A sweat microfluidic patch, visualized model and preparation method thereof

Abstract

This invention discloses a sweat microfluidic patch, its visualization model, and its preparation method. The visualization model uses characteristic data related to physiological indices in sweat as input variables to establish a correlation model with the microfluidic patch's scale. This model has advantages such as simple structure, strong compatibility, and easy acquisition of relevant data. The sweat microfluidic patch consists of an adhesion layer, a microfluidic main layer, a scale layer, and an encapsulation layer. Its simple structure requires no external power supply and meets the requirements for rapid detection and accurate reading. The preparation method involves pre-coating pigments and detection reagents onto the microfluidic main layer, and then encapsulating it after the solution has completely dried. This preparation method is simple, has low production costs, requires no additional investment, and is suitable for continuous industrial production.

Description

Technical Field

[0001] This invention relates to a sweat microfluidic patch, specifically to a sweat microfluidic patch, its visualization model, and its preparation method, belonging to the novel medical field. Background Technology

[0002] Currently, there is increasing attention being paid to the link between human physiological markers and individual health status. Human sweat contains abundant electrolytes, metabolites, proteins, hormones, and exogenous substances. The presence or changes in the concentration of these biomarkers can affect or predict the incidence of diseases. Related analytical and detection technologies, which can collect and detect biomarkers non-invasively, have become important tools for analyzing human biomarker levels and health status. For example, cortisol in human sweat is believed to be associated with major depressive disorder and stress disorder; the presence of cytokines in sweat is significant for detecting inflammation; chloride ion concentration can detect whether infants have cystic fibrosis; sodium ion concentration can reflect the level of electrolytes and the probability of hyponatremia; and glucose concentration can reflect blood sugar levels. Wearable devices on the market, such as smartwatches and smart bracelets, can monitor various physiological markers. However, these wearable devices are made of rigid materials, which can cause discomfort to users to some extent and limit their use to the wrist. Furthermore, electronic devices are expensive. Researchers both domestically and internationally have conducted extensive work to detect physiological markers in human sweat. Researchers have reportedly designed a flexible wearable patch that integrates electrochemical detection units, including those for potassium ions, sodium ions, glucose, and lactic acid, enabling real-time monitoring of human sweat composition. However, electrochemical methods for detecting sweat components require complex data acquisition and transmission modules, increasing manufacturing complexity and cost. In particular, most recently developed wearable microfluidic patches employ colorimetric methods to detect chloride ions, pH, glucose, and lactic acid in sweat. However, this method requires image acquisition equipment and relies on lighting conditions, and cannot provide users with direct, accurate readings. The discontinuous nature of sweat secretion on the human body makes its collection and analysis uncontrollable, prone to evaporation and contamination. Traditional distance-based detection methods require quantitative liquid administration each time, making them unsuitable for wearable, real-time sweat detection. Therefore, developing a low-cost sweat detection patch capable of high-throughput, multi-time-period, visualized detection, and real-time alerts is imperative. Summary of the Invention

[0003] To address the problems existing in the prior art, the first objective of this invention is to provide a sweat microfluidic patch, which comprises an adhesive layer, a microfluidic main layer, a scale layer, and an encapsulation layer. This patch has a simple structure, is easy to use, requires no external power supply, and can meet the requirements for rapid detection and accurate readings.

[0004] The second objective of this invention is to provide a visualization model for a sweat microfluidic patch. This model uses characteristic data related to physiological indices in sweat as input variables to establish a correlation model with the microfluidic patch scale. This model has a simple structure, strong compatibility, and readily available data, enabling rapid and accurate analysis and visualization of sweat composition.

[0005] The third objective of this invention is to provide a method for preparing a sweat microfluidic patch. This method involves pre-coating a pigment and a detection agent onto a microfluidic master layer, and then encapsulating the patch after the solution has completely dried. This preparation method is simple, has low production costs, requires no additional investment, and is suitable for continuous industrial production.

[0006] To achieve the above technical objectives, this invention provides a visualization model of a sweat microfluidic patch, comprising the following steps:

[0007] Step 1): Obtain characteristic data related to physiological indices in sweat;

[0008] Step 2): Noise reduction and preprocessing of characteristic data related to physiological indices in sweat;

[0009] Step 3): Obtain correlation data between physiological indices and microfluidic patch scale.

[0010] Step 4): Establish a visualization model between characteristic data related to physiological indices in sweat and microfluidic patch scale;

[0011] Step 5): Perform performance testing and parameter optimization on the model obtained in Step 4).

[0012] This invention extracts feature data related to physiological indices from sweat components, and uses physiological indices as a bridge to establish a visualization model between sweat and microfluidic patch scale. This model can quickly and accurately analyze the components in sweat and visualize these components. The model has good regression performance, high accuracy, and an error precision of less than 1%.

[0013] As a preferred embodiment, the characteristic data related to physiological indices in the sweat include: sweat volume, glucose concentration, and chloride ion concentration.

