Preparation method of array type silver nanoring SERS enhancement substrate and application thereof in serum metabolic fingerprint analysis method

Abstract

This invention provides a method for preparing an array-type silver nanoring SERS-enhanced substrate, comprising a self-assembly step of a polystyrene microsphere monolayer template; a step of preparing a silicon nanocone array using reactive ion etching; a step of removing residual template by high-temperature annealing; and a step of depositing a silver layer by magnetron sputtering. This invention also provides a method for serum metabolic fingerprint analysis using the above-mentioned array-type silver nanoring SERS-enhanced substrate. Furthermore, this invention provides the application of the above-mentioned method for serum metabolic fingerprint analysis using the array-type silver nanoring SERS-enhanced substrate in the construction of disease diagnostic models. The method provided by this invention has the advantages of simple operation, rapid detection, low sample requirement, high sensitivity, and high repeatability, and can be used for early screening, disease activity monitoring, and auxiliary diagnosis of autoimmune diseases.

Description

Technical Field

[0001] This invention belongs to the field of biological detection and relates to a surface-enhanced Raman spectroscopy technique, specifically a method for preparing an array of silver nanoring SERS-enhanced substrate and its application in the analysis of serum metabolic fingerprints. Background Technology

[0002] The era of precision medicine places higher demands on the early screening, accurate diagnosis, and dynamic monitoring of autoimmune diseases. Autoimmune diseases are diverse, have complex pathogenesis, and exhibit strong individual heterogeneity. Their diagnosis and assessment often rely on the comprehensive judgment of multiple laboratory indicators. However, existing conventional testing methods still have certain limitations in early identification, disease activity assessment, and personalized precision treatment, and are unable to comprehensively and in real time reflect changes in the body's metabolic state.

[0003] Systemic lupus erythematosus (SLE), a typical chronic systemic autoimmune disease, presents with complex and diverse clinical manifestations, affecting multiple organs including the skin, joints, kidneys, nervous system, and hematopoietic system. Currently, clinical diagnosis primarily relies on laboratory tests such as antinuclear antibodies, autoantibody profiles, complement levels, and inflammatory markers. However, the sensitivity and specificity of these indicators in early disease screening, activity assessment, and personalized precision diagnosis remain insufficient, failing to fully meet the clinical need for real-time mapping of the overall metabolic characteristics of the disease. Therefore, developing a rapid, non-invasive, highly sensitive novel detection method that reflects the overall metabolic characteristics of the disease is of great significance for the precise diagnosis and dynamic monitoring of autoimmune diseases.

[0004] In recent years, metabolic fingerprinting analysis methods based on blood and other bodily fluid samples have attracted widespread attention. Serum contains a large number of small molecule metabolites related to the occurrence and development of diseases, and changes in their composition and abundance can reflect the body's immune status and pathophysiological changes to a certain extent. Therefore, highly sensitive detection of serum metabolic characteristics holds promise for establishing novel analytical methods for the diagnosis of autoimmune diseases. Surface-enhanced Raman spectroscopy (SERS) is considered a highly promising analytical technique in the field of biomedical detection due to its advantages such as ultra-high sensitivity, no need for complex labeling, fast detection speed, and ability to provide molecular "fingerprint" information. SERS technology utilizes the localized surface plasmon resonance effect generated by noble metal nanostructures to significantly enhance the Raman scattering signal of target molecules, thereby achieving highly sensitive detection of trace metabolites.

[0005] Existing SERS detection systems primarily rely on liquid-phase dispersed noble metal nanoparticles as reinforcing substrates, such as silver nanoparticles (AgNPs), gold nanoparticles (AuNPs), and their composite nanostructures. While these systems possess high enhancement capabilities, they still face several limitations in practical applications. For instance, liquid-phase nanoparticles are prone to random aggregation, leading to uneven distribution of "hot spots" and consequently poor signal stability and repeatability. Furthermore, the detection process typically requires complex mixing, incubation, and separation steps, making it cumbersome and time-consuming. In addition, in complex biological samples, large protein molecules and impurities can easily interfere with the interaction between the target metabolites and nanoparticles, thereby reducing detection sensitivity and accuracy.

