Smartphone-controlled biosensor for viral respiratory infectious diseases: Screening and response

Abstract

Outbreaks of emerging viral respiratory infectious diseases (VRIDs) including coronavirus disease 2019 (COVID-19) seriously endanger people's health. However, the traditional nucleic acid detection required professionals and larger instruments and antigen-antibody detection suffered a long window period of target generation. To facilitate the VRIDs detection in time for common populations, a smartphone-controlled biosensor, which integrated sample preparation (electromembrane extraction), biomarker detection (red-green-blue model) and remote response technology (a built-in APP), was developed in this work. With the intelligent biosensor, VRIDs could be recognized in the early stage by using endogenous hydrogen sulfide as the biomarker. Importantly, it only took 15 min to accomplish the whole process of screening and response to VRIDs. Moreover, the experimental data showed that this smartphone-controlled biosensor was suitable for ordinary residents and could successfully differentiate non-communicable respiratory diseases from VRIDs. To the best of our knowledge, this is the first time that a smartphone-controlled biosensor for screening and response to VRIDs was reported. We believe that the present biosensor will help ordinary residents jointly deal with the challenges brought by COVID-19 or other VRIDs in the future.

Graphical abstract

1. Introduction

Viral respiratory infectious diseases (VRIDs) are a series of diseases caused by different respiratory viruses, normally including influenza (infected with influenza virus), acute upper respiratory tract infections (infected with rhinovirus), severe acute respiratory syndrome (SARS) and novel coronavirus pneumonia (COVID-19) (infected with coronavirus), etc. VRIDs often spread rapidly, seriously endangering people's health and bringing great economic losses [1]. For example, with the emergence of highly concealed and contagious variant strains of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), the coronavirus disease 2019 (COVID-19) has been threatening people's lives and the development of the social economy around the world [2]. From an epidemiological point of view, the risk of VRIDs can be reduced from three aspects: controlling the source of infection, cutting off the way of transmission, and protecting the susceptible population [3]. Timely and accurate diagnosis is the priority in the prevention and control of respiratory infectious diseases. Once a person was diagnosed or screened as positive for VRIDs, some necessary measures should be taken to stop further transmission, including quarantining at home, and seeking medical help. In this case, developing a point-of-care sensor that could screen VRIDs and instantly make responses (e.g., disinfection of the residence place and suggestions for personnel isolation) is urgently needed for common residents.

So far, there have been mainly two methods for the diagnosis of VRIDs [4]. The first one is the “gold standard” for the diagnosis: Nucleic acid testing (NAT). The implementation of NAT requires professionals and instruments, and the NAT process traditionally includes sample preparation, gene extraction, and polymerase chain reaction (PCR) [5]. Another method is antigen-antibody testing. At the early stage of virus infection (generally called the window period), there may be a few antigens and nearly no antibody generated in the infectious body. Moreover, usually antigens will change with the appearance of variant strains. Thus, this method has a limit on testing time and accuracy [6,7]. In short, neither of these methods is suitable for common residents to detect the VRIDs at any time or anywhere they need and disinfect the residence place after detection in time. To address the cumbersome detection, some point-of-care approaches for antigen detection have been reported [8], which facilitate the prevention of VRIDs. However, to the best of our knowledge, a point-of-care sensor for the earlier and more accurate screening of VRIDs and the simultaneous response remains a challenge in this field.

Nowadays, some paper-based or on-chip optical detection methods (colorimetry, fluorescence, etc.) have already been developed and used for the point-of-care testing of some chronic and bacterial infectious diseases. Generally, these methods have been achieved via detecting biomarkers or specific substances related to the disease [9,10]. VRIDs often cause lung injury. It has been proved in various experimental models that hydrogen sulfide has the potential to prevent acute lung injury in pulmonary inflammation [[11], [12], [13], [14]]. Recently, some studies have found that the availability of hydrogen sulfide would significantly change in COVID-19 patients [15,16]. Besides, relevant literature has proved that in the in-vitro models of respiratory syncytial virus (RSV) infection, the regulation of endogenous hydrogen sulfide significantly affected cell response and virus replication. At the same time, RSV infection affected the generation of hydrogen sulfide in airway epithelial cells [17]. As a regulatory factor in organisms, endogenous hydrogen sulfide plays an important role in the pathophysiological mechanism and has a variety of protective functions including anti-inflammatory, antioxidant stress, and antivirus [18,19]. Therefore, the concentration of hydrogen sulfide in the blood will also change in some other diseases, including cardiovascular diseases (coronary artery lesions and atherosclerosis, etc.) [20,21], chronic kidney diseases [22], and Alzheimer's disease [23]. As a potential biomarker, the preliminary diagnosis or screening of VRIDs could be achieved by detecting the level of endogenous hydrogen sulfide in plasma [15,16]. According to relevant literature, the concentration of sulfide in blood or plasma is in the range of 30–300 μmol L⁻¹ [24]. At present, the reported detection methods of endogenous hydrogen sulfide mainly include inductively coupled plasma mass spectrometers (ICP-MS) [25], methylene blue spectrophotometry (MBSP) [26], high-performance liquid chromatography (HPLC) coupled to a UV-VIS [27], etc. However, it is difficult for common residents to use these precise instruments to accomplish the rapid detection of endogenous hydrogen sulfide. How to achieve further response towards the detection results should be considered as well.

