Sweat electrolyte detection method based on CRISPR technology

Abstract

The invention relates to the technical field of detection, and provides a sweat electrolyte detection method based on a CRI SPR technology, and the method comprises the following steps: firstly, collecting a sweat sample through a flexible patch, and carrying out impurity removal and pH value adjustment on sweat; then, respectively carrying out enrichment treatment on sodium ions and potassium ions in the sweat by utilizing the magnetic nanoparticles of which the surfaces are modified with specific recognition ligands; then, introducing the enriched sample into a designed CRI SPR detection system, and generating a fluorescence signal through a specific reaction of the CRI SPR system; then, monitoring the generated fluorescence signal in real time, removing noise interference in the signal by using a denoising algorithm, and correcting a signal time sequence error by using a time axis correction algorithm; and finally, converting the processed fluorescence signals into concentration values of sodium ions and potassium ions in sweat by utilizing the trained random forest model.

Description

Technical Field

[0001] The present invention relates to the field of detection technologies, and more specifically, to a method for detecting sweat electrolytes based on CRISPR technology. Background Art

[0002] Sweat is an easily accessible biological sample, which contains abundant electrolytes and metabolites. The concentrations of sodium ions and potassium ions in sweat are closely related to the body's water balance, electrolyte status, and metabolic changes during exercise. Real-time monitoring of the dynamic changes of these electrolytes can not only provide key data for optimizing exercise performance, but also enable efficient monitoring of the human physiological state in a non-invasive manner under resource-limited conditions.

[0003] In Document 1 (Pirovano P, et al., A wearable sensor for the detection of sodium and potassium in human sweat during exercise, Talanta, 2020), a technology platform SwEatch based on wearable sensors was reported. This platform uses ion-selective electrode technology and can monitor the changes in the concentrations of sodium ions and potassium ions in sweat in real time. Research shows that this platform exhibits good sensitivity and selectivity during exercise and can track the dynamic changes of sodium ions and potassium ions in sweat, providing technical support for sweat electrolyte detection.

[0004] CRISPR technology is a gene editing tool discovered in bacteria and archaea and has received extensive attention due to its high targeting specificity and powerful nucleic acid detection ability. The CRISPR-Cas system can achieve highly sensitive nucleic acid detection by targeting specific DNA or RNA sequences and cleaving their nucleic acids, and is one of the important technologies in the field of molecular diagnosis.

[0005] In Document 2 (Kumaran A, et al., Advancements in CRISPR-Based Biosensing for Next-Gen Point of Care Diagnostic Application, Biosensors, 2023), the applications of CRISPR technology in point-of-care testing platforms were systematically reviewed, including fluorescence, colorimetric, and electrochemical detection methods. The document mentioned that CRISPR detection platforms represented by SHERLOCK and DETECTR have been able to achieve highly sensitive detection of viruses and bacteria, with a sensitivity reaching the attomolar (aM) level, providing technical guarantee for point-of-care diagnosis.

[0006] Although the existing technologies have made some progress in the detection of sodium and potassium ions in sweat, there are still some limitations. For example, traditional sweat detection platforms have limited capabilities in multi-target detection and high-sensitivity detection, and it is difficult to cope with signal interference caused by fluctuations in sweat secretion rate. With the rapid development of CRISPR technology, how to integrate it into the sweat electrolyte detection system to improve detection sensitivity, targeting, and dynamic response capabilities remains an urgent problem to be solved. Summary of the Invention

[0007] To overcome the above-mentioned defects of the existing technologies, the present invention provides a method for detecting sweat electrolytes based on CRISPR technology. Sweat samples are collected through a flexible patch, and magnetic nanoparticles are used to enrich sodium and potassium ions in sweat. Combined with the CRISPR detection system to generate fluorescence signals, the Kalman filter algorithm is used for real-time noise reduction and the dynamic time warping algorithm is used to correct the fluctuations in sweat secretion rate. Finally, the fluorescence signals are converted into electrolyte concentration values through a random forest model, solving the problems of low sensitivity in multi-target detection and large influence by signal interference in the existing sweat detection system, and realizing high-sensitivity and high-precision real-time sweat electrolyte monitoring.

