides a method for preparing Fe2O3 / TiO2 heterojunction nanomaterials (denoted as FeTi), including the following steps: 200 mg MIL-101(Fe) was dispersed in 96 mL of 0.1 M HCl, and 4 mL of TiBALDH aqueous solution (50 wt%) was added. The mixture was stirred at room temperature for 3 hours.

[0048] After the reaction was completed, the product was collected by centrifugation, washed and dried, and then calcined at 500℃ for 2 hours under nitrogen atmosphere by heating at 5℃ / min to obtain FeTi.

[0049] Example 4: Preparation of Cr2O3 / TiO2 heterojunction nanomaterials This embodiment provides a method for preparing Cr2O3 / TiO2 heterojunction nanomaterials (denoted as CrTi), including the following steps: 800 mg of MIL-101(Cr) was calcined in air at 500°C for 3 hours (5°C / min) to obtain Cr2O3.

[0050] The obtained Cr2O3 was redispersed in 96 mL of 0.1 M HCl, and 4 mL of TiBALDH aqueous solution (50 wt%) was added. The mixture was stirred at room temperature for 3 hours.

[0051] After the reaction was completed, the product was collected by centrifugation, washed and dried, and then calcined at 500℃ for 2 hours under nitrogen atmosphere by heating at 5℃ / min to obtain CrTi.

[0052] Example 5: Preparation of CuO / TiO2 heterojunction nanomaterials This embodiment provides a method for preparing CuO / TiO2 heterojunction nanomaterials (denoted as CuTi), including the following steps: 100 mg HKUST-1 was dispersed in 40 mL of anhydrous ethanol, 4 mL of tetrabutyl titanate was added, and the mixture was transferred to a 100 mL polytetrafluoroethylene-lined high-pressure reactor and reacted at 150 °C for 16 hours.

[0053] After the reaction was completed, the product was collected by centrifugation, washed and dried, and then calcined in air at 500℃ for 3 hours at a rate of 5℃ / min to obtain CuTi.

[0054] Example 6: Comparative Evaluation of Matrix Performance (1) Preparation of matrix suspension: The CoTi, ZnTi, FeTi, CrTi, CuTi materials prepared in Examples 1-5, as well as commercial TiO2 nanoparticles, ZIF-67 prepared in Example 1 and MOF-derived Co3O4, were dispersed in deionized water and prepared into matrix suspensions with a concentration of 2 mg / mL.

[0055] (2) Detection of standard metabolites: A mixed standard solution containing serine (Ser), phenylalanine (Phe), tryptophan (Trp), valine (Val), lysine (Lys), and lactose (Lac) was prepared (each substance concentration was 5 mM). 1 μL of matrix suspension was mixed with 1 μL of standard solution on a mass spectrometry target plate and dried. Data were acquired using a Bruker UltrafleXtreme MALDI-TOF mass spectrometer (355 nm Nd:YAG laser) in positive ion reflectance mode.

[0056] (3) Salt tolerance and protein tolerance test: The above standard metabolite mixture was dissolved in 1 M NaCl solution and 5 mg / mL bovine serum albumin (BSA) solution, respectively. After mixing with each matrix suspension at a 1:1 volume ratio, the mixture was then tested on the target.

[0057] (4) Photocurrent response test: The transient photocurrent response of CoTi and other control matrices under periodic illumination was tested on a CHI760E electrochemical workstation with a 500 W xenon lamp as the light source.

[0058] Figure 3 The image shows a bar chart comparing the signal intensity of the CoTi matrix with other control matrices for the detection of standard metabolites. Figure 3 This demonstrates that the heterojunction material prepared by this strategy can be successfully used for metabolite detection.

[0059] Figure 4 The image shows the signal stability of the CoTi matrix and other control matrices in detecting metabolite standards under high salt (1 M NaCl) and high protein (5 mg / mL BSA) conditions. Figure 4 This demonstrates that CoTi possesses strong salt resistance and resistance to protein interference, making it effective for serum metabolite detection.

[0060] Example 7: Application of CoTi matrix in laser desorption / ionization mass spectrometry (LDI MS) for detecting serum metabolic fingerprints Serum sample pretreatment and mass spectrometry data acquisition Clinical serum samples were collected from collaborating hospitals, including healthy controls (HC) and samples from patients with clinically diagnosed bronchial asthma (BA), chronic obstructive pulmonary disease (COPD), interstitial lung disease (ILD), and lung cancer (LCa). The collection and use of all samples were approved by the relevant ethics committees. Samples were stored at -80°C.