[0014] As a preferred embodiment, the noise reduction and preprocessing process includes: I) labeling and classifying the feature data; II) cleaning outlier data in the feature data, setting gradient intervals for the cleaned data, and performing normalization. The feature data is classified according to volume, glucose concentration, and chloride ion concentration. When the obtained feature data significantly exceeds or falls below the acquisition range, it is cleaned and removed. The main purpose of normalization is to eliminate the influence of dimensions on the feature data, converting it into a dimensionless expression and simplifying calculations.

[0015] As a preferred embodiment, the changes in the microfluidic patch scale with the amount of sweat, the changes in the microfluidic patch scale with the amount of glucose concentration, and the changes in the microfluidic patch scale with the amount of chloride ion concentration are described.

[0016] As a preferred embodiment, the visualization model between the feature data related to physiological indices in the sweat and the microfluidic patch scale is a linear model, a nonlinear model, or a machine learning model.

[0017] As a preferred embodiment, the process of establishing the machine learning model includes:

[0018] i) 50-70% of the data related to physiological indices in sweat are used as training set data, 10-20% of the data are used as test set data, and the rest are used as validation set data;

[0019] ii) Establish a neural network model for both using the training set data as the input variable and the microfluidic patch scale as the output variable. The number of neurons in the hidden layer is 5 to 15, and the training algorithm is as follows;

[0020] iii) Use test set data to test the accuracy of the neural network model.

[0021] As a preferred approach, the process of model performance testing and parameter optimization is as follows: the actual predictive ability of the model is tested using validation set data, and the parameters are optimized using the LM algorithm.

[0022] The present invention also provides a sweat microfluidic patch, obtained from the above-mentioned visualization model, comprising: an adhesion layer, a microfluidic main layer, a scale layer, and an encapsulation layer; the microfluidic main layer is one of a single-channel, dual-channel, and multi-channel parallel structure; the adhesion layer, the microfluidic main layer, the scale layer, and the encapsulation layer are stacked sequentially from bottom to top.

[0023] As a preferred embodiment, the adhesion layer is provided with a liquid inlet; the microfluidic main layer is provided with a liquid inlet area, a pigment area, a reagent area, a reference channel and a reaction channel; the liquid inlet area is located directly above the liquid inlet.

[0024] As a preferred embodiment, the pigment area and the reagent area are independently located on opposite sides of the liquid inlet area.

[0025] As a preferred embodiment, one end of the pigment area is connected to the liquid inlet area, and the other end is connected to the reference channel.

[0026] As a preferred embodiment, one end of the reagent area is connected to the liquid inlet area, and the other end is connected to the reaction channel.

[0027] As a preferred embodiment, the scale layer is provided with a reference scale, a health reminder scale, and a marker concentration scale.

[0028] As a preferred embodiment, the reference scale is located on one side of the reference channel, the marker concentration scale is located on one side of the reaction channel, and the health reminder scale is located between the reference scale and the marker concentration scale.

[0029] The reading method of the sweat microfluidic patch described in this invention is as follows: the distance the pigment moves with the sweat is recorded as R on the reference scale (8), and the length of the precipitation of the marker in the sweat is recorded as H on the health reminder scale (9); when R reaches a certain area (Rn, n=1,2,3,…) on the reference scale (8), the distance L of the precipitation generated by the sweat marker should be read from the nth reading bar (Ln) on the marker concentration scale (10) to determine the marker concentration.

[0030] Furthermore, the reading method for the health reminder scale is as follows: when R equals H, the concentration of the marker is the dividing point between a healthy state and an unhealthy state. When the concentration of the disease marker is higher than the dividing point, R greater than H indicates a healthy state, and R less than H indicates a sub-healthy or diseased state. When the concentration of the disease marker is lower than the dividing point, R less than H indicates a healthy state, and R greater than H indicates a sub-healthy or diseased state.

[0031] As a preferred embodiment, multiple microfluidic master layers are connected in parallel and share the same liquid inlet area. A slow-release agent is added at the connection points, enabling multi-time-segment detection of the microfluidic master layer. An appropriate amount of slow-release agent can effectively slow down the flow of liquid. By setting a fixed concentration gradient for the slow-release agent and adding different concentrations of slow-release agent to different connection points, multi-time-segment detection of the microfluidic master layer can be achieved.

[0032] As a preferred embodiment, the slowing agent is at least one of polyvinyl alcohol, sodium polystyrene sulfonate, sodium carboxymethyl cellulose, and Triton solution.

[0033] The present invention also provides a method for preparing a sweat microfluidic patch, the main process of which includes: injecting a colored solution into the pigment region of the microfluidic main layer and injecting a detection reagent into the reagent region; after all the injected reagents have dried completely, they are adhered to the adhesive layer and covered with a scale layer and an encapsulation layer, thus obtaining the patch.