[0008] To achieve the above object, the present invention provides the following technical solutions:

[0009] A method for detecting sweat electrolytes based on CRISPR technology, comprising the following steps:

[0010] Step S1, collecting a sweat sample through a flexible patch;

[0011] Step S2, removing impurities and adjusting the pH value of the collected sweat sample, removing impurity particles through a filter membrane, and using a buffer solution containing EDTA and an ion regulator to adjust the sweat sample to the ion strength and pH range required by the CRISPR reaction system;

[0012] Step S3, using magnetic nanoparticles surface-modified with sodium and potassium ion-specific recognition ligands to enrich sodium and potassium ions in sweat respectively;

[0013] Step S4, introducing the enriched samples into the CRISPR detection systems for sodium and potassium ions respectively;

[0014] Step S5, monitoring the fluorescence signals generated by the CRISPR detection system in real time, and using the Kalman filter algorithm to remove the real-time noise of the fluorescence detection signals of sodium and potassium ions respectively; at the same time, using an improved dynamic time warping algorithm to eliminate the signal timing deviation caused by the fluctuations in sweat secretion rate, wherein the expression of the improved dynamic programming cumulative distance formula is:

[0015]

[0016] In the formula, P(i,j) is the path matching value of the i-th row and j-th column, σ is the extended range parameter of the Gaussian weight, k is the time step, λ1 is the weight parameter of the sweat secretion rate constraint, λ2 is the weight parameter of the signal second derivative constraint, s k+i is the sweat secretion rate at the k+i-th moment, s k+j is the sweat secretion rate at the k+j-th moment, is the second derivative of the signal to be aligned at the k+i-th moment, is the second derivative of the standard signal at the k+j-th moment;

[0017] Step S6, through the pre-trained random forest model, the processed dual-channel fluorescence signals are respectively converted into the real-time concentration values of sodium ions and potassium ions in sweat.

[0018] As a further solution of the present invention, the CRISPR detection system includes an RNA aptamer that specifically recognizes sodium ions or potassium ions, a Cas13 protein, and a fluorescently labeled RNA probe. When the RNA aptamer specifically binds to the corresponding ion, the nuclease activity of the Cas13 protein is activated, thereby cleaving the RNA probe to release a fluorescent signal.

[0019] As a further solution of the present invention, in step S5, the specific execution of the Kalman filtering algorithm includes the following steps:

[0020] Step S51, respectively construct four-dimensional state vectors for the fluorescence detection signals of sodium ions and potassium ions. The state vector includes the true value of the signal, the signal change rate, the sweat secretion rate and its change rate;

[0021] Step S52, establish an observation equation based on the measured fluorescence signal and sweat secretion sensor data. The observation equation reflects the mapping relationship between the state quantity and the measured value. Among them, the equation of the observation equation is:

[0022]

[0023] In the formula, z k is the observed value, β is the influence coefficient of sweat secretion on the signal, x k is the signal value of the ion concentration at the current moment, is the signal value change rate of the ion concentration at the current moment, s k is the sweat secretion rate at the current moment, is the change rate of the sweat secretion rate at the current moment, v k is the measurement noise at the current moment;

[0024] Step S53: Model the system noise as a Gaussian distribution considering the influence of sweat secretion fluctuations, and model the measurement noise as a Gaussian distribution considering environmental interference. Determine the mean and variance parameters of the two Gaussian distributions through statistical analysis of historical detection data;

[0025] Step S54: Execute the prediction step. Calculate the prior state estimate according to the state equation, construct a state transition matrix based on the coupling relationship between the true signal value and the sweat secretion rate, substitute the state vector at the previous moment into the state equation to obtain the prior estimate at the current moment, and calculate the prior error covariance. Among them, the equation expression of the state equation is:

[0026]

[0027] In the formula, x k+1 is the signal value of the ion concentration at the next moment, is the change rate of the signal value of the ion concentration at the next moment, s k+1 is the sweat secretion rate at the next moment, is the change rate of the sweat secretion rate at the next moment, Δt is the sampling time interval, α is the coupling coefficient, x k is the signal value of the ion concentration at the current moment, is the change rate of the signal value of the ion concentration at the current moment, s k is the sweat secretion rate at the current moment, is the change rate of the sweat secretion rate at the current moment, w k is the system noise;

[0028] Step S55: Execute the update step. Calculate the Kalman gain according to the observation equation, substitute the deviation between the observed value and the prior estimate into the Kalman gain equation for state correction, update the posterior error covariance, and return to Step S54 to continue executing the prediction-update loop until the signal detection ends after one iteration.