In the present age of digital information, smartphones have been inseparable from our daily life. Whether in terms of information processing (detection) or transmission (response), smartphones could finish it in the shortest time [28]. Therefore, a system based on smartphone could meet the needs of intelligent screening and response to VRIDs for common residents: i) Quantification of the endogenous hydrogen sulfide, a potential biomarker of VRIDs, can be realized by chromogenic reaction combined with the camera function. After photographing the chromogenic image, the color information is processed with the self-developed application (APP) installed on the smartphone using the red-green-blue (RGB) model [29]. By associating the values of three primary colors, the concentration of endogenous hydrogen sulfide is obtained. ii) Once the detection result shows a risk of the presence of VRIDs, the APP automatically starts the further response including suggestions for disinfection, personnel isolation, etc.

Because of its features including low cost and easy operation, the paper-based chip has been used widely in environmental pollutant analysis, clinical diagnosis, and other fields [30,31]. In this work, a paper-based portable platform was designed to achieve rapid detection. Endogenous hydrogen sulfide in plasma samples was selected as an indicator of VRIDs, but the matrices in plasma (e.g., colored red blood cells) would strongly affect the chromogenic reaction and thus the detection of endogenous sulfide. To eliminate the inference effect of matrix components, a sample pretreatment procedure was integrated into the paper-based portable platform. As a sample pretreatment method, electromembrane extraction (EME) is an electrokinetic migration process, which has the advantages of short extraction time, excellent enrichment, and purification ability [32]. At present, the paper-based EME has been applied for extraction of the copper ions from complicated matrices [33]. Therefore, EME shows great potential in solving the problem of matrix interference in color detection.

In short, a smartphone-controlled biosensor including a portable screening platform and a remote emergency response system was constructed to reduce the risk of VRIDs infection. The sensor consisted of a paper-based platform and a built-in response APP. After the screening of VRIDs, the APP will proceed with the risk assessment immediately according to the test result. When the biomarker level is abnormal, the response APP will suggest the subject go to a hospital for further diagnosis. At the same time, the APP can remotely switch on the disinfection equipment in advance at home or office to complete the disinfection (Fig. 1 ). We believe that the biosensor integrating sample preparation, biomarker detection and remote response technology will help ordinary residents to deal with the challenges brought by COVID-19 or other VRIDs in the future.

Fig. 1.

Schematic illustration of the smartphone-controlled screening and remote response system.

2. Experimental section

2.1. Reagents and materials

Hydrochloric acid (HCl), sodium hydroxide (NaOH), sodium nitroprusside (SNP), and sodium sulfide (Na₂S) were purchased from Sinopharm Chemical Reagent Co., Ltd. (Shanghai, China). 1-Heptanol, 1-octanol, 1-nonanol, 1-decanol, 1-undecanol, cetyltrimethylammonium bromide (CTAB), polyvinylpyrrolidone (PVP), and sodium dodecyl sulfate (SDS) were purchased from Aladdin Chemical Reagent Co. (Shanghai, China). Whatman 1 qualitative filter paper was purchased from Zhengcheng Experimental Instrument Co., Ltd. (Shanghai, China). The polypropylene (PP) membrane (approximately 100 μm in thickness) was purchased from Membrana (Wuppertal, Germany). The 9 V battery was purchased from Nanfu Battery Co., Ltd. (Fujian, China).