[0034] As a preferred embodiment, the present invention also provides a detailed preparation method for a sweat microfluidic patch, the main steps of which are as follows: For a microfluidic patch for detecting chloride ions, the preparation process is as follows: 1) Select a porous medium as the microfluidic master layer, cut it into the required shape and divide it into sections; 2) Put potassium chromate solution and silver nitrate solution into empty ink cartridges of an inkjet printer respectively, print potassium chromate solution 20 times on the reaction channel of the microfluidic master layer with an inkjet printer, and after the solution dries, print silver nitrate solution 20 times. Potassium chromate reacts with silver nitrate to form a reddish-brown silver chromate precipitate; 3) Add pigment solution to the pigment area of ​​the microfluidic master layer; 4) After all reagents are completely dry, adhere the microfluidic layer to the adhesion layer, and cover it with a scale layer and an encapsulation layer to obtain the patch.

[0035] The preparation process of the microfluidic patch for glucose detection is as follows: 1) Select a porous medium as the microfluidic master layer, cut it into the required shape and divide it into sections; 2) Fill the empty ink cartridges of the inkjet printer with ferric chloride solution, potassium ferrocyanide solution and ascorbic acid solution respectively. Print ferric chloride solution 20 times on the reaction channel of the microfluidic master layer, and then print potassium ferrocyanide solution 20 times in the corresponding position. Ferric chloride reacts with potassium ferrocyanide to generate blue Prussian blue precipitate. Finally, print ascorbic acid solution in the same position. Prussian blue precipitate reacts with ascorbic acid to generate white Prussian white precipitate; 3) Add glucose oxidase solution dropwise to the reagent area of ​​the microfluidic master layer; 4) Add red pigment solution dropwise to the pigment area. After the solution dries, the patch is obtained; 5) After all reagents are completely dry, adhere the microfluidic patch layer by layer to the adhesion layer, and cover it with a scale layer and an encapsulation layer.

[0036] In order to improve the accuracy and speed of detection, the size, shape and number of inkjet prints of the microfluidic master control layer were further optimized in this invention.

[0037] As an optimization approach, lasers were used to cut the microfluidic master layer into different shapes: straight, serpentine, and spiral. When 10 μL of 50 mM chloride ion solution was added, the straight filter paper produced the longest precipitate, while the serpentine filter paper had the best tensile properties, and the spiral filter paper had the smallest area ratio.

[0038] As a preferred embodiment, the liquid inlet area (3) is a circle with a diameter of 5 mm, the pigment area (4) and the reagent area (5) are circles with a diameter of 4 mm, and the reference channel (6) and the reaction channel (7) are 35 mm long and 1.5 mm wide.

[0039] As a preferred method, in order to measure the effect of the number of times the inkjet printer prints the reagent on the length of the precipitate, the reagent was printed 5, 7, 10, 12, 15, 17, 20 and 25 times respectively. Then, the same volume and concentration of chloride ion solution were added. As the number of printing times increased, the length of the precipitate became shorter. The precipitate length generated by printing the reagent 20 times was most suitable for measuring the concentration of chloride ion solution in sweat.

[0040] As a preferred embodiment, the present invention also provides a microfluidic patch for multi-time period sweat detection, the main preparation process of which is as follows:

[0041] As a preferred embodiment, the colored solution is a magenta solution and / or food coloring.

[0042] As a preferred embodiment, the detection reagent is at least one of potassium chromate, silver nitrate, ferric chloride, potassium ferrocyanide, ascorbic acid, and glucose oxidase solution.

[0043] The sweat microfluidic patch provided by this invention offers an instant detection solution for diseases such as cystic fibrosis and hyperglycemia, helping patients or potential patients to perform real-time self-examination and preventative self-diagnosis. This patch creatively establishes a visual model between data related to physiological indices in sweat and the microfluidic patch's scale, and uses machine learning methods to optimize the model, enhancing its accuracy. Furthermore, the patch is soft and skin-friendly, provides accurate testing, requires no external power supply or display, is inexpensive, and is disposable, effectively expanding the patch's application scenarios and avoiding the risk of cross-infection caused by mixing patches.

[0044] Compared with the prior art, the beneficial technical effects of the technical solution of the present invention are as follows:

[0045] 1) The visualization model provided by this invention has strong compatibility and can be adjusted according to actual needs. The data sources required by the model are wide-ranging and the structure is simple. The accuracy and applicability of the model can be further improved through machine learning algorithms. It can quickly and accurately analyze and visualize the composition of sweat.

[0046] 2) The patch provided by the present invention consists of an adhesion layer, a microfluidic main layer, a scale layer and an encapsulation layer. The patch has a simple structure, is easy to use, does not require an external power supply, and can meet the requirements of rapid detection and accurate reading.

[0047] 3) In the technical solution provided by this invention, the pigment and detection reagent are pre-coated onto the microfluidic master layer, and then encapsulated after the solution is completely dried. This preparation method is simple, has low production cost, requires no additional investment, and is suitable for continuous industrial production. Attached Figure Description

[0048] Figure 1 This is a schematic diagram of the sweat microfluidic patch in Example 1.

