Microfluidic Sliding Paper-Based Device for Point-of-Care Determination of Albumin-to-Creatine Ratio in Human Urine

A novel assay platform consisting of a microfluidic sliding double-track paper-based chip and a hand-held Raspberry Pi detection system is proposed for determining the albumin-to-creatine ratio (ACR) in human urine. It is a clinically important parameter and can be used for the early detection of related diseases, such as renal insufficiency. In the proposed method, the sliding layer of the microchip is applied and the sample diffuses through two parallel filtration channels to the reaction/detection areas of the microchip to complete the detection reaction, which is a simple method well suited for self-diagnosis of ACR index in human urine. The RGB (red, green, and blue) value intensity signals of the reaction complexes in these two reaction zones are analyzed by a Raspberry Pi computer to derive the ACR value (ALB and CRE concentrations). It is shown that the G + B value intensity signal is linearly related to the ALB and CRE concentrations with the correlation coefficients of R2 = 0.9919 and R2 = 0.9923, respectively. It is additionally shown that the ALB and CRE concentration results determined using the proposed method for 23 urine samples were collected from real suffering chronic kidney disease (CKD) patients are in fine agreement with those acquired operating a traditional high-reliability macroscale method. Overall, for point-of-care (POC) CKD diagnosis and monitoring in clinical applications, the results prove that the proposed method offers a convenient, real time, reliable, and low-spending solution for POC CKD diagnosis.


Introduction
Chronic kidney disease (CKD) affects a gradual loss of normal kidney function and is generally associated with diabetes, high blood pressure, or hypertension. In the early stage of its development, CKD shows very few signs or symptoms. However, as the disease progresses, CKD patients may experience nausea and vomiting, fatigue and sleep problems, shortness of breath, lower back pain, and the swelling of the feet and ankles. In severe cases, CKD may also result in vascular disease, cognitive dysfunction, anemia, osteoporosis, and fractures [1]. CKD is generally diagnosed through blood and urine tests. Human urine consists of around 95% water and a large number of smaller constituents, such as uric acid, ALB, CRE, sodium, potassium, ammonia, nitrogen, magnesium, and others [2][3][4][5][6][7][8]. Urine may also include precipitates of calcium, phosphorus, and other compound crystals [9]. In healthy humans, the albumin content in urine is less than 30 mg/g, and any value higher than this may indicate potential kidney disease [10]. Thus, routine urine examinations are advantages over the benchtop techniques described above, including a significantly lower cost, faster throughput, simpler operation, lower reagent consumption, and greater portability [59][60][61][62]. As a result, they have significant potential for the realization of point-of-care testing (POCT) systems [63][64][65][66][67].
Accordingly, this study proposes a simple assay platform for determining the ALB and CRE concentrations in real human urine using a sliding microfluidic paper-based chip and intrusion detection system with Raspberry Pi. Briefly, 50 µL of real human urine is dripped into the sample reservoir of the paper-based microchip and diffuses to two reaction zones containing BCG and picric acid, respectively. The chip is then heated at 40 • C and held for 3 min to actuate the reaction of colorimetric. The Raspberry Pi with software is analyzed the RGB color value intensity of the two reaction areas and the concentrations of ALB and CRE are derived using simple calibration formulae prepared in advance using control samples with known concentrations. The presented detection platform system is confirmed feasibility by comparing the ALB and CRE detection results for 23 real-world urine samples collected from CKD patients with the results obtained from a certified detection technique. The specimen used in this study is a non-invasive sample of urine that can be easily collected without pain and requiring no special equipment. As a result, it has significant potential for point-of-care testing (POCT) or home health monitoring and diagnosis.