All blood and plasma samples were obtained from 50 healthy volunteers and collected in tubes containing EDTA to avoid clotting. For the detection method development, 25 plasma samples were mixed, and the other samples were used for method validation. All samples were preserved at 4 °C. All volunteers signed an informed consent form and agreed to publish the experimental data publicly. This study was approved by the Ethics Committee of Tongji Medical College, Huazhong University of Science and Technology on November 25, 2020 ([2020] S284).

2.2. Construction of the paper-based platform

The screening function of the sensor was based on the paper-based platform. The construction of the platform was given as follows:

Construction of the platform framework: To ensure the stability of the sensor during the detection process, the platform framework was independently designed by 3D printing with resin material. As shown in Fig. S1, the framework contained two shells that formed a whole cuboid with a size of 44 mm × 24 mm × 5 mm. In the middle of each shell, a rectangle window with a size of 10 mm × 6 mm was presented. For connecting the paper chip to the power supply, two channels (4 mm × 1 mm) were reserved on both sides of the shell.

Design of the paper chip: Hydrophilic paper (Whatman 1 qualitative filter paper) was used for the construction of the paper chip. Two paper layers with the shape of “±” were first cut, which contained four parts (Fig. 2 a). The head (part-I) on the paper was designed for fixing the electrode, and two horizontal bars (the part-II and part-IV) were designed to fix the chip into the framework. To confine the sample or acceptor solution within the fine area of the part-III with a size of 6 mm × 10 mm, the head, as well as two horizontal bars, were fabricated as hydrophobic areas by immersing in molten paraffin at 80 °C for 2 s. It is noted that, when constructing the hydrophobic area of the paper, a copper wire electrode with a length of 15 mm was also immersed in paraffin. The end of the wire was close to the middle of the hydrophobic head. After fabrication of the acceptor layer and donor layer, the supported liquid membrane (SLM) for EME was prepared. The size of the PP membrane was 12 mm × 14 mm (Fig. 2a), which was larger than the hydrophilic area (6 mm × 10 mm) to prevent liquid leakage. The membrane solvent in SLM was 5 μL of 1-octanol mixed with CTAB, which was cast at the center of the support membrane. The paper chip was then obtained by fixing the acceptor layer, donor layer, and SLM into the 3D-printing framework. The acceptor layer was on the top, the donor layer was on the bottom, and the SLM was in the middle of the two paper layers (Fig. 2b).

Fig. 2.

a) Schematic model of the paper-based platform; b) SEM image of the neat PP membrane; c) Schematic diagram of hydrogen sulfide movement; d) Linear relationships of r with hydrogen sulfide in water; e) Change of r with different concentrations of hydrogen sulfide in plasma; f) Effects of different solvents and surfactants on the target recoveries (conditions: 9 V, 12 min, pH value of the acceptor solution was 12, pH value of the donor solution was 5).

Connection to the power supply: The framework with the paper chip inside was placed on the power supply. The electrodes were connected to the power supply from the power supply holes. In this study, the acceptor layer was connected to the positive electrode, and the donor layer was connected to the negative electrode. In this way, the paper-based platform was successfully constructed.

2.3. Optimization of paper-based EME

To solve the problem of matrix interference in plasma, EME was used for the separation of targets. Before the optimization of EME, the endogenous hydrogen sulfide in the mixed plasma sample was detected with MBSP. The standard curve was shown in Fig. S2. Using this method, the hydrogen sulfide in the mixed plasma sample was measured to be 33.2 ± 1.2 μmol L⁻¹. In this mixed plasma sample, 30.0 μmol L⁻¹ S²⁻ was spiked (sodium sulfide as sulfur ion donor). This spiked plasma was then used for the optimization of paper-based EME. In the EME process (Fig. 2c), the acceptor solution was 20 μL of sodium hydroxide solution (pH = 12). The donor solution was a mixture of the 10 μL plasma sample and 10 μL hydrochloric acid (pH = 4.7). The pH value of the donor phase was around 5.0. After adding the sample and acceptor to the paper layer, the power supply (9 V) was switched on and the separation was conducted for 12 min. The concentration of hydrogen sulfide in the acceptor phase was detected with the spectrophotometry method. During the optimization, the method of the single-factor test was used to analyze the critical variables (EME voltage, pH of the acceptor, and extraction time). Only one parameter was changed as designed and the other parameters were kept constant. The spiked recovery was calculated for the evaluation of optimal parameters.