[0049] Figure 2 This represents the linear relationship between the amount of solution absorbed by the sweat microfluidic patch and the distance the pigment moves in Example 2.

[0050] Figure 3 Example 3 was used to investigate the linear relationship between the length of the generated silver chloride precipitate and different concentrations of sodium chloride solution at absorption volumes of 10 μL, 12 μL, and 15 μL for the sweat microfluidic patch.

[0051] Figure 4 The linear relationship between the addition of different amounts of 60mM sodium chloride solution to the sweat microfluidic patch in Example 4 and the length of the generated silver chloride precipitate was investigated.

[0052] Figure 5 Example 5 was used to test the linear relationship between the length of the Prussian blue precipitate generated from 10 μL, 12 μL, and 15 μL glucose solutions of different concentrations on the sweat microfluidic patch.

[0053] Figure 6 This study examines the performance and applications of different shaped sweat microfluidic patches in Example 6.

[0054] Figure 7 This is a schematic diagram of the sweat microfluidic patch used to monitor the hydration status of the human body in Example 7.

[0055] Figure 8 The sweat microfluidic patch in Example 8 is used to detect the chloride ion concentration in sweat using three methods of stimulating sweat: ion penetration, thermal stimulation, and exercise.

[0056] Figure 9 This is a chloride ion concentration detection diagram for a delayed-action sweat microfluidic patch in Example 9.

[0057] Figure 10 The detection performance of the sweat microfluidic patch after folding in Example 10.

[0058] Figure 11 This is a rapid calibration diagram of the neural network visualization model of the sweat microfluidic patch in Example 11.

[0059] Figure 12 This is a simplified neural network diagram of the neural network visualization model of the sweat microfluidic patch in Example 11.

[0060] Figure 13 This is a comparison of the sweat chloride ion concentration detected by the sweat microfluidic patch with dual-scale reading and the single-channel distance method chloride ion detection patch described in Example 12.

[0061] Figure 14This is a comparison of the detection performance of the sweat microfluidic patch dual-scale reading method described in Example 13 with that of commercial glucose test strips. Detailed Implementation

[0062] To illustrate the technical methods employed to achieve the intended objectives and advantages of this invention, a detailed and comprehensive description is provided below in conjunction with the accompanying drawings and preferred embodiments. Obviously, the described embodiments are merely a part of the embodiments of this invention, and not all of them. Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without inventive effort are within the scope of protection of this invention.

[0063] Example 1

[0064] The bottom of the sweat microfluidic patch is an adhesive layer (1), which is made of laser-cut double-sided adhesive into a specific shape. A circular hole is cut into the double-sided adhesive as the sweat inlet (13). Above the adhesive layer (1) is the microfluidic main layer (2), which is made of laser-cut filter paper into a specific shape. The microfluidic main layer consists of an inlet area (3), a pigment area (4), a reagent area (5), a reference channel (6), and a reaction channel (7). The inlet area (3) is used to draw sweat into the paper-based microfluidic chip. A pigment solution is added to the reagent area (4). An enzyme solution or other pre-reaction reagents are added to the reagent area (5). The reaction channel (7) contains reaction reagents for sweat physiological markers. The reaction reagents are printed directly onto the filter paper by an inkjet printer. Above the filter paper layer is a scale layer (11), which consists of a reference scale (8), a health reminder scale (9), and a sweat marker concentration scale (10). The scale layer (11) is made of printable PET self-adhesive stickers, and the scales are printed onto the PET self-adhesive stickers by an inkjet printer. Above the scale layer (11) is an encapsulation layer (12), which is waterproof, abrasion-resistant, and protects the wearable patch from external environmental influences.

[0065] Example 2

[0066] The linear relationship between the amount of solution absorbed by the sweat microfluidic patch and the distance the pigment migrates was investigated. The procedure involved adding 5 μL, 7 μL, 10 μL, 12 μL, 15 μL, and 20 μL of ultrapure water. The distance the pigment migrated along the water flow path in the sweat volume detection channel was measured. Four replicates were performed for each solution volume. A linear fit was then performed between the solution volume and the pigment migration distance, yielding the linear equation y = 0.2521x - 0.8060 for solution volumes from 5 μL to 15 μL, with R0. 2 =0.9904, linear equation R 2A value greater than 0.95 indicates a good linear fit, meaning there's a linear relationship between solution volume (5 μL to 15 μL) and pigment migration distance. However, solution volumes exceeding 15 μL exceed the range of the wearable patch. Therefore, we can measure the pigment migration distance to reflect perspiration volume. Figure 2 A reference scale (8) was created to detect the amount of sweat.