[0061] Pretreatment procedure: Take 100 μL of thawed serum sample, add 300 μL of pre-cooled acetonitrile, and vortex for 3 minutes to fully precipitate proteins. Then centrifuge at 13000 rpm for 15 minutes at 4°C. Take 200 μL of supernatant for subsequent mass spectrometry analysis.

[0062] Mass spectrometry data acquisition: The treated serum supernatant was mixed with the CoTi matrix suspension (5 mg / mL) prepared in Example 1 at a volume ratio of 1:1. After vortexing, 1 μL of the mixture was spotted onto a MALDI target plate and dried at room temperature to form a uniform spot. Data acquisition was performed using a Bruker UltrafleXtreme MALDI-TOF mass spectrometer. The parameters were set as follows: positive ion reflectance mode, mass scan range mass-to-charge ratio (m / z) 50-1000, laser frequency 1000 Hz, laser energy 70%, laser focus size set to "small". 1000 laser scans were accumulated at randomly selected locations on each spot. Three technically replicate spots were prepared and acquired independently for each serum sample to obtain a serum metabolic fingerprint, as shown below. Figure 5 As shown.

[0063] Example 8: Application of CoTi matrix in constructing disease diagnostic models Based on Example 7, machine learning model construction and diagnostic performance verification were performed.

[0064] (1) Data preprocessing and feature extraction: All samples (including HC, BA, COPD, ILD, LCa) were randomly divided into a training set (80%) and an independent validation set (20%). Peak identification of the raw mass spectrometry data was performed using Bruker FlexAnalysis software (signal-to-noise ratio S / N>3). The mass spectrometry peaks of all samples in the training set were aligned to form a peak list containing all possible m / z features. Subsequently, peak area extraction and total ion current (TIC) normalization were performed on the mass spectrometry data of all samples to finally obtain a data matrix with rows of samples and columns representing the m / z feature peak intensities.

[0065] (2) Full-spectrum model construction: The preprocessed training set data matrix was imported into the Python data mining platform. Four algorithms were used to construct a five-class classification model: neural network (NN, with 100 hidden neurons), support vector machine (SVM, radial basis function kernel), K-nearest neighbor (KNN, k=5), and Naive Bayes (NB). The five-class classification model was optimized on the training set using 10-fold cross-validation.

[0066] (3) Construction and Validation of Diagnostic Feature Panel: To improve the clinical applicability of the model, features were selected from the training set data. The selection criteria were: in 10-fold cross-validation, the frequency of the feature being selected into the model was greater than 70%, and the p-value of the one-way ANOVA was less than 0.05. Finally, 23 key mass-to-charge ratio (m / z) features were selected (as shown in Table 1) to form the diagnostic feature panel. Based on these 23 features, the NN model was retrained to obtain a simplified model.

[0067] Table 1 shows the 23 key m / z features selected. Figure 6 The image shows a t-SNE dimensionality reduction visualization of five groups of samples (HC, BA, COPD, ILD, LCa) based on a diagnostic feature panel, which shows obvious inter-group separation.

[0068] Figure 7 The image shows the receiver operating characteristic (ROC) curves of a neural network (NN) model performing five-class classification based on a diagnostic feature panel on the training set and independent validation sets. Figure 7 As can be seen, the simplified model has an AUC of 0.950 in the five-class classification on the training set and maintains excellent performance in the independent validation set with an AUC of 0.956, achieving an overall accuracy of 89.7%. This indicates that the diagnostic feature panel can fully represent the metabolic differences among the five classes of samples and has strong diagnostic discriminative power.

[0069] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A method for preparing a MOF-derived heterojunction, characterized in that, The preparation method includes the following steps: S1. The MOF precursor is prepared into a metal oxide by a first-stage heat treatment, and then the metal oxide is reacted with a titanium source to obtain the product; or, the MOF precursor is reacted directly with a titanium source without the first-stage heat treatment to obtain the product. S2. The product is subjected to a second-stage heat treatment in an inert atmosphere or air to obtain a MOF-derived heterojunction, wherein the MOF-derived heterojunction is a MOF-derived metal oxide / TiO2 heterojunction.

2. The preparation method according to claim 1, characterized in that, The MOF precursor includes at least one of ZIF-67, ZIF-8, MIL-101(Fe), MIL-101(Cr) and HKUST-1; The MOF-derived metal oxide / TiO2 heterojunction includes at least one of Co3O4 / TiO2 heterojunction, ZnO / TiO2 heterojunction, Fe2O3 / TiO2 heterojunction, Cr2O3 / TiO2 heterojunction, and CuO / TiO2 heterojunction.