Control Samples and Reagents
ALB and CRE control samples were acquired from National Cheng Kung University Hospital (NCKUH, Tainan, Taiwan) together with the corresponding reagents required for reaction purposes. For ALB detection, small strips of filter paper (Advantec Grade No. 4A Hardened Ashless Filter Paper, TOYO ROSHI KAISHA LTD., Tokyo, Japan) were coated with a reagent solution consisting of BCG dye (6.6 mmol/L) mixed with ethanol (70%) and then diluted to approximately pH 4.1 using PBS buffer (0.1 M). Under appropriate reaction conditions (temperature and time), a reaction complex was thus formed as:

Albumin+Bromocresol Green
Similarly, to facilitate CRE detection, small strips of filter paper were coated in a mixed reagent solution consisting of 50 mmol PA (Picric Acid) Reagent 2 (pH = 6.5) and 1 mol NaOH Reagent 1 (pH > 13.5). Colorimetric detection was subsequently performed in accordance with the following Jaffé reaction: Figure 1a illustrates the basic structure of the sliding microfluidic double-track paperbased chip proposed in the present study. As shown, the microchip consisted of five main layers and had overall dimensions of 50 mm × 20 mm × 4.5 mm. The packaging layer consisted of a polyethylene terephthalate (PET) layer coated with a gold membrane for aesthetic purposes and incorporating an opening for the sliding actuator. The second layer was also fabricated of PET and consisted of a large open structure with two color bars adhering to the two side branches to facilitate a qualitative analysis of the ALB and CRE concentrations of the urine sample, respectively. The third and fourth layers of the microchip (the sliding layers) were fabricated of polymethyl methacrylate (PMMA) and contained a sample chamber with dimensions of 5.5 mm × 5 mm, two reaction/detection areas with dimensions of 2.5 mm × 2.5 mm, and two filtration channels with dimensions of 0.3 mm × 5 mm connecting the sample chamber and reaction/detection zones, respectively. Finally, the fifth layer of the microchip was also fabricated of PMMA and served to improve the rigidity of the device and ensure a uniform distribution of the urine sample on the filter paper strips within the two reaction/detection zones. gent-coated strips of filter paper were cut to size (2.5 mm × 2.5 mm) and placed in the two reaction/detection zones, and two small pieces of blood separation membrane (F1, Advanced Microdevices Pvt. Ltd., Saha, India) were introduced into the two microfiltration channels. Two PET layers were bonded using an oxygen plasma treatment and were then sealed with the PMMA layers using a thermal compression technique to form the final microchip assembly (see Figure 1b,d).  Figure 2 shows the hand-held detection system developed in the present study with dimensions of approximately 150 mm × 75 mm × 40 mm and a weight of around 492 g. As shown, the main components of the system which included a Raspberry Pi computer (Raspberry Pi 4 Model B, Tainan, Taiwan), a Raspberry Pi Camera (Sony IMX219, 8-megapixel, Raspberry Pi, Taiwan Taiwan), a temperature controller (DHT 11, TAI-WANIOT, Co., Ltd., Yunlin, Taiwan), a touch switch module (TTP223-BA6, Tainan, Taiwan), a voltage controller (DC-DC, MP1584, Taiwan), two rectangular LED white light sources (CL-S257W8, ShenZhen Caijing, Co., Ltd., Shenzhen, China), and a lithium battery (3.7 V, 4000 mAh, 606090, Tainan, Taiwan).

Hand-Held Detection System
Note that two (rather than one) LEDs were installed in the detection box in order to improve the illumination uniformity of the reaction zone and avoid light halo effects. The system also incorporated an insertion slot for the sliding microchip and a Raspberry Pi smart tablet (Raspberry Pi 3.5″, Taiwan) mounted on top of the detection box (not shown) for visualizing the ALB and CRE concentrations detection results. The various layers within the microchip were designed using AutoCAD (2020) and CorelDRAW (2020) software and were manufactured by the cutting machine with CO 2 laser (GIANT TECH.INC Co., Ltd., Taipei, Taiwan). In the assembly process, the reagentcoated strips of filter paper were cut to size (2.5 mm × 2.5 mm) and placed in the two reaction/detection zones, and two small pieces of blood separation membrane (F1, Advanced Microdevices Pvt. Ltd., Saha, India) were introduced into the two microfiltration channels. Two PET layers were bonded using an oxygen plasma treatment and were then sealed with the PMMA layers using a thermal compression technique to form the final microchip assembly (see Figure 1b,d). Figure 2 shows the hand-held detection system developed in the present study with dimensions of approximately 150 mm × 75 mm × 40 mm and a weight of around 492 g. As shown, the main components of the system which included a Raspberry Pi computer (Raspberry Pi 4 Model B, Tainan, Taiwan), a Raspberry Pi Camera (Sony IMX219, 8-megapixel, Raspberry Pi, Taiwan Taiwan), a temperature controller (DHT 11, TAIWANIOT, Co., Ltd., Yunlin, Taiwan), a touch switch module (TTP223-BA6, Tainan, Taiwan), a voltage controller (DC-DC, MP1584, Taiwan), two rectangular LED white light sources (CL-S257W8, Shen-Zhen Caijing, Co., Ltd., Shenzhen, China), and a lithium battery (3.7 V, 4000 mAh, 606090, Tainan, Taiwan).