[0006] The aforementioned existing technologies suffer from numerous drawbacks, including poor signal stability, severe interference from complex serum backgrounds, complex operation, and insufficient data reliability. Summary of the Invention

[0007] To address the aforementioned technical problems in the existing technology, this invention provides a method for preparing an array-type silver nanoring SERS-enhanced substrate and its application in serum metabolic fingerprint analysis. This method for preparing an array-type silver nanoring SERS-enhanced substrate and its application in serum metabolic fingerprint analysis aims to solve the technical problems of poor sensitivity and repeatability in existing SERS technologies for detecting metabolites.

[0008] This invention provides a method for preparing an array of silver nanoring SERS-enhanced substrates, comprising the following steps:

[0009] 1) Self-assembly steps of a polystyrene microsphere monolayer template: First, polystyrene (PS) microspheres with a diameter of 100~140 nm are mixed with ethanol at a volume ratio of 1:(1~3) and ultrasonically dispersed to form a precursor suspension; then, deionized water is completely covered on the surface of a 2cm×2cm~4cm×4cm silicon (100) wafer; next, the precursor solution is dropped onto one end of the wafer using a pipette, so that the polystyrene microspheres can spontaneously migrate to the other end of the liquid surface; a tightly packed polystyrene microsphere monolayer template is formed on the silicon wafer surface;

[0010] 2) A step for fabricating a silicon nanocone array using reactive ion etching: The polystyrene microsphere monolayer template obtained in step 1) is dried at 50-65°C and then placed in a reactive ion etching (RIE) apparatus for etching; the etching gas is a plasma of SF6, Ar, and CHF3, with flow rates of 6, 15, and 11 sccm / min for SF6, Ar, and CHF3, respectively; the etching power is 150 W; the chamber pressure is 2.25 Pa; and the etching time is 51 seconds; a highly ordered silicon nanocone array (Si NCA) is formed on the silicon wafer.

[0011] 3) A step of removing residual template by high-temperature annealing: The etched silicon nanocone array obtained in step 2) is cleaned with ethanol and then annealed in a muffle furnace at 580~620℃; a pure and uncontaminated silicon nanocone array template is obtained.

[0012] 4) A step of magnetron sputtering to deposit a silver layer: On the prepared silicon nanocone array template, a silver layer is deposited by a magnetron sputtering system with a deposition current of 30mA and a deposition rate of about 21nm / min, and the thickness of the silver layer is precisely controlled to 5-20nm.

[0013] This invention also provides a method for serum metabolic fingerprint analysis using the above-mentioned arrayed silver nanoring SERS-enhanced substrate, comprising the following steps:

[0014] 1) A procedure for collecting serum samples: Collect 3–5 mL of peripheral venous blood from the subject using disposable sterile vacuum blood collection tubes, allow the blood to coagulate naturally at room temperature, and then centrifuge at 3000 rpm at 4°C to separate the supernatant serum.

[0015] 2) A step for pre-processing serum samples: The upper serum obtained in step 1) is added to an ultrafiltration centrifuge tube with a molecular weight cutoff of 30 kDa, and centrifuged at 10,000 g at 4°C to collect the serum ultrafiltrate.

[0016] 3) A step for detecting serum ultrafiltrate samples: 5 μL of serum ultrafiltrate was dropped onto the surface of the prepared arrayed silver nanoring SERS-enhanced substrate. The serum ultrafiltrate was scanned and detected by a micro Raman spectroscopy system using SERS mapping. 200 SERS spectra were collected by scanning multiple regions on the substrate surface point by point to obtain the serum SERS metabolic fingerprint spectrum, and the obtained spectral data were saved.

[0017] 4) Includes a preprocessing step for serum SERS metabolic fingerprint data: For each SERS spectrum in the serum SERS metabolic fingerprint, the SERS spectrum is first unified to 400–1800 cm⁻¹ using linear interpolation. -1The first-order processed spectrum is obtained by tracing the uniform wavenumber range within the range. Then, the moving average residual method is used to remove cosmic ray peak outliers from the first-order processed spectrum to obtain the second-order processed spectrum. The second-order processed spectrum is then baseline-corrected using the adaptive smoothing penalized least squares method to obtain the third-order processed spectrum. Finally, the corrected third-order processed spectrum is smoothed using the Savitzky-Golay filter to obtain the fourth-order processed spectrum. After performing the above preprocessing on all SERS spectra in the serum SERS metabolic fingerprint, all four-order processed spectra are summed and averaged, and the summed and averaged spectrum is used as the representative serum SERS metabolic fingerprint.