[0029] As a further solution of the present invention, the specific execution of the dynamic time warping algorithm in Step S5 includes the following steps:

[0030] Step A1: Align the fluorescence signal sequences of sodium ions and potassium ions after Kalman filtering processing with the signal sequences under the corresponding standard sweat secretion patterns respectively;

[0031] Step A2: Construct an accumulated distance matrix. The calculation of the distance matrix uses a mixed distance metric considering signal amplitude and derivative, and introduces the sweat secretion rate as a constraint condition. Among them, the formula of the mixed distance metric is:

[0032]

[0033] Wherein, D(i,j) is the element at the i-th row and j-th column in the cumulative distance matrix, S1(i) is the amplitude of the signal to be aligned at time i, S2(j) is the amplitude of the standard signal at time j, is the derivative value of the signal to be aligned at time i, is the derivative value of the standard signal at time j, s i is the sweat secretion rate at time i, s j is the sweat secretion rate at time j, ω1, ω2 and ω3 are weight coefficients;

[0034] Step A3, searching for the optimal matching path based on the improved dynamic programming algorithm, and the improvement includes: adding path constraints based on sweat secretion kinetics, setting an adaptive step size range, and introducing hard constraints of signal feature points;

[0035] Step A4, performing a non-linear time axis transformation on the signal according to the optimal matching path to eliminate the timing deviation caused by the fluctuation of the sweat secretion rate;

[0036] Step A5, performing spline interpolation on the corrected signal to obtain a standardized signal on a unified time axis.

[0037] As a further solution of the present invention, by collecting sweat samples of healthy subjects under standard experimental conditions for detection, fluorescence signal data is obtained; the obtained signal data is normalized and time-aligned; the signal mean value of all samples at each time point is calculated to obtain a standard signal sequence.

[0038] As a further solution of the present invention, the fluorescence signal after being processed in step S5, after being normalized and denoised, is used as the input data of the random forest model in step S6, wherein the specific construction steps of the random forest model in step S6 include the following steps:

[0039] Step S61, extracting the time-domain features of the sodium ion and potassium ion fluorescence signals, including signal peak value, mean value, standard deviation, and sweat secretion cycle characteristics;

[0040] Step S62, extracting the frequency-domain features of the sodium ion and potassium ion fluorescence signals, including power spectral density, main frequency component, and sweat secretion frequency characteristics;

[0041] Step S63, inputting the extracted time-domain and frequency-domain feature vectors into the random forest model;

[0042] Step S64, setting the splitting rule of the decision tree based on the mean square error minimization criterion for ion concentration prediction;

[0043] Step S65, calculating the final concentration values of sodium ions and potassium ions in sweat by weighted-averaging the prediction results of multiple decision trees.

[0044] Compared with the prior art, the beneficial effects of the present invention, a kind of... are as follows:

[0045] The present invention introduces the Kalman filtering algorithm in signal processing, and respectively establishes four-dimensional state vectors for the fluorescence signals of sodium ions and potassium ions, including the true value of the signal, the signal change rate, the sweat secretion rate and its change rate. By constructing the observation equation and the state transition equation, combined with the real-time monitoring data, the signal change trend is accurately predicted and the noise is effectively removed. In contrast, the prior art lacks systematic modeling and filtering processing for real-time dynamic signals, is easily interfered by environmental noise and signal fluctuations, resulting in unstable detection results and low accuracy.

[0046] The present invention uses the dynamic time warping algorithm to correct the fluorescence signal sequence, combines the dynamic constraint conditions of the sweat secretion rate, eliminates the signal timing deviation caused by the fluctuation of the sweat secretion rate, and realizes the signal alignment on the unified time basis through the non-linear time axis transformation. In contrast, the prior art lacks systematicness in diverse signal correction, cannot adapt to the complex signal changes caused by the fluctuation of the sweat secretion rate, resulting in significant errors in the detection results. The present invention effectively improves the accuracy and robustness of the detection and realizes a higher level of sweat electrolyte monitoring. Brief Description of the Drawings

[0047] Figure 1 It is an exploded view of the combination of PEDOT and POT ion-selective electrodes in Document 1.

[0048] Figure 2 It is an exploded view of the SwEatch platform structure in Document 1.

[0049] Figure 3 It is a diagram of the targeting binding and sequence-specific cleavage mechanism of the CRISPR-Cas system in Document 2.

[0050] Figure 4 It is a schematic flow chart of a method for detecting sweat electrolytes based on CRISPR technology according to the present invention.

[0051] Figure 5 It is a schematic flow chart of constructing a random forest model for a method for detecting sweat electrolytes based on CRISPR technology according to the present invention.