Hand-Held Detection System
Note that two (rather than one) LEDs were installed in the detection box in order to improve the illumination uniformity of the reaction zone and avoid light halo effects. The system also incorporated an insertion slot for the sliding microchip and a Raspberry Pi smart tablet (Raspberry Pi 3.5", Taiwan) mounted on top of the detection box (not shown) for visualizing the ALB and CRE concentrations detection results.

ALB and CRE Concentrations Detection Process
In the detection process, the sliding PMMA element in the proposed mic extended and 50 μL of real human urine was dripped into the sample rese sliding element was then retracted, and the urine sample was allowed to di rally through the filtration channels until it reached the two reaction/detec Once the filter paper strips in the reaction/detection zones were fully wetted proximately 10 s), the microchip was inserted into the detection box and the te set to 40 °C for heating, the chip was left for 3 minutes to actuate the colorimet in both reaction/detection zones. The complex compounds formed in the two areas were observed by the Raspberry Pi camera and the resulting images faced to the Raspberry Pi computer. The RGB color intensities of the two c (ALB + BCG and CRE + Picrate) were then analyzed and the ALB and CRE con were determined using calibration formulae prepared in advance using ALB control samples with known concentrations. Finally, the ALB and CRE con were displayed on the Raspberry Pi tablet on top of the detection box.
Notably, the use of a sliding chip design minimizes the effects of evaporat the sample filtration and colorimetric reaction stages and therefore improve intensity value of the colorimetric images and enhances the detection perform result. Furthermore, in contrast to traditional benchtop analytic techniques, th microchip and detection box enable the ALB and CRE concentrations of the ur to be determined simultaneously in a single detection process. As a result, the detection system has substantial potential for low-spending POCT application CKD practice.
In the present study, the ALB and CRE concentrations are determined by berry Pi computer based on an inspection of the RGB intensity signal. H shown in Figure 1a, the microchip also incorporates two color bars adjacent t tion/detection zones to facilitate a qualitative determination of the ALB and centrations. It is thus anticipated that the sliding microchip can also be used standalone system for CKD diagnosis without the need for the full function

ALB and CRE Concentrations Detection Process
In the detection process, the sliding PMMA element in the proposed microchip was extended and 50 µL of real human urine was dripped into the sample reservoir. The sliding element was then retracted, and the urine sample was allowed to diffuse naturally through the filtration channels until it reached the two reaction/detection zones. Once the filter paper strips in the reaction/detection zones were fully wetted (after approximately 10 s), the microchip was inserted into the detection box and the temperature set to 40 • C for heating, the chip was left for 3 min to actuate the colorimetric reaction in both reaction/detection zones. The complex compounds formed in the two detection areas were observed by the Raspberry Pi camera and the resulting images were interfaced to the Raspberry Pi computer. The RGB color intensities of the two compounds (ALB + BCG and CRE + Picrate) were then analyzed and the ALB and CRE concentrations were determined using calibration formulae prepared in advance using ALB and CRE control samples with known concentrations. Finally, the ALB and CRE concentrations were displayed on the Raspberry Pi tablet on top of the detection box.
Notably, the use of a sliding chip design minimizes the effects of evaporation during the sample filtration and colorimetric reaction stages and therefore improves the RGB intensity value of the colorimetric images and enhances the detection performance as a result. Furthermore, in contrast to traditional benchtop analytic techniques, the proposed microchip and detection box enable the ALB and CRE concentrations of the urine sample to be determined simultaneously in a single detection process. As a result, the presented detection system has substantial potential for low-spending POCT applications in clinical CKD practice.
In the present study, the ALB and CRE concentrations are determined by the Raspberry Pi computer based on an inspection of the RGB intensity signal. However, as shown in Figure 1a, the microchip also incorporates two color bars adjacent to the reaction/detection zones to facilitate a qualitative determination of the ALB and CRE concentrations. It is thus anticipated that the sliding microchip can also be used as a crude standalone system for CKD diagnosis without the need for the full functionality of the Raspberry Pi-based detection system (e.g., a simple hotplate is sufficient).