The spiked extraction recovery (SER) of the hydrogen sulfide in plasma was calculated by equation (1), and the recovery in this paper was SER:

$$SER=\frac{C_2-C_1}{C_3}\times 100\%$$ (1)

C₁ was the recovery concentration of hydrogen sulfide before standard addition. C ₂ was the recovery concentration of hydrogen sulfide after standard addition. C ₃ was the spiked concentration.

2.4. Rapid quantification of biomarkers

2.4.1. Optimization of color reaction

To realize the rapid detection of biomarkers without relying on large-scale instruments, the quantification of the target was mainly based on the color reaction in this work. The conditions for color reaction were also optimized, including the concentration of chromogenic reagent, reaction time, and pH of the reaction system. Typically, 20 μL of 40 mmol L⁻¹ chromogenic reagent was added directly to the acceptor solution on the hydrophilic area of the paper. After the desired reaction time (∼30 s), the color was observed visually and recorded by a smartphone.

2.4.2. Image capture and processing

In this study, the smartphone of Honor Magic 3 was used to capture the digital images after the color reaction. The attributes of the image from Honor Magic 3 were 3072 × 4096 pixels, aperture value f/1.9, exposure time 1/100 s and ISO-320, respectively. The smartphone lens was placed horizontally about 8 cm above the color rendering area when photographing. It is noted that the edge of the platform was exactly aligned with the edge of the phone screen at this distance. The images were saved in Joint Photographic Experts Group format (.jpg).

The images were processed with Adobe Photoshop CC 2018 software (PS). Briefly, the color region (6 mm × 10 mm) was selected by the tailoring tool, and the average RGB values in the region of interest (ROI) were recorded. Previous works had shown that the RGB measurements could be used for the quantitative determination of the target in samples [34]. To control the effects of the photographing conditions (e.g. the light intensity) for stable outcomes of picture processing, a quantitative RGB normalization method was developed [35]. Thus, the transformation of RGB to normalized RGB was conducted by the following equation (2) in this study:

$$r=\frac{R+B}{R+G+B}$$ (2)

This normalized RGB value (r) and hydrogen sulfide concentration showed a good linear relationship.

2.5. Analytical curve for construction of the biosensor

Before the APP design, analytical parameters should be obtained. First, a standard curve for testing the concentrations of hydrogen sulfide in plasma was established. Firstly, the endogenous hydrogen sulfide in the mixed plasma sample was detected as 33.2 ± 1.2 μmol L⁻¹ with the MBSP. The hydrogen sulfide concentrations in the plasma were then adjusted to10 μmol L⁻¹, 15 μmol L⁻¹, 37.5 μmol L⁻¹, 75 μmol L⁻¹, 150 μmol L⁻¹, 200 μmol L⁻¹, 250 μmol L⁻¹, and 300 μmol L⁻¹ by dilution with PBS buffer (pH = 7.4) or addition of sulfide. The hydrogen sulfide in plasma was separated by the paper-based platform, and then the color reaction was performed. r was then obtained after the capture of color reaction images with a smartphone and processing of the images with PS. The standard curve was established by plotting the r values versus the concentration of the hydrogen sulfide. In this way, the analytical parameters of the standard curve for APP design were obtained.

2.6. Design of the APP

APP was developed based on the Android Studio (Google) integrated development tool. The function of APP mainly included three parts: i) acquisition of images, ii) conversion of color signal into RGB information and then conversion of RGB information into the concentration of the biomarker, and iii) the feedback of the result (e.g. connection with the disinfection equipment).

The layout of the APP interface was mainly divided into two parts (the upper part was the image display area, and the lower part was the parameter display area and the function area). The photographing function was directly written into the APP program. Click the “Photograph” button to wake up the camera. In addition to taking photos, the local photos can also be imported into the APP for detection.

The RGB mode was used to process the color area in this study. ROI was set as the 3 mm × 3 mm area in the middle of the rendering area (6 mm × 10 mm) according to the optimization of the detection area. To select and calculate the ROI accurately, the length-to-width ratio of the photographing area (there has a border indication on the screen when photographing) was set to 3:5.5 according to the length-to-width ratio of the platform. During photographing, the long and short sides of the platform completely coincide with the photographing border. Lock the coordinates and calculate the average value of R, G and B (by the pixel value of each channel) in the center (the detection size was converted from the actual 3 mm × 3 mm area) of the image display area. The average value of R, G and B was brought into equation (2) to obtain the r value. Then the standard curve of the r value and hydrogen sulfide concentration was used to obtain the concentration of the target. To eliminate the occurrence of abnormal values, the range of the r value and the range of detectable concentration was set at 0.66–0.76 and 0–300 μmol L⁻¹ according to the standard curve, respectively. Once the measured value was out of this range, the screen would display that the detection value was invalid.