[0067] Example 3

[0068] The linear relationship between chloride ion concentrations (10 μL, 12 μL, and 15 μL) and precipitate length was investigated. Chloride ion solutions of different concentrations (20 mM, 30 mM, 50 mM, 60 mM, 75 mM, 100 mM, 120 mM, and 150 mM) were added to a wearable chloride ion detection patch. The length of the white silver chloride precipitate formed on the biomarker detection channel was measured. Four replicates were performed for each concentration. Linear fitting was then performed on the precipitate length for chloride ion concentrations of 10 μL, 12 μL, and 15 μL. The linear equation for chloride ion concentrations from 20 mM to 120 mM was obtained as y = 0.0092x + 0.0349, R0. 2 =0.9829, 12μL is y = 0.0119x + 0.1625, R 2 =0.9900, 15μL is y = 0.0138x + 0.3004, R 2 =0.9836, three sets of linear equations R 2 The values ​​all being greater than 0.95 indicate a good linear fit, suggesting a linear relationship between chloride ion concentration (20 mM to 120 mM) and the length of precipitate formation. However, chloride ion concentrations above 120 mM exceed the range of the wearable patch. A chloride ion concentration exceeding 60 mM in human sweat increases the probability of cystic fibrosis. A chloride ion concentration detection range of 20 mM to 120 mM can provide immediate alerts for cystic fibrosis. Therefore, we can use the length of precipitate formation and the amount of sweat to reflect the chloride ion concentration in sweat, thus providing an immediate alert for cystic fibrosis. Figure 3The scale for measuring physiological markers of sweat on the chloride ion detection patch is set (8). The scales 5 to 15 on the left represent solution volumes of 5 μL to 15 μL. The reading reached by the pigment tip is the current amount of sweat contained in the paper-based chip. The blue, green, and red scales on the right represent chloride ion concentrations from 20 mM to 120 mM for 10 μL, 12 μL, and 15 μL, respectively. When the pigment tip on the left reaches scale 10, the current chloride ion concentration can be read from the blue scale reached by the silver chloride precipitate tip on the right. Afterward, when the pigment tip reaches scales 12 and 15, the chloride ion concentrations are read from the green and red scales reached by the silver chloride precipitate tip on the right, respectively. The reason for setting multiple scales corresponding to different sweat volumes is that human sweating is a continuous process, and we cannot control the amount of sweat during testing. Setting a single point may not reach or exceed the set scale line, while setting multiple points can reduce this situation. Furthermore, setting multiple points also allows for analysis of changes in human chloride ion concentration over time.

[0069] Example 4

[0070] The linear relationship between different volumes of 60mM chloride ion solution in a sweat microfluidic patch and the formation of precipitate was investigated. 5μL, 7μL, 10μL, 12μL, and 15μL of 60mM chloride ion solution were added to the chloride ion detection wearable patch. The length of the white silver chloride precipitate formed on the biomarker detection channel was measured. Four replicates were performed at each concentration point. A linear fit was then performed on the relationship between different volumes of 60mM chloride ion solution and the precipitate length, yielding the linear equation y = 0.1086x - 0.4945, R0. 2 =0.9874, linear equation R 2 A value greater than 0.95 indicates a good linear fit, suggesting a linear relationship between the 60 mM chloride ion solution (5 μL to 15 μL) and the length of precipitate formation. The probability of cystic fibrosis is higher when the chloride ion concentration in human sweat exceeds 60 mM. Therefore, a precipitate formation length above the linear equation indicates a possible diagnosis of cystic fibrosis, while a length below the linear equation indicates an unlikely diagnosis. Figure 4 Make a health reminder scale for the chloride ion detection patch (9). The 60mM chloride ion scale line corresponds one-to-one with the reference scale line on the left. When the reference scale line where the pigment tip is located on the left is equal to the 60mM chloride ion scale line where the silver chloride precipitate tip is located on the right, the chloride ion concentration is 60mM. Therefore, when the reference scale line where the pigment tip is located on the left is greater than the 60mM chloride ion scale line where the silver chloride precipitate tip is located on the right, the chloride ion concentration exceeds 60mM. In this case, the person being tested is very likely to have cystic fibrosis.