[0052] In the figure, PMMA Gasket: PMMA gasket; PEDOT Transducing Layer: PEDOT conductive layer; Dielectric Insulating Ink: dielectric insulating ink; Screen Printed Conductive Carbon InkLayer on PET Substrate: screen-printed conductive carbon ink layer on PET substrate; Target Specific BindingProtein: target-specific binding protein; ss RNA, ss DNA, ds DNA: single-stranded RNA, single-stranded DNA, double-stranded DNA; Effector Cas Protein: effector Cas protein; Bound Target Gene: bound target gene; TargetSpecific Cleavage: target-specific cleavage; Sequence Specific Binding: sequence-specific binding; Sequence Specific Trans-cleavage: sequence-specific trans-cleavage; Target Nucleic Acid: target nucleic acid; Reporter Probe: reporter probe; Binding with Cas Complex: binding with Cas complex; Cleavage: cleavage; Reporter Free Complex: released reporter probe complex. Detailed implementation mode

[0053] The following will clearly and completely describe the technical solutions in this embodiment in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0054] Example 1

[0055] As Figure 4 shown, a method for detecting sweat electrolytes based on CRISPR technology includes the following steps:

[0056] Step S1, collecting a sweat sample through a flexible patch;

[0057] Step S2, removing impurities and adjusting the pH value of the collected sweat sample, removing impurity particles through a filter membrane, and using a buffer solution containing EDTA and an ion regulator to adjust the sweat sample to the ion strength and pH range required for the CRISPR reaction system;

[0058] Step S3: Use magnetic nanoparticles with sodium ion- and potassium ion-specific recognition ligands on their surfaces to enrich sodium ions and potassium ions in sweat respectively;

[0059] Step S4: Introduce the enriched samples into the CRISPR detection systems for sodium ions and potassium ions respectively;

[0060] Step S5: Monitor the fluorescence signals generated by the CRISPR detection systems in real time, and use the Kalman filtering algorithm to remove the real-time noise from the fluorescence detection signals of sodium ions and potassium ions respectively; at the same time, use the improved dynamic time warping algorithm to eliminate the signal timing deviation caused by the fluctuation of sweat secretion rate;

[0061] Step S6: Through a pre-trained random forest model, convert the processed dual-channel fluorescence signals into the real-time concentration values of sodium ions and potassium ions in sweat respectively.

[0062] In step S1 of the embodiment of the present invention, a flexible hydrogel patch is used to collect human surface sweat. The flexible hydrogel patch is made of polyacrylate hydrogel material. The surface of the patch is provided with a microfluidic channel array covering the sweat gland-dense area. The microfluidic channel array is made of polydimethylsiloxane material. The inner wall of the channel is formed into a hydrophilic surface through plasma treatment to promote the directional flow of sweat; a hydrophobic isolation layer is provided at the bottom of the patch. The isolation layer is provided with a through-hole array corresponding to the positions of sweat glands. The through-holes are in one-to-one correspondence and communication with the microfluidic channels; an independent chamber for detecting sodium and potassium ions in sweat is provided in the microfluidic channel. The inner wall of the chamber is modified with an inert group to prevent ion adsorption.

[0063] In step S2 of the embodiment of the present invention, the pretreatment is carried out in a constant-temperature microfluidic reaction chamber, and the flow rate is controlled by a peristaltic pump. Among them, the specific steps of the pretreatment are:

[0064] Step S21: The sweat sample passes through a protein filtration membrane with a pore size of 5 microns and a fine filtration membrane with a pore size of 0.22 microns in sequence to remove interfering substances such as cell debris, proteins, and amino acids in sweat;

[0065] Step S22: Mix the filtered sweat with an EDTA buffer solution in a volume ratio of 1:1. The composition of the buffer solution includes EDTA, sodium chloride, potassium chloride, and HEPES, and is used to adjust the pH value and ionic strength of sweat.

[0066] In step S3 of the embodiments of the present invention, the specific structure of the magnetic nanoparticle system in the third step is as follows: the magnetic nanoparticles are composed of an Fe3O4 core and a silica shell layer, and the particle surface is respectively modified with a sodium ion-specific recognition ligand and a potassium ion-specific recognition ligand; the ligand is connected to the silane coupling agent on the particle surface through a thiolated oligonucleotide sequence; the magnetic nanoparticles are controlled by an electromagnetic coil arranged outside the detection chamber, and the dynamic magnetic field generated by the electromagnetic coil is perpendicular to the sweat flow direction for capturing and enriching target ions.