[0018] Furthermore, the subjects were patients with systemic lupus erythematosus.

[0019] Furthermore, during the detection process in step 3), the excitation wavelength of the Raman spectrometer was set to 532 nm, the laser power was set to 20 mW, the integration time was 1000 ms, and the mapping scan step size was set to 10 μm.

[0020] This invention also provides the application of the above-mentioned method for serum metabolic fingerprint analysis using an array-type silver nanoring SERS-enhanced substrate in the construction of disease diagnostic models.

[0021] Specifically, in the process of building a disease diagnosis model, representative serum SERS metabolic fingerprint profiles of multiple subjects were obtained, and the obtained profiles were used to construct a sample dataset.

[0022] The sample dataset is then divided into training and test sets according to the proportions. Multiple candidate algorithms are selected, and a disease diagnosis model is built using each candidate algorithm. The disease diagnosis models built by various candidate algorithms are compared and screened using the training and test sets, and the optimal model is selected as the final disease diagnosis model.

[0023] The candidate algorithms include random forest, decision tree, support vector machine, logistic regression model, and K-nearest neighbor algorithm.

[0024] Specifically, the disease diagnosis model constructed using the random forest algorithm was selected as the final disease diagnosis model.

[0025] Compared with existing technologies, the technical effects of this invention are positive and obvious.

[0026] 1. The array-type silver nanoring SERS enhancement substrate constructed in this invention has a regular and ordered periodic micro-nano structure. Uniform and controllable nano gaps are formed between the silver nanorings. Under laser excitation, it can generate a significant local surface plasmon resonance effect, thereby forming a high-density and highly uniform SERS enhancement "hot spot", which significantly improves the enhancement efficiency of Raman signal.

[0027] 2. This invention, through the design of an arrayed silver nanoring structure, effectively solves the problems of random distribution of "hot spots," large signal fluctuations, and insufficient repeatability in traditional randomly aggregated liquid-phase nanoparticle systems. Compared with traditional colloidal SERS systems, the arrayed substrate constructed in this invention exhibits superior signal uniformity and detection repeatability, significantly improving the reliability and comparability of detection results.

[0028] 3. The silver nanoring structure in this invention has a unique ring edge electric field coupling effect, which can form a stronger local electromagnetic field enrichment ability at the edge and interstitial regions of the nanoring, thereby improving the signal amplification ability of low-abundance metabolite molecules and enhancing the detection sensitivity of weak metabolic features in complex biological samples.

[0029] 4. This invention further incorporates a serum ultrafiltration pretreatment step, effectively removing large molecular weight proteins and high molecular weight background interference from serum through a 30 kDa ultrafiltration system, while enriching small molecule metabolite signals, resulting in clearer and more stable serum metabolism-related SERS characteristic peaks. This pretreatment method, together with the arrayed silver nanoring SERS substrate, forms a synergistic sensitizing effect, further improving the sensitivity and specificity of detection.

[0030] 5. This invention adopts a solid-phase SERS detection mode, which only requires adding the ultrafiltered serum sample to the surface of the arrayed silver nanoring substrate to complete the detection. This avoids the complex nanoparticle mixing, aggregation induction and centrifugation separation process in the traditional liquid-phase SERS system, which significantly simplifies the operation process, shortens the detection time and improves the standardization of the detection process.

[0031] 6. This invention combines microscopic Raman mapping scanning with automated machine learning analysis strategies to acquire serum SERS metabolic fingerprints through multi-point spatial acquisition. It also combines random forest, decision tree, support vector machine, logistic regression and k-nearest neighbor models for automatic screening and optimization, achieving efficient analysis of complex high-dimensional spectral data and improving the accuracy, stability and clinical application potential of autoimmune disease classification models. Attached Figure Description

[0032] Figure 1 This is a scanning electron microscopy image and elemental distribution map of the array-type silver nanoring substrate of Embodiment 1 of the present invention.

[0033] Figure 2 This is a representative serum SERS metabolic fingerprint spectrum with different retention amounts in Example 2 of the present invention, along with a comparison of peak intensity and signal-to-noise ratio.