[0071] Example 5

[0072] The linear relationship between 10 μL, 12 μL, and 15 μL of glucose solution at different concentrations and the length of precipitate formation was investigated using a sweat microfluidic patch. Glucose solutions of different concentrations (100 μM, 200 μM, 300 μM, 500 μM, and 700 μM) were added to the wearable glucose detection patch, and the length of the blue Prussian blue precipitate formed on the biomarker detection channel was measured. Four replicates were performed for each concentration. Linear fitting was then performed on the precipitate length for 10 μL, 12 μL, and 15 μL of glucose solution at different concentrations. The linear equation for glucose concentrations from 100 μM to 500 μM was obtained as y = 0.0018x + 0.0117, R0. 2 =0.9980, 12μL is y = 0.0020x + 0.1073, R 2 =0.9752, 15μL is y = 0.0036x - 0.0663, R 2 =0.9996, three sets of linear equations R 2 The values ​​all being greater than 0.95 indicate a good linear fit, suggesting a linear relationship between glucose concentration (100μM to 500μM) and precipitate formation length. However, glucose concentrations above 500μM exceed the range of the wearable patch. Human sweat chloride ion concentrations exceeding 200μM increase the probability of hyperglycemia. A glucose concentration detection range of 100μM to 500μM can provide immediate alerts for hyperglycemia. Therefore, we can use the precipitate formation length and sweat volume to reflect the glucose concentration in sweat, thus providing an immediate alert for hyperglycemia. Figure 3 The scale for measuring sweat physiological markers on the glucose detection patch is set (8). The scale from 5 to 15 on the left represents the solution volume from 5 μL to 15 μL. The scale reading reached by the pigment tip is the amount of sweat contained in the paper chip. The blue, green, and red scale lines on the right represent chloride ion concentrations from 100 μM to 500 μM, respectively, for 10 μL, 12 μL, and 15 μL. When the pigment tip on the left reaches scale 10, the current glucose concentration can be read from the blue scale line reached by the Prussian blue precipitate tip on the right. Afterward, when the pigment tip reaches scales 12 and 15, the glucose concentration can be read from the green and red scale lines reached by the Prussian blue precipitate tip on the right, respectively.

[0073] Example 6

[0074] The sweat microfluidic patch is fabricated using lasers and templates, thus offering high scalability. We investigated the performance of three channel types—spiral, serpentine, and linear—at widths of 1 mm, 1.5 mm, and 2 mm for chloride ion detection. After adding 10 μL of 60 mM chloride ion solution, the figure shows that the 1 mm linear channel has the longest detection distance for chloride ions. Under the same chloride ion concentration, a narrower channel width results in a longer deposition distance, thus increasing accuracy, but also shortening the measurement range. The linear channel has a slightly longer deposition distance than the other two channels, possibly because the curved channel affects the solution flow rate. We then compared the ratio of the effective distance to the patch length for the three channel shapes on the paper-based chip, finding that the spiral channel had the largest ratio. Therefore, under the same patch length, the spiral channel can collect more sweat for detection. By utilizing the properties of paper-based chips of different shapes, we can make them suitable for application to different parts of the human body. For example, the snake-shaped type can be applied to the neck area because of its excellent stretchability, the spiral type can collect more sweat in a very short length and is suitable for application to areas with high sweating, while the straight type is easy to make with scales for easy reading, but needs to be applied to flat areas.

[0075] Example 7

[0076] To verify the practicality of the sweat microfluidic patch, it was used to detect changes in chloride ion content in sweat produced during exercise in both normal and dehydrated states. Three sets of experiments were conducted on the same individual in both normal and dehydrated states. In the normal state, hydration was initiated 1 hour prior to exercise; in the dehydrated state, a round of exercise was performed, resulting in sweating and significant thirst. After applying the chloride ion detection patch to the same area, the same amount of exercise was performed, followed by a 10-minute rest before testing. The graph shows that by comparing the ratio of the 60mM chloride ion scale to the reference scale, the ratio in the normal state was significantly lower than that in the dehydrated state. This indicates that the chloride ion concentration in sweat is significantly lower in the normal state than in the dehydrated state. Therefore, this patch can be used to detect a person's hydration status during exercise and provide timely reminders to rehydrate.

[0077] Example 8

[0078] This study compared the concentration of chloride ions in sweat detected by a sweat microfluidic patch using three methods of stimulating sweat production. Iontophoresis involves using a DC power source to deliver pilocarlin cations from a hydrogel at the positive electrode to the skin, stimulating sweat glands to induce sweating. This method can induce sweating at rest. Thermal stimulation involves sweating through activities such as drinking hot water or using a sauna. Exercise induces sweating through physical activity. By comparing the ratio of the 60mM chloride ion concentration to the reference concentration, we can see that the ratio for sweating during exercise is higher than that of iontophoresis or thermal stimulation. This indicates that sweating during exercise leads to a higher detected chloride ion concentration, possibly due to changes in the body's hydration state during exercise. Since both iontophoresis and thermal stimulation involve sweating at rest, the changes in chloride ion concentration are relatively small. Therefore, iontophoresis or thermal stimulation is the preferred method for stimulating sweating in cystic fibrosis testing.