[0067] In step S4 of the embodiments of the present invention, the specific composition of the CRISPR detection system is as follows: the sodium ion-specific RNA aptamer and the potassium ion-specific RNA aptamer are both screened from a random RNA library through the SELEX technique. The core sequence region of the aptamer contains a conserved domain for recognizing the target ion, and the domain forms a specific stem-loop structure; the Cas13 protein is a highly active mutant obtained by directed evolution; the 5' end and 3' end of the RNA probe are respectively labeled with fluorescent groups of different wavelengths for distinguishing the detection signals of sodium ions and potassium ions.

[0068] The CRISPR detection system includes an RNA aptamer that specifically recognizes sodium ions or potassium ions, a Cas13 protein, and a fluorescently labeled RNA probe. When the RNA aptamer specifically binds to the corresponding ion, the nuclease activity of the Cas13 protein is activated, thereby cleaving the RNA probe to release a fluorescent signal.

[0069] The structure of the RNA aptamer in the embodiments of the present invention includes a first functional region, which is composed of a conserved stem-loop structure and specifically binds to sodium ions or potassium ions.

[0070] The structure of the RNA aptamer in the embodiments of the present invention further includes a second functional region, which is composed of a variable base sequence and binds to the active site of the Cas13 protein.

[0071] The structure of the RNA aptamer in the embodiments of the present invention further includes a third functional region, whose nucleotide sequence is complementary to the RNA probe; when the target ion binds to the first functional region, it causes a conformational change in the RNA aptamer, exposing the second functional region and activating the Cas13 protein.

[0072] The Cas13 protein in the embodiments of the present invention is a double mutant obtained by directed evolution. Among them, the first mutation site is located in the catalytic domain to improve its cleavage activity against the RNA probe; the second mutation site is located in the binding domain to enhance its binding ability to the RNA aptamer; the double mutant undergoes a conformational change after recognizing the RNA aptamer-ion complex, exposing the nuclease active site.

[0073] The design of the fluorescent RNA probe in the embodiments of the present invention includes labeling the FAM fluorescent group on the RNA probe for sodium ion detection, and its sequence is complementary to the third functional region of the sodium ion-specific RNA aptamer.

[0074] The design of the fluorescent RNA probe in the embodiments of the present invention further includes labeling the Cy5 fluorescent group on the RNA probe for potassium ion detection, and its sequence is complementary to the third functional region of the potassium ion-specific RNA aptamer; the sequences of the two RNA probes are optimized to avoid mutual interference.

[0075] In step S5 of the embodiments of the present invention, the specific execution of the Kalman filter algorithm includes the following steps:

[0076] Step S51, constructing four-dimensional state vectors for the fluorescence detection signals of sodium ions and potassium ions respectively, and the state vectors include the true signal value, the signal change rate, the sweat secretion rate and its change rate;

[0077] Step S52, establishing an observation equation based on the measured fluorescence signal and the sweat secretion sensor data, and the observation equation reflects the mapping relationship between the state quantity and the measured value. Among them, the equation of the observation equation is:

[0078]

[0079] In the formula, z k is the observed value, β is the influence coefficient of sweat secretion on the signal, x k is the signal value of the ion concentration at the current moment, is the signal value change rate of the ion concentration at the current moment, s k is the sweat secretion rate at the current moment, is the change rate of the sweat secretion rate at the current moment, v k is the measurement noise at the current moment;

[0080] Step S53, modeling the system noise as a Gaussian distribution considering the influence of sweat secretion fluctuations, modeling the measurement noise as a Gaussian distribution considering environmental interference, and determining the mean and variance parameters of the two Gaussian distributions through statistical analysis of historical detection data;

[0081] Step S54, performing the prediction step, calculating the prior state estimate according to the state equation, constructing the state transition matrix based on the coupling relationship between the true signal value and the sweat secretion rate, substituting the state vector of the previous moment into the state equation to obtain the prior estimate of the current moment, and calculating the prior error covariance. Among them, the equation expression of the state equation is:

[0082]

[0083] In the formula, xk+1 is the signal value of the ion concentration at the next moment, is the rate of change of the signal value of the ion concentration at the next moment, s k+1 is the secretion rate of sweat at the next moment, is the rate of change of the sweat secretion rate at the next moment, Δt is the sampling time interval, α is the coupling coefficient, x k is the signal value of the ion concentration at the current moment, is the rate of change of the signal value of the ion concentration at the current moment, s k is the secretion rate of sweat at the current moment, is the rate of change of the sweat secretion rate at the current moment, w k is the system noise;

[0084] Step S55, execute the update step, calculate the Kalman gain according to the observation equation, substitute the deviation between the observed value and the prior estimate into the Kalman gain equation for state correction, update the posterior error covariance, and return to step S54 to continue executing the prediction-update loop until the signal detection ends after one iteration.