[0034] Figure 3 This is a representative SERS spectrum of 4-MBN in Example 3 of the present invention and at 2229 cm⁻¹. -1 Peak intensity distribution at [location].

[0035] Figure 4 This is a 200-scan thermogram of the SERS metabolic fingerprint and a PCC distribution map of an SLE patient in Example 3 of the present invention.

[0036] Figure 5 The data consists of pretreated serum SERS metabolic fingerprint heatmaps and average spectra of 40 healthy controls and 40 SLE patients in Example 4 of this invention, as well as classification results from five machine learning models. Detailed Implementation

[0037] The following description, with reference to the accompanying drawings, illustrates several preferred embodiments of the present invention to make its technical content clearer and easier to understand. The present invention can be embodied in many different forms, and the scope of protection of the present invention is not limited to the embodiments mentioned herein.

[0038] Example 1: Preparation of an array of silver nanoring SERS-enhanced substrate

[0039] Step 1: A silicon nanocone array template was prepared by gas-liquid interface self-assembly combined with reactive ion etching and then further deposited.

[0040] Specifically, polystyrene microspheres with a diameter of 120 nanometers are mixed with ethanol at a volume ratio of 1:2 and ultrasonically dispersed to form a precursor suspension.

[0041] After covering a 2cm × 2cm silicon (100) wafer with deionized water, a precursor suspension was slowly dropped onto one end of the wafer using a pipette. This allowed polystyrene microspheres to spontaneously migrate and form a tightly packed polystyrene microsphere monolayer template using a gas-liquid interface self-assembly method. The silicon wafer covered with polystyrene microspheres was dried at 60°C for 20 minutes and then placed in a reactive ion etching (RIE) apparatus. Etching was performed using SF6, Ar, and CHF3 plasmas at flow rates of 6, 15, and 11 sccm / min, respectively, with an etching power of 150W, a chamber pressure of 2.25 Pa, and an etching time of 51 seconds. The reactive ion etching process etched downwards into the unprotected silicon regions, resulting in a highly ordered silicon nanocone array.

[0042] The etched silicon nanocone array was cleaned with ethanol and annealed in a muffle furnace at 600°C for 2 hours. The styrene microspheres remaining at the top of the silicon nanocones were completely removed by high-temperature calcination, resulting in a pure and uncontaminated silicon nanocone array template.

[0043] Finally, a silver layer was deposited on the prepared silicon nanocone array by magnetron sputtering at a deposition current of 30 mA and a deposition rate of approximately 21 nm / min.

[0044] Depend on Figure 1 It can be seen that the substrate surface exhibits a regular array-like structure, with the silver nanoring array being uniformly distributed and neatly arranged, showing no obvious structural defects. Through... Figure 1 The high-resolution morphology image and the corresponding Ag and Si elemental surface scan images show that the silver element represented by the green signal is mainly concentrated in the ring-shaped region, while the silicon element represented by the red signal is uniformly distributed in the bottom substrate region. The superimposed image further confirms the successful construction of silver nanorings on the silicon nanocone array.

[0045] Step 2: The arrayed silver nanoring SERS-enhanced substrate was sealed and stored under nitrogen gas at 4°C.

[0046] Example 2: Pretreatment method for serum samples

[0047] Step 1: All serum samples were collected from subjects in the morning on an empty stomach. 5 mL of peripheral venous blood was collected from each subject using a disposable sterile vacuum blood collection tube. The blood was allowed to clot naturally at room temperature for 30 min, then centrifuged at 3000 rpm for 10 min at 4°C to obtain the supernatant serum. All serum samples were immediately stored in a −80°C ultra-low temperature freezer.

[0048] Step 2: The serum sample stored at -80℃ was slowly thawed and gently mixed at 4℃. Preliminary experiments were performed using ultrafiltration centrifuge tubes with cutoff values ​​of 5 kDa, 10 kDa, 30 kDa, 50 kDa, and 100 kDa, respectively, and the SERS signals of the ultrafiltrate obtained at each cutoff value were compared. Representative SERS metabolic fingerprint profiles and detailed analyses under different cutoff conditions are shown below. Figure 2 As shown.