[0079] Example 9

[0080] A delayed-action sweat microfluidic patch can be used to remind people to rehydrate when dehydrated. To verify that the added solution that slows down the liquid flow does not affect the wearable patch's ability to detect the concentration of physiological markers in sweat, we designed the following experiment. A chloride ion detection wearable patch with an inlet connected to two detection channels was prepared. One channel (channel 1) was not connected with sodium polystyrene sulfonate solution, and the other channel (channel 2) was connected with 15% w / v sodium polystyrene sulfonate solution. The device was dried in a 35°C oven for 30 min. Then, a 20 mM to 60 mM chloride ion solution was injected into the inlet of the chloride ion detection wearable patch at a flow rate of 2 μL / min using a syringe pump. The chloride ion solution first flowed into channel 1, and then into channel 2 after about 5 min. Figure 9 As shown, by comparing the distances of the silver chloride precipitate on the right side when the pigment tip reaches 0.8 cm in channels 1 and 2, the difference between the two distances is within the allowable error range. Therefore, the effect of the added solution that slows down the liquid flow on the ability of the paper-based wearable patch to detect the concentration of sweat physiological markers is negligible. We obtained a linear relationship between chloride ion concentration and precipitation distance by linearly fitting the distance of the silver chloride precipitate on the right side when the pigment tip reaches 0.8 cm in 20 mM to 60 mM chloride ion solutions. Then, we marked a red line at 0.8 cm on the left side of the detection channel and made a chloride ion concentration scale on the right side of the detection channel based on the linear relationship between chloride ion concentration and precipitation distance. When the pigment on the left side reaches the 0.8 cm red line, the chloride ion concentration can be read from the scale corresponding to the precipitation distance on the right side. Since we've previously demonstrated that the concentration of chloride ions in human sweat increases due to dehydration, we can attach our designed time-delayed chloride ion detection wearable patch to the human body to monitor dehydration in real time. To verify this function, we attached the time-delayed chloride ion detection wearable patch to the back of test subjects and had them exercise. Figure 9 As shown, when the tester exercised for 10 minutes, sweat had already flowed into channel 1, which did not have sodium polystyrene sulfonate solution added, while no sweat had entered channel 2, which had 15% w / v sodium polystyrene sulfonate solution added. When the tester exercised for 16 minutes, the pigment in channel 1 had reached the 0.8 cm mark, allowing for the reading of sweat chloride ion concentration, while sweat had just entered channel 2. When the tester exercised for 29 minutes, the pigment in channel 2 had reached the 0.8 cm mark, allowing for the reading of sweat chloride ion concentration. By comparing the chloride ion concentrations in channel 1 and channel 2 for four testers, it can be seen that the chloride ion concentration in channel 2 was higher than that in channel 1. The increase varied due to the different sweating rates of each individual. Of course, we can set up more channels to further detect the chloride ion concentration in sweat during exercise over a longer period, thereby providing real-time reminders for people to replenish water when dehydrated.

[0081] Example 10

[0082] The sweat microfluidic patch was tested for resistance to external interference. The patch was bent once, twice, and three times, and then 15 μL of aqueous solution was added to simulate the flow of sweat on filter paper. Each experiment had at least three parallel sets. The average distance of the pigment front was compared with the distance of the pigment front of an unbent paper-based patch after adding 15 μL of aqueous solution. The error was within 0.1 mm, which is within an acceptable range. This indicates that the sweat microfluidic patch can withstand multiple bends and compressions without affecting its detection performance.

[0083] Example 11

[0084] A neural network visualization model of a sweat microfluidic patch was developed, and the Neural NetFitting function in MATLAB 2018b was used for neural network fitting and prediction. Experimental data was imported into the homepage, with 80 sets of data named "into" containing the following input variables: filter paper width (mm), potassium chromate concentration (mM), number of prints, liquid inlet volume (μL), and chloride ion concentration (mM). The output variables were the same: liquid inlet distance (cm) and chloride ion reaction distance (cm), with 80 sets of data named "out". The Neural NetFitting function was opened, and the data was placed into the input and output fields. This test used 10 hidden layer neurons for computation, with the artificial neural network comprising 70% of the training set, 15% of the test set, and 15% of the validation set. The Levenberg-Marquardt training algorithm was used, yielding a training accuracy plot and a linear regression plot. The error accuracy was below 1%, indicating good fitting performance. Then, solutions of different volumes (5, 7, 10, 12, 15, and 20 μL) were dropped onto a paper-based patch with a channel width of 1.5 mm, and the distances were measured. These distances were then compared with the distances predicted by machine learning, and a good fit was obtained.

[0085] Example 12

[0086] This study compares the single-channel distance method with the dual-channel dual-scale reading method of a sweat microfluidic patch, where both patches are applied to the skin to detect chloride ion concentration in sweat. The single-channel distance method, an existing technology, typically requires adding a fixed volume of solution for detection. For discontinuous and non-uniform sweating, it can only determine the amount of marker based on the distance of the reaction color, not the specific concentration. Our designed dual-channel dual-scale reading chloride ion detection patch can read the marker concentration even with irregular sweating rates. As shown in the figure, when the red pigment tip of the left channel reaches the left scale value of 10, the concentration can then be read from the blue scale value reached by the white precipitate tip of the right channel. The chloride ion concentration detected in the sweat shown in the figure is approximately 30 mM.