3. The preparation method according to claim 2, characterized in that, When the MOF precursor is ZIF-67 or MIL-101(Cr), the preparation method includes the following steps: S1. ZIF-67 or MIL-101(Cr) is subjected to a first-stage heat treatment in air to obtain metal oxide nanoparticles; the metal oxide nanoparticles are dispersed in a dilute acid aqueous solution, a solution containing a titanium source is added, and the reaction is stirred to obtain the product. S2. The product is subjected to a second-stage heat treatment in an inert atmosphere to obtain a MOF-derived heterojunction; When the MOF precursor is MIL-101(Fe), the preparation method includes the following steps: S1. Disperse MIL-101(Fe) in a dilute acid aqueous solution, add a solution containing a titanium source, stir the reaction, and obtain the product; S2. The product is subjected to a second-stage heat treatment in an inert atmosphere to obtain a MOF-derived heterojunction.

4. The preparation method according to claim 3, characterized in that, The first stage of heat treatment includes: heating from room temperature to 350℃-550℃ at a heating rate of 1℃ / min-5℃ / min, and holding at 350℃-550℃ for 1h-5h. The second stage of heat treatment includes: heating from room temperature to 350℃-550℃ at a heating rate of 1℃ / min-5℃ / min, and holding at 350℃-550℃ for 2h-5h. The concentration of the solution containing the titanium source is 45 wt%-55 wt%, wherein the titanium source includes titanate bis(ammonium lactic acid) dihydroxyl. The concentration of the dilute acid aqueous solution is 0.05 M - 0.2 M; The ratio of the metal oxide nanoparticles, dilute acid aqueous solution, and titanium-containing solution is 180 mg-220 mg: 92 mL-100 mL: 3 mL-5 mL.

5. The preparation method according to claim 2, characterized in that, When the MOF precursor is ZIF-8 or HKUST-1, the preparation method includes the following steps: ZIF-8 or HKUST-1 was dispersed in an organic solvent, a titanium source was added, and a first-stage hydrothermal reaction was carried out to obtain the product. The product was subjected to a second-stage heat treatment in air to obtain a MOF-derived heterojunction.

6. The preparation method according to claim 5, characterized in that, The organic solvent is anhydrous ethanol; The titanium source is tetrabutyl titanate; The ratio of the MOF precursor, organic solvent and titanium source is 80 mg-120 mg: 30 mL-50 mL: 3 mL-5 mL; The first stage of hydrothermal reaction includes: reacting at 140℃-160℃ for 14h-18h; The second stage of heat treatment includes: heating from room temperature to 350℃-550℃ at a heating rate of 1℃ / min-5℃ / min, and holding at 350℃-550℃ for 2h-5h.

7. A MOF-derived heterojunction obtained by the preparation method according to any one of claims 1 to 6, characterized in that, The MOF-derived heterojunction comprises MOF-derived metal oxide and TiO2, and a tight heterojunction interface is formed between the MOF-derived metal oxide and TiO2.

8. The application of the MOF-derived heterojunction as described in claim 7 as a matrix in the detection of serum metabolic fingerprints by laser desorption / ionization mass spectrometry.

9. The application of the MOF-derived heterojunction of claim 7 in constructing a disease diagnostic model, characterized in that, The application method includes: The MOF-derived heterojunction was dispersed in deionized water to prepare a matrix suspension. Mix the serum sample with an organic solvent, vortex, and centrifuge to collect the supernatant. The supernatant is mixed with the matrix suspension and vortexed to obtain a mixture; The mixture was spotted onto a mass spectrometry target plate and dried at room temperature to form co-crystallized spots. LDI MS was used to collect data from the spots to obtain serum metabolic fingerprint profiles; The serum metabolic fingerprint was combined with machine learning algorithms to construct a disease diagnosis model.

10. The application according to claim 9, characterized in that, The disease diagnosis model includes a respiratory disease diagnosis model and / or a lung cancer staging diagnosis model. The respiratory diseases include at least one of bronchial asthma, chronic obstructive pulmonary disease, interstitial lung disease, and lung cancer. The lung cancer staging includes early lung cancer, intermediate lung cancer, and late lung cancer. The machine learning algorithm includes one or more combinations of neural networks, support vector machines, K-nearest neighbors, and Naive Bayes.

Images