[0049] Depend on Figure 2 As shown in the left figure, the SERS spectra at each cutoff level all exhibit similar metabolic fingerprint characteristic peaks, but the signal intensities differ significantly. Among them, the spectrum at the 30 kDa cutoff level (red line) shows the highest signal intensity at each characteristic peak. Figure 2 The bar chart and line graph on the right further quantify the signal intensity and signal-to-noise ratio (SNR) at different cutoff values, showing that the signal intensity is highest and the SNR reaches its peak at a cutoff value of 30 kDa. Therefore, 30 kDa is determined to be the optimal cutoff value for serum sample ultrafiltration.

[0050] Step 3: Collect serum samples from patients diagnosed with systemic lupus erythematosus (SLE) by a professional clinician. The samples were obtained from Shanghai East Hospital and approved by the hospital's ethics committee. Add 200 μL of serum to an ultrafiltration centrifuge tube with a molecular weight cutoff of 30 kDa and centrifuge at 10,000 g for 10 min at 4 ℃. After ultrafiltration, large molecular weight proteins and high molecular weight impurities are retained above the filter membrane, while small molecular weight metabolites pass through the filter membrane into the filtrate. Collect the ultrafiltrate of SLE patient serum for subsequent SERS metabolic fingerprint analysis.

[0051] Example 3: Construction of a serum SERS metabolic fingerprint detection system for SLE patients

[0052] Step 1: Prepare a 1 mM 4-mercaptobenzonitrile (4-MBN) ethanol solution as a standard analyte. Add 5 μL of this solution to the surface of the arrayed silver nanoring SERS-enhanced substrate obtained in Example 1. Place the substrate in a micro Raman spectroscopy system, setting the excitation wavelength to 532 nm, laser power to 20 mW, integration time to 1000 ms, and mapping step size to 10 μm. Collect 200 SERS spectra point-by-point. From the first scan, sequentially accumulate the first n spectra (n = 1, 5, 10, 20, ..., 200), and calculate the 2229 cm⁻¹ value at each accumulation. -1 The relative standard deviation (RSD) of the characteristic peak intensity at 2229 cm⁻¹. Representative SERS spectrum of 4-MBN and 2229 cm⁻¹. -1 Peak intensity analysis at such location Figure 3 As shown.

[0053] Depend on Figure 3 As can be seen in the upper left figure, compared to bare 4-MBN (black line), the SERS spectrum (red line) of 4-MBN on the arrayed silver nanoring substrate exhibits a significantly enhanced characteristic peak signal. The heatmap in the upper right figure shows that the intensity of each characteristic peak signal remains stable and shows no significant attenuation in 200 consecutive measurements. The scatter plot in the lower left figure indicates the 2229 cm⁻¹... -1 The peak intensity fluctuated little in 200 measurements. The lower right figure shows that the RSD gradually stabilized with the increase of the number of cumulative measurements, reaching a plateau at 200 measurements (RSD of about 3%). This indicates that the substrate has excellent signal reproducibility and stability. Therefore, 200 measurements were determined to be the effective number of scans.

[0054] Step 2: Take 5 μL of the SLE patient serum ultrafiltrate obtained in Example 2 and drop it onto the surface of the arrayed silver nanoring SERS-enhanced substrate. SERS mapping scanning was performed using a micro Raman spectroscopy system. The excitation wavelength was set to 532 nm, laser power to 20 mW, integration time to 1000 ms, and mapping step size to 10 μm. Multiple regions on the substrate surface were scanned point-by-point, collecting a total of 200 SERS spectra. The SERS spectrum of the SLE patient serum ultrafiltrate was obtained and the data was saved. The SERS metabolic fingerprint of one SLE patient, including 200 scans of thermal imaging and PCC analysis, is shown below. Figure 4 As shown.

[0055] Depend on Figure 4 As shown in the left figure, the SERS signal of serum ultrafiltrate remained stable throughout 200 consecutive measurements, with no significant attenuation or drift. The right figure shows that PCC rose rapidly and then stabilized, indicating that the 200 scans were sufficient to characterize a single sample.

[0056] Example 4: Machine learning was applied to the extracted SERS metabolic fingerprint to construct a disease diagnosis model.

[0057] Step 1: Data preprocessing and analysis of serum SERS metabolic fingerprint profiles were completed in the Python 3.12 environment.