Optimization of ALB and CRE Reaction Conditions
According to the findings of previous studies by the present group [24], the colorimetric reaction within the detection box was performed at the 40 • C for the waiting time of 3 min. Moreover, the G + B value intensity signal of the detection areas was found to be correlated mostly strongly with the ALB and CRE concentrations of the urine sample [35,36]. However, the RGB intensity of the reaction complexes also depends on the NaOH concentration (CRE reaction) and BCG concentration (ALB reaction). Accordingly, an experimental investigation was performed to determine the optimal NaOH and BCG concentrations using ALB and CRE control samples with known concentrations in the range of 10-300 mg/dL and 0.75-10 mg/dL, respectively.
In previous studies [24,35,36,68], the optimal reaction temperature for the detection of ALB and CRE by a paper-based device was mostly set to 37 • C. However, the current study used a slip-hybrid PMMA/paper microchip to integrate ALB and CRE detection on this microchip. Therefore, the optimal reaction temperature for this study was set at 40 • C to allow efficient and continuous heating from PMMA to paper-based components. In the optimal reaction time, the optimal reaction time was set to 3 min. For a more detailed description of the optimal reaction temperature and reaction time, please refer to our previous studies [24,35,36]. Figure 3 shows the G + B intensity value variation with the CRE concentration for four distinct values of the NaOH concentration (1-4 M) in the NaOH/picric acid reagent. For all values of the CRE concentration, the G + B intensity value reduces with a raising NaOH concentration, owing to the lower solubility of picric acid in more alkaline environments. Moreover, for each NaOH concentration, the G + B intensity value reduces approximately linearly with an increasing CRE concentration. However, regression analysis inspection results show that the reagent with a NaOH concentration of 3 M yields the highest correlation coefficient (R 2 = 0.9923) between the G + B intensity value and the CRE concentration. Thus, the concentration of NaOH was selected as 3 M in all of the remaining experiments.
Biosensors 2022, 12, x FOR PEER REVIEW 6 of 1 of 3 min. Moreover, the G + B value intensity signal of the detection areas was found to be correlated mostly strongly with the ALB and CRE concentrations of the urine sample [35,36]. However, the RGB intensity of the reaction complexes also depends on the NaOH concentration (CRE reaction) and BCG concentration (ALB reaction). Accordingly, an experimental investigation was performed to determine the optimal NaOH and BCG concentrations using ALB and CRE control samples with known concentrations in the range of 10-300 mg/dL and 0.75-10 mg/dL, respectively. In previous studies [24,35,36,68], the optimal reaction temperature for the detection of ALB and CRE by a paper-based device was mostly set to 37 °C. However, the curren study used a slip-hybrid PMMA/paper microchip to integrate ALB and CRE detection on this microchip. Therefore, the optimal reaction temperature for this study was set at 40 °C to allow efficient and continuous heating from PMMA to paper-based components. In the optimal reaction time, the optimal reaction time was set to 3 min. For a more detailed description of the optimal reaction temperature and reaction time, please refer to ou previous studies [24,35,36]. Figure 3 shows the G + B intensity value variation with the CRE concentration fo four distinct values of the NaOH concentration (1-4 M) in the NaOH/picric acid reagent For all values of the CRE concentration, the G + B intensity value reduces with a raising NaOH concentration, owing to the lower solubility of picric acid in more alkaline envi ronments. Moreover, for each NaOH concentration, the G + B intensity value reduce approximately linearly with an increasing CRE concentration. However, regression analysis inspection results show that the reagent with a NaOH concentration of 3 M yields the highest correlation coefficient (R 2 = 0.9923) between the G + B intensity value and the CRE concentration. Thus, the concentration of NaOH was selected as 3 M in all o the remaining experiments.