[0085] The construction method of the state transition matrix in the embodiments of the present invention is as follows: establish a functional relationship between the true signal value and the sweat secretion rate based on the physiological model of sweat secretion; linearize the functional relationship to obtain the coupling coefficient between state variables; fill the coupling coefficient into the corresponding position of the state transition matrix; the state transition matrix reflects the dynamic relationship between state variables at adjacent moments.

[0086] The calculation method of the Kalman gain in the embodiments of the present invention is as follows: calculate the optimal gain value based on the observation noise covariance matrix and the prior error covariance matrix; the optimal gain value minimizes the mean square error of the state estimate; the observation noise covariance matrix is determined by analyzing environmental interference factors, and the prior error covariance matrix is determined by evaluating the uncertainty of the system model.

[0087] The specific execution of the dynamic time warping algorithm in step S5 in the embodiments of the present invention includes the following steps:

[0088] Step A1, align the fluorescence signal sequences of sodium ions and potassium ions after Kalman filtering processing with the signal sequences in the corresponding standard sweat secretion patterns respectively, where the standard signal sequences are obtained based on a large amount of experimental data statistics;

[0089] Step A2, construct an accumulated distance matrix, and the calculation of the distance matrix uses a mixed distance metric considering signal amplitude and derivative, and introduces the sweat secretion rate as a constraint condition, where the formula of the mixed distance metric is:

[0090]

[0091] In the formula, D(i, j) is the element in the i-th row and k-th column of the cumulative distance matrix, S1(i) is the amplitude of the signal to be aligned at time i, and S2(j) is the amplitude of the standard signal at time j. is the derivative value of the signal to be aligned at time i. is the derivative value of the standard signal at time j, s i is the sweat secretion rate at time i, s j is the sweat secretion rate at time j, and ω1, ω2, and ω3 are weight coefficients;

[0092] Step A3: Search for the optimal matching path based on the improved dynamic programming algorithm. The improvement includes: adding a path constraint based on sweat secretion kinetics, setting an adaptive step size range, and introducing a hard constraint of signal feature points. Among them, the expression of the improved dynamic programming cumulative distance formula is:

[0093]

[0094] In the formula, P(i, j) is the path matching value of the i-th row and j-th column, σ is the extended range parameter of the Gaussian weight, k is the time step, λ1 is the weight parameter of the sweat secretion rate constraint, λ2 is the weight parameter of the signal second derivative constraint, s k+i is the sweat secretion rate at the k + i-th moment, s k+j is the sweat secretion rate at the k + j-th moment, is the second derivative of the signal to be aligned at the k + i-th moment, is the second derivative of the standard signal at the k + j-th moment;

[0095] Step A4: Perform a non-linear time axis transformation on the signal according to the optimal matching path to eliminate the timing deviation caused by the fluctuation of the sweat secretion rate;

[0096] Step A5: Perform spline interpolation on the corrected signal to obtain a standardized signal on the unified time axis.

[0097] In the embodiment of the present invention, sweat samples of healthy subjects are collected and detected under standard experimental conditions to obtain fluorescence signal data; the obtained signal data is normalized and time-aligned; the signal mean value of all samples at each time point is calculated to obtain a standard signal sequence.

[0098] The fluorescence signal processed in step S5 in the embodiment of the present invention, after being normalized and denoised, is used as the input data of the random forest model in step S6. Among them, the specific construction steps of the random forest model in step S6 include the following steps:

[0099] Step S61, extract the time-domain features of the sodium ion and potassium ion fluorescence signals, including signal peak value, mean value, standard deviation, and sweat secretion cycle characteristics;

[0100] Step S62, extract the frequency-domain features of the sodium ion and potassium ion fluorescence signals, including power spectral density, main frequency component, and sweat secretion frequency characteristics;

[0101] Step S63, input the extracted time-domain and frequency-domain feature vectors into the random forest model, which consists of multiple decision trees, and each decision tree randomly samples the training data and feature subsets based on the bootstrap method;

[0102] Step S64, set the splitting rules of the decision tree based on the mean square error minimization criterion for ion concentration prediction;

[0103] Step S65, calculate the final concentration values of sodium ions and potassium ions in sweat by weighted averaging the prediction results of multiple decision trees.