[0058] First, for each SERS spectrum in the serum SERS metabolic fingerprint, linear interpolation was used to unify the SERS spectra to 400–1800 cm⁻¹. -1 The first-order processed spectrum is obtained by tracing the uniform wavenumber range within the range. Then, the moving average residual method is used to remove cosmic ray peak outliers from the first-order processed spectrum to obtain the second-order processed spectrum. The second-order processed spectrum is then baseline-corrected using the adaptive smoothing penalized least squares method to obtain the third-order processed spectrum. Finally, the corrected third-order processed spectrum is smoothed using the Savitzky-Golay filter to obtain the fourth-order processed spectrum. After performing the above preprocessing on all SERS spectra in the serum SERS metabolic fingerprint, all four-order processed spectra are summed and averaged, and the summed and averaged spectrum is used as the representative serum SERS metabolic fingerprint.

[0059] Representative SERS metabolic fingerprint profiles were obtained. Metabolic fingerprint heatmap analysis was performed on serum samples from 40 healthy controls (HC) and 40 SLE patients (all from Shanghai East Hospital, meeting the medical disease definition and passing medical ethics review). The pre-processed serum SERS metabolic fingerprint heatmaps and average spectra of the 80 samples are shown below. Figure 5 As shown. By Figure 5As can be seen, the left figure shows the comparison of SERS spectral thermograms and difference spectral analysis between SLE patients (40 cases) and healthy controls (40 cases), and there are significant differences in characteristic peaks between the two types of samples in the fingerprint area.

[0060] Step 2: In the process of building the machine learning model, representative serum SERS metabolic fingerprints of multiple subjects were obtained, and a sample dataset was constructed using the obtained fingerprints. Then, the 80 sample datasets were divided into training and test sets in a 7:3 ratio. Several candidate algorithms were then selected (specifically Random Forest (RF), Decision Tree (DT), Support Vector Machine (SVM), Logistic Regression (LR), and k-Nearest Neighbor (kNN) algorithms). A disease diagnosis model was constructed using each candidate algorithm. The disease diagnosis models constructed by various candidate algorithms were compared and screened using the training and test sets, and the optimal model was selected as the final disease diagnosis model.

[0061] Simultaneously, an automatic hyperparameter optimization strategy was used to adjust the model parameters to improve the model's classification performance and generalization ability. The performance of each algorithm was evaluated to determine the optimal model. The results of SLE diagnosis using SERS metabolic fingerprinting are as follows: Figure 5 As shown.

[0062] Depend on Figure 5 As shown in the radar chart at the top center, Random Forest (RF) performs best in all five metrics: accuracy, sensitivity, and specificity. The ROC curve at the bottom left shows that Logistic Regression (LR) has the highest AUC (0.986), followed by Random Forest (RF) at 0.962, KNN and SVM at 0.979, and Decision Tree (DT) at 0.833. The bottom right shows the confusion matrix for the Random Forest test set; the accuracy rate for identifying healthy controls is 100.0%, while the accuracy rate for identifying SLE patients is 91.7%. These results demonstrate that the SERS metabolic fingerprint obtained by the method of this invention has high specificity (100.0%) and high sensitivity (91.7%) in diagnosing SLE.

[0063] Ultimately, the disease diagnosis model constructed using the random forest algorithm was selected as the final disease diagnosis model.

[0064] The above embodiments are merely preferred embodiments of the present invention and are not intended to limit the present invention in any form or substance. It should be noted that any technical solution that can be obtained by those skilled in the art based on the concept of the present invention through logical analysis, reasoning, or limited experimentation on the basis of the prior art should be within the scope of protection defined by the claims.