Calibration Equations for ALB and CRE Detection
The proposed assay platform was calibrated using ALB control samples with know concentrations of 0.75, 2, 4, 6, and 10 mg/dL, respectively, and CRE control samples wi concentrations of 10, 50, 100, 200, and 300 mg/dL, respectively. For all of the samples, t G + B intensity value were acquired under the optimal reaction conditions described the previous section. Figures 5 and 6 present the mean G + B intensity values obtained f the ALB and CRE control samples, respectively, over five repeated measurements in ea case. The regression analysis results in Figure 5 indicate that the G + B intensity value ( is related to the ALB concentration (X) as Y = −10.932X + 405.418 with a correlation coe ficient of R 2 = 0.9919. Similarly, in Figure 6, the G + B intensity is related to the CRE co centration as Y = −0.414X + 336.192 with a correlation coefficient of R 2 = 0.9923. In bo cases, the correlation coefficients have a high value close to 1. In other words, the ca bration curves provide a reliable means of predicting the ALB and CRE concentratio from the measured G + B intensity values.

Calibration Equations for ALB and CRE Detection
The proposed assay platform was calibrated using ALB control samples with known concentrations of 0.75, 2, 4, 6, and 10 mg/dL, respectively, and CRE control samples with concentrations of 10, 50, 100, 200, and 300 mg/dL, respectively. For all of the samples, the G + B intensity value were acquired under the optimal reaction conditions described in the previous section. Figures 5 and 6 present the mean G + B intensity values obtained for the ALB and CRE control samples, respectively, over five repeated measurements in each case. The regression analysis results in Figure 5 indicate that the G + B intensity value (Y) is related to the ALB concentration (X) as Y = −10.932X + 405.418 with a correlation coefficient of R 2 = 0.9919. Similarly, in Figure 6, the G + B intensity is related to the CRE concentration as Y = −0.414X + 336.192 with a correlation coefficient of R 2 = 0.9923. In both cases, the correlation coefficients have a high value close to 1. In other words, the calibration curves provide a reliable means of predicting the ALB and CRE concentrations from the measured G + B intensity values.

Calibration Equations for ALB and CRE Detection
The proposed assay platform was calibrated using ALB control samples with know concentrations of 0.75, 2, 4, 6, and 10 mg/dL, respectively, and CRE control samples wit concentrations of 10, 50, 100, 200, and 300 mg/dL, respectively. For all of the samples, th G + B intensity value were acquired under the optimal reaction conditions described i the previous section. Figures 5 and 6 present the mean G + B intensity values obtained fo the ALB and CRE control samples, respectively, over five repeated measurements in eac case. The regression analysis results in Figure 5 indicate that the G + B intensity value (Y is related to the ALB concentration (X) as Y = −10.932X + 405.418 with a correlation coe ficient of R 2 = 0.9919. Similarly, in Figure 6, the G + B intensity is related to the CRE con centration as Y = −0.414X + 336.192 with a correlation coefficient of R 2 = 0.9923. In bot cases, the correlation coefficients have a high value close to 1. In other words, the cal bration curves provide a reliable means of predicting the ALB and CRE concentration from the measured G + B intensity values.