[0104] Embodiment 2

[0105] As Figure 5 shown, the specific construction steps of the random forest model in Step S6 include the following steps:

[0106] Step S61, extract the time-domain features of the sodium ion and potassium ion fluorescence signals, including signal peak value, mean value, standard deviation, and sweat secretion cycle characteristics;

[0107] Step S62, extract the frequency-domain features of the sodium ion and potassium ion fluorescence signals, including power spectral density, main frequency component, and sweat secretion frequency characteristics;

[0108] Step S63, input the extracted time-domain and frequency-domain feature vectors into the random forest model;

[0109] Step S64, set the splitting rules of the decision tree based on the mean square error minimization criterion for ion concentration prediction;

[0110] Step S65, calculate the final concentration values of sodium ions and potassium ions in sweat by weighted averaging the prediction results of multiple decision trees.

[0111] The following is a Python code example for obtaining the final concentration values of sodium ions and potassium ions in sweat using the random forest model. Please note that this example is only a starting point and may need to be adjusted according to the actual situation and device interface in practical applications;

[0112]

[0113]

[0114]

[0115]

[0116]

[0117]

[0118]

[0119]

[0120] This code is only for example, and in actual application, appropriate modifications and adjustments need to be made according to specific situations.

[0121] As described above, it is only the specific implementation manner of this application, but the protection scope of this application is not limited thereto. Any person skilled in the art in the technical scope disclosed by this application can easily think of changes or substitutions, which should all be covered by the protection scope of this application. Therefore, the protection scope of this application should be subject to the protection scope of the claims described.

[0122] Finally: The above description is only the preferred embodiment of the present invention and is not used to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principle of the present invention should all be included in the protection scope of the present invention.

Claims

1. A method for detecting sweat electrolytes based on CRISPR technology, characterized in that, It includes the following steps: Step S1: Collect a sweat sample through a flexible patch; Step S2: Remove impurities and adjust the pH value of the collected sweat sample. Remove impurity particles through a filter membrane, and use a buffer solution containing EDTA and an ion regulator to adjust the sweat sample to the ion strength and pH range required by the CRISPR reaction system; Step S3: Use magnetic nanoparticles surface-modified with sodium-ion and potassium-ion specific recognition ligands to enrich sodium ions and potassium ions in sweat respectively; Step S4: Introduce the enriched samples into the CRISPR detection systems for sodium ions and potassium ions respectively; Step S5: Monitor the fluorescence signals generated by the CRISPR detection systems in real time, and use the Kalman filtering algorithm to remove the real-time noise of the fluorescence detection signals of sodium ions and potassium ions respectively; meanwhile, use the improved dynamic time warping algorithm to eliminate the signal timing deviation caused by the fluctuation of the sweat secretion rate. Among them, the expression formula for improving the dynamic programming cumulative distance formula is: Wherein, P(i,j) is the path matching value of the i-th row and the j-th column, σ is the extended range parameter of the Gaussian weight, k is the time step, λ1 is the weight parameter of the sweat secretion rate constraint, λ2 is the weight parameter of the signal second derivative constraint, s k+i is the sweat secretion rate at the k+i-th moment, s k+j is the sweat secretion rate at the k+j-th moment, is the second derivative of the signal to be aligned at the k+i-th moment, is the second derivative of the standard signal at the k+j-th moment; Step S6: Through a pre-trained random forest model, convert the processed dual-channel fluorescence signals into the real-time concentration values of sodium ions and potassium ions in sweat respectively.

2. The method for detecting sweat electrolytes based on CRISPR technology according to claim 1, wherein The CRISPR detection system includes an RNA aptamer that specifically recognizes sodium ions or potassium ions, Cas13 protein, and a fluorescently labeled RNA probe. When the RNA aptamer specifically binds to the corresponding ion, it activates the nuclease activity of Cas13 protein, thereby cleaving the RNA probe to release a fluorescence signal.