[0087] Example 13

[0088] A comparison of glucose detection between commercial glucose test strips (manufactured by Guilin Youlite Medical Electronics Co., Ltd.) and sweat microfluidic patches. Commercial glucose test strips use a colorimetric method to detect glucose concentration, with a detection range of 2mM to 100mM, while our designed glucose detection patch has a detection range of 0.1mM to 0.5mM, which better matches the glucose concentration detection range of sweat. Furthermore, Figure 13The results showed that commercial glucose test strips, which use a colorimetric method to detect glucose concentration, are affected by light conditions. The colors obtained in bright and dark environments differ significantly. When comparing the G values ​​in the RGB values, the G values ​​for 5mM glucose and 10mM glucose show a noticeable decrease in both light and dark areas, which seriously affects the accuracy of concentration detection. In contrast, our designed glucose detection patch uses a dual-scale reading method to detect glucose concentration, which is unaffected by light conditions. The glucose concentration can be obtained simply by reading the distance of color change in both channels, and the glucose concentration value read is the same regardless of whether it is in bright or dark environments.

Claims

1. A method of building a visual model of a sweat microfluidic patch, characterized in that, The method comprises the following steps: Step 1): obtaining feature data related to physiological indexes in sweat; Step 2): denoising and preprocessing the feature data related to physiological indexes in sweat; Step 3): obtaining correlation data between physiological indexes and microfluidic patch scales; Step 4): establishing a visualization model between the feature data related to physiological indexes in sweat and the microfluidic patch scales; Step 5): detecting the performance of the model obtained in step 4) and optimizing the parameters; The feature data related to physiological indexes in sweat includes sweat volume, glucose concentration and chloride concentration; The denoising and preprocessing process includes: I) labeling and classifying the feature data; II) cleaning abnormal data in the feature data, setting a gradient interval for the cleaned data, and performing normalization processing; The correlation data between physiological indexes and microfluidic patch scales includes: the change amount of the microfluidic patch scale with the change of sweat volume, the change amount of the microfluidic patch scale with the change of glucose concentration, and the change amount of the microfluidic patch scale with the change of chloride concentration; The visualization model between the feature data related to physiological indexes in sweat and the microfluidic patch scales is a linear model, a nonlinear model or a machine learning model; The machine learning model is established by: i) using 50-70% of the feature data related to physiological indexes in sweat as training set data, 10-20% of the data as test set data, and the remaining data as validation set data; ii) establishing a neural network model of the training set data as input variable and the microfluidic patch scale as output variable, with 5-15 hidden layer neurons and Levenberg-Marquardt training algorithm; When the feature data related to physiological indexes in sweat is sweat volume, the microfluidic patch scale is the distance of the pigment moving with the sweat; when the feature data related to physiological indexes in sweat is glucose concentration and chloride concentration, the microfluidic patch scale is the distance of the precipitate produced by the sweat marker; iii) testing the accuracy of the neural network model using the test set data.

2. The method of establishing a visual model of a sweat microfluidic patch of claim 1, wherein: The model performance detection and parameter optimization process is to detect the actual prediction ability of the model by the validation set data, and to optimize the parameters by L-M algorithm.

3. A sweat microfluidic patch, characterized by: The model established based on the method of claim 1 or 2 comprises an adhesion layer, a microfluidic main layer, a scale layer and a packaging layer; the microfluidic main layer is one of single-channel, double-channel and multi-channel parallel structure; the adhesion layer, the microfluidic main layer, the scale layer and the packaging layer are stacked from bottom to top; The adhesion layer is provided with a liquid inlet; the microfluidic main layer is provided with a liquid inlet area, a pigment area, a reagent area, a reference channel and a reaction channel; the liquid inlet area is located directly above the liquid inlet; the pigment area and the reagent area are independently distributed on both sides of the liquid inlet area; one end of the pigment area is connected to the liquid inlet area, and the other end is connected to the reference channel; one end of the reagent area is connected to the liquid inlet area, and the other end is connected to the reaction channel; The scale layer is provided with a reference scale, a health reminding scale and a marker concentration scale; the reference scale is located on one side of the reference channel, the marker concentration scale is located on one side of the reaction channel, and the health reminding scale is located between the reference scale and the marker concentration scale; the scale reading of the reference scale reached by the front end of the pigment on the left side is the amount of sweat contained in the current paper-based chip; the marker concentration scales of different colors on the right side respectively correspond to the glucose concentration under different amounts of sweat and the chloride ion concentration under different amounts of sweat.

4. The sweat microfluidic patch of claim 3, wherein: A plurality of microfluidic main layers are connected in parallel and share the same liquid inlet area, and a slow-release agent is added at the connection, thereby obtaining a multi-period detection microfluidic main layer.

5. The sweat microfluidic patch of claim 3 or 4, wherein: The preparation process is as follows: colored solution is injected into the pigment area of the microfluidic main layer, and detection reagent is injected into the reagent area; after all the injected reagents are completely dried, they are adhered to the adhesion layer, and the scale layer and the packaging layer are covered, thereby obtaining the paper-based chip.

Images