Claims

1. A method for preparing an array of silver nanoring SERS-enhanced substrates, characterized in that... Includes the following steps: 1) Self-assembly steps of a polystyrene microsphere monolayer template: First, polystyrene microspheres with a diameter of 100~140 nm are mixed with ethanol at a volume ratio of 1:(1~3) and ultrasonically dispersed to form a precursor suspension; then, deionized water is completely covered on the surface of a 2cm×2cm~4cm×4cm silicon (100) wafer; next, the precursor solution is dropped onto one end of the wafer using a pipette, so that the polystyrene microspheres can spontaneously migrate to the other end of the liquid surface; a tightly packed polystyrene microsphere monolayer template is formed on the silicon wafer surface; 2) A step for preparing silicon nanocone arrays using reactive ion etching: The polystyrene microsphere monolayer template obtained in step 1) is dried at 50~65℃ and then placed in a reactive ion etching device for etching; The etching gas was a plasma of SF6, Ar and CHF3, with flow rates of 6, 15 and 11 sccm / min for SF6, Ar and CHF3, respectively. The etching power was 150 W, the chamber pressure was 2.25 Pa and the etching time was 51 seconds. A highly ordered silicon nanocone array was formed on the silicon wafer. 3) A step of removing residual template by high-temperature annealing: The etched silicon nanocone array obtained in step 2) is cleaned with ethanol, and then annealed in a muffle furnace at 580~620℃ for 1~3 hours; a pure and uncontaminated silicon nanocone array template is obtained. 4) A step of magnetron sputtering to deposit a silver layer: On the prepared silicon nanocone array template, a silver layer is deposited by a magnetron sputtering system with a deposition current of 30mA and a deposition rate of about 21nm / min, and the thickness of the silver layer is precisely controlled to 5-20nm.

2. A method for serum metabolic fingerprint analysis using the arrayed silver nanoring SERS-enhanced substrate as described in claim 1, characterized in that... Includes the following steps: 1) A procedure for collecting serum samples: Collect 3-5 mL of peripheral venous blood from the subject using disposable sterile vacuum blood collection tubes, allow the blood to clot naturally at room temperature, and then centrifuge at 3000 rpm at 4°C to separate the supernatant serum. 2) A step for pre-processing serum samples: The upper serum obtained in step 1) is added to an ultrafiltration centrifuge tube with a molecular weight cutoff of 30 kDa, and centrifuged at 10,000 g at 4°C to collect the serum ultrafiltrate. 3) A procedure for detecting serum ultrafiltrate samples: 5 μL of serum ultrafiltrate was dropped onto the surface of the prepared arrayed silver nanoring SERS-enhanced substrate. The serum ultrafiltrate was scanned and detected by a micro Raman spectroscopy system using SERS mapping. 200 SERS spectra were collected by scanning the substrate surface point by point, and the serum SERS metabolic fingerprint spectrum was constructed using the 200 SERS spectra. The obtained spectral data were saved. 4) Includes a preprocessing step for serum SERS metabolic fingerprint data: For each SERS spectrum in the serum SERS metabolic fingerprint, the SERS spectrum is first unified to 400–1800 cm⁻¹ using linear interpolation. -1 The first processed spectrum is obtained by uniform wavenumber interval within the range; the second processed spectrum is obtained by removing cosmic ray peak outliers from the first processed spectrum using the moving average residual method; the third processed spectrum is obtained by baseline correction of the second processed spectrum using the adaptive smoothing penalized least squares method; and the fourth processed spectrum is obtained by smoothing the corrected third processed spectrum using the Savitzky-Golay filter. After performing the above preprocessing on all SERS spectra in the serum SERS metabolic fingerprint, the spectra from all four processing steps were summed and averaged, and the summed and averaged spectra were used as the representative serum SERS metabolic fingerprint.

3. The method according to claim 2, characterized in that, The subjects were patients with systemic lupus erythematosus.

4. The method according to claim 2, characterized in that, During the detection process in step 3), the excitation wavelength of the Raman spectrometer was set to 532 nm, the laser power was set to 20 mW, the integration time was 1000 ms, and the mapping scan step size was set to 10 μm.

5. The method for serum metabolic fingerprint analysis using an array-type silver nanoring SERS-enhanced substrate as described in claim 2 is applied in the construction of disease diagnostic models.

6. The application according to claim 5, characterized in that: In the process of building a disease diagnosis model, representative serum SERS metabolic fingerprints of multiple subjects were obtained, and the obtained fingerprints were used to construct a sample dataset. The sample dataset is then divided into training and test sets according to the proportions. Multiple candidate algorithms are selected, and a disease diagnosis model is built using each candidate algorithm. The disease diagnosis models built by various candidate algorithms are compared and screened using the training and test sets, and the optimal model is selected as the final disease diagnosis model. The candidate algorithms include random forest, decision tree, support vector machine, logistic regression model, and K-nearest neighbor algorithm.

7. The application according to claim 6, characterized in that: The disease diagnosis model constructed using the random forest algorithm was selected as the final disease diagnosis model.

Images