3. The method for detecting sweat electrolytes based on CRISPR technology according to claim 1, wherein In Step S5, the specific execution of the Kalman filtering algorithm includes the following steps: Step S51: Construct a four-dimensional state vector for the fluorescence detection signals of sodium ions and potassium ions respectively. The state vector includes the true value of the signal, the signal change rate, the sweat secretion rate, and its change rate; Step S52: Establish an observation equation based on the measured fluorescence signal and sweat secretion sensor data. The observation equation reflects the mapping relationship between the state quantity and the measurement value. Among them, the equation of the observation equation is: where z k is the observed value, β is the influence coefficient of sweat secretion on the signal, x k is the signal value of the ion concentration at the current moment, is the change rate of the signal value of the ion concentration at the current moment, s k is the secretion rate of sweat at the current moment, is the change rate of the sweat secretion rate at the current moment, v k is the measurement noise at the current moment; Step S53: Model the system noise as a Gaussian distribution considering the influence of sweat secretion fluctuations, and model the measurement noise as a Gaussian distribution considering environmental interference. Determine the mean and variance parameters of the two Gaussian distributions through statistical analysis of historical detection data; Step S54: Execute the prediction step. Calculate the prior state estimate according to the state equation, construct a state transition matrix based on the coupling relationship between the true value of the signal and the sweat secretion rate, substitute the state vector at the previous moment into the state equation to obtain the prior estimate at the current moment, and calculate the prior error covariance. Among them, the equation expression of the state equation is: where x k+1 is the signal value of the ion concentration at the next moment, is the change rate of the signal value of the ion concentration at the next moment, s k+1 is the secretion rate of sweat at the next moment, is the change rate of the sweat secretion rate at the next moment, Δt is the sampling time interval, α is the coupling coefficient, x k is the signal value of the ion concentration at the current moment, is the change rate of the signal value of the ion concentration at the current moment, s k is the secretion rate of sweat at the current moment, is the change rate of the sweat secretion rate at the current moment, w k is the system noise; Step S55: Execute the update step. Calculate the Kalman gain according to the observation equation, substitute the deviation between the observed value and the prior estimate into the Kalman gain equation for state correction, update the posterior error covariance, and return to Step S54 to continue executing the prediction-update loop until the signal detection ends after one iteration.

4. The method for detecting sweat electrolytes based on CRISPR technology according to claim 1, wherein The specific execution of the dynamic time warping algorithm in Step S5 includes the following steps: Step A1: Align the fluorescence signal sequences of sodium ions and potassium ions after Kalman filtering with the signal sequences in the corresponding standard sweat secretion patterns respectively. Step A2: Construct an accumulated distance matrix. The calculation of the distance matrix uses a hybrid distance metric considering signal amplitude and derivative, and introduces the sweat secretion rate as a constraint condition. Among them, the formula of the hybrid distance metric is: Where, D(i,j) is the element in the i-th row and j-th column of the cumulative distance matrix, S1(i) is the amplitude of the signal to be aligned at time i, S2(j) is the amplitude of the standard signal at time j, is the derivative value of the signal to be aligned at time i, is the derivative value of the standard signal at time j, s i is the sweat secretion rate at time i, s j is the sweat secretion rate at time j, and ω1, ω2, and ω3 are weight coefficients; Step A3: Search for the optimal matching path based on an improved dynamic programming algorithm. The improvement includes: adding a path constraint based on sweat secretion kinetics, setting an adaptive step size range, and introducing a hard constraint on signal feature points. Step A4: Perform a non-linear time axis transformation on the signal according to the optimal matching path to eliminate the timing deviation caused by the fluctuation of the sweat secretion rate. Step A5: Perform spline interpolation on the corrected signal to obtain a standardized signal on a unified time axis.

5. The method for detecting sweat electrolytes based on CRISPR technology according to claim 1 or 4, characterized in that, Detect the fluorescence signal data by collecting sweat samples from healthy subjects under standard experimental conditions; perform normalization processing and time alignment on the obtained signal data. Calculate the signal mean at each time point for all samples to obtain a standard signal sequence.

6. The sweat electrolyte detection method based on CRISPR technology according to claim 1, wherein, The fluorescence signal after being processed in Step S5, after normalization and denoising processing, is used as the input data for the random forest model in Step S6. Among them, the specific construction steps of the random forest model in Step S6 include the following steps: Step S61: Extract the time domain features of the fluorescence signals of sodium ions and potassium ions, including signal peak value, mean value, standard deviation, and sweat secretion cycle features. Step S62: Extract the frequency domain features of the fluorescence signals of sodium ions and potassium ions, including power spectral density, main frequency component, and sweat secretion frequency features. Step S63: Input the extracted time domain and frequency domain feature vectors into the random forest model. Step S64: Set the splitting rule of the decision tree based on the mean square error minimization criterion for ion concentration prediction. Step S65: Calculate the final concentration values of sodium ions and potassium ions in sweat by weighted averaging the prediction results of multiple decision trees.

Images