Application of Proposed Assay Platform to Real-World Urine Samples
Human urine samples were collected from 23 adult CKD patient volunteer tional Cheng Kung University Hospital (NCKUH, Taiwan). For each sample, the provided the age and gender of the patient, the ALB and CRE concentrations, corresponding ACR value (see Table 1). Note that the ALB and CRE concentratio determined using a conventional method on a benchtop biochemistry analyze chi-7600, Hitachi High-Technologies Co., Minato, Japan). The ALB and CRE concentrations of each sample were determined using posed assay platform under the optimal reaction conditions described in Section the calibration equations presented in Section 3.2. The corresponding results ar in Figure 7a,b, where the Y-axis in each figure shows the benchmark results prov NCKUH and the X-axis indicates the measurement results obtained using the p paper-based platform. It can be seen that a good agreement exists between the tw measurements in both cases (i.e., R 2 = 0.9933 for ALB and R 2 = 0.9980 for CRE). T feasibility of the proposed platform for practical ALB and CRE determination tions is confirmed. Table 2 presents the comparison of several properties betw current platform detection system and the developed methods for ALB and CR

Application of Proposed Assay Platform to Real-World Urine Samples
Human urine samples were collected from 23 adult CKD patient volunteers at National Cheng Kung University Hospital (NCKUH, Taiwan). For each sample, the hospital provided the age and gender of the patient, the ALB and CRE concentrations, and the corresponding ACR value (see Table 1). Note that the ALB and CRE concentrations were determined using a conventional method on a benchtop biochemistry analyzer (Hitachi-7600, Hitachi High-Technologies Co., Minato, Japan). The ALB and CRE concentrations of each sample were determined using the proposed assay platform under the optimal reaction conditions described in Section 3.1 and the calibration equations presented in Section 3.2. The corresponding results are shown in Figure 7a,b, where the Y-axis in each figure shows the benchmark results provided by NCKUH and the X-axis indicates the measurement results obtained using the proposed paper-based platform. It can be seen that a good agreement exists between the two sets of measurements in both cases (i.e., R 2 = 0.9933 for ALB and R 2 = 0.9980 for CRE). Thus, the feasibility of the proposed platform for practical ALB and CRE determination applications is confirmed. Table 2 presents the comparison of several properties between the current platform detection system and the developed methods for ALB and CRE detection in urine samples.

Conclusions
This study has presented an effective and low-cost approach for determining the ALB and CRE concentrations in single human urine samples using a simple sliding microfluidic paper-based device and a hand-held detection system based on RGB color intensity analysis. Notably, the ALB and CRE concentrations can be detected through a single colorimetric reaction conducted at a low temperature of 40 °C for just 3 min. The experimental results have shown that, given an optimal composition of the ALB and CRE reagents, the G + B intensity value of the reaction compound varies linearly (R 2 = 0.9919)

Conclusions
This study has presented an effective and low-cost approach for determining the ALB and CRE concentrations in single human urine samples using a simple sliding microfluidic paper-based device and a hand-held detection system based on RGB color intensity analysis. Notably, the ALB and CRE concentrations can be detected through a single colorimetric reaction conducted at a low temperature of 40 • C for just 3 min. The experimental results have shown that, given an optimal composition of the ALB and CRE reagents, the G + B intensity value of the reaction compound varies linearly (R 2 = 0.9919) with the ALB concentration over the range of 0.75-10 mg/dL. Similarly, the G + B intensity varies linearly (R 2 = 0.9923) with the CRE concentration over the range of 10-300 mg/dL. Furthermore, the ALB and CRE concentration measurements obtained for the urine samples of 23 CKD patient volunteers are in excellent agreement (R 2 = 0.9933 and R 2 = 0.9980, respectively) with the benchmark values obtained using a conventional macroscale technique.
The results indicate that the proposed assay platform provides a feasible alternative to conventional benchtop techniques for determining the ALB and CRE concentrations in human urine samples and computing the ACR accordingly. As such, it provides a feasible solution for the low-cost, rapid, and straightforward detection of CKD in both clinical and POC contexts. In the present study, the ALB and CRE concentrations are derived from a computer-based analysis of the RGB intensity of the reaction compounds. However, the microchip also incorporates two color bars which enable a crude assessment of the ALB and CRE concentrations to be made through a simple naked-eye inspection. Thus, the potential of the proposed microchip for preliminary CKD diagnosis in non-clinical contexts or regions of the world with poorly-developed medical infrastructures is further enhanced.

Data Availability Statement:
The data presented in this study are available upon request from the corresponding author.

Conflicts of Interest:
The authors declare no conflict of interest.