After the target concentration was obtained, the APP would further respond to the detection results. First, 30–300 μmol L⁻¹ was preset as the normal range in the program, and then two countermeasures were set after the comparison results appeared. If a value was within the normal range, it indicated that the infection risk would be low in this current and displayed a green prompt to end testing. On the contrary, it would prompt the warning interface and request to open the disinfection equipment. To realize the remote connection of the self-developed APP and equipment, the APP was able to connect to the “smart remote-control” software, which could jump to the control interface of the device. It should be noted that the disinfection equipment used in this study was equipped with intelligent WiFi cloud control technology, which could be directly controlled with the APP [36]. For traditional equipment requiring infrared control, it is necessary to purchase an intelligent infrared remote controller to connect with the APP to control the equipment.

2.7. Method validation

To verify the screening accuracy of this smartphone-controlled biosensor, 11 plasma samples were detected by the MBSP and this biosensor at the same time. Further, to verify that the response system of the smartphone can operate normally, an air purifier as disinfection equipment was chosen to connect with the APP in advance, and then three samples were tested for the remote control of the equipment.

To verify the practical feasibility of this biosensor for screening VRIDs, a simple verification test was designed. Four volunteers were selected for on-site detection of the endogenous hydrogen sulfide by the biosensor. At the same time, whether the volunteers would have respiratory symptoms within the next five days was observed. Moreover, five people (who had no experimental experience related to this experiment) were randomly selected to operate the biosensor to prove its serviceability for ordinary residents.

3. Results and discussion

3.1. Proof of concept

The importance of fabrication of a screening platform (Fig. 1) was confirmed by the detection of the biomarker in aqueous solution (Fig. 2d) and plasma (Fig. 2e). As seen, with the color reaction, the hydrogen sulfide concentration and r showed a good linear relationship in aqueous solution (R² ≥ 0.99), while this color reaction was strongly inhibited by the color of the plasma (resulting in a failure detection of the biomarker in plasma). These experimental data would provide the potential for biomarker detection in plasma if the matrix interference could be efficiently removed. Therefore, the fabrication of an intergraded platform including the separation and detection of the biomarker was one of the challenges in this work.

3.2. Parameters of paper-based platform

The platform was designed for common residents to screen the VRIDs. Paper was a kind of promising testing material for the construction of chips in the rapid detection of various targets. Herein, a new separation technology named paper-based EME was used. The papers with pore sizes of 0.7, 1.0, 1.2, 1.6, and 2.0 μm were selected for the fabrication of the paper-based platform. It was seen in Fig. S3 that the types of papers (with different pore sizes) showed no effect on the EME efficiency.

As a paper-based platform for ordinary residents to use, the sample volume required for detection should be controlled at a very low level. To find the most appropriate sample volume for the experimental effect, a sample (ferrous sulfate solution) was added in drops to the hydrophilic area (10 μL each time). It is seen in Fig. S4 that the liquid covered the whole color area and had a clear color rendering effect when the volume was 40 μL, so the total volume of acceptor solution and chromophore solution was chosen as 40 μL. When the volume ratio of the two solutions was 1:1 (each was 20 μL), it had a uniform color rendering effect. For the donor solution on the bottom of the paper layers, it was necessary to avoid liquid leakage. When the volume was less than 20 μL, the liquid only spread horizontally and without the vertical infiltrate. Therefore, to facilitate solution addition, the volume of three solutions (acceptor solution, chromophore solution, and donor solution) were all chosen at 20 μL. Besides, compared with lateral flow immunoassay to screen the VRIDs [37], the separation of samples on the paper-based platform did not need exclusive transport channels and thus avoided the residue of the samples.

3.3. Optimization of the biomarker separation

In this work, the separation of the biomarker was achieved by paper-based EME. The SLM is one of the important parameters that largely determines the selectivity, efficiency, and stability of the EME. Alcohols are effective membrane solvents to separate inorganic anions in EME [38]. However, our preliminary experiment showed that the pure alcohols failed to function as SLMs to separate the hydrogen sulfide from plasma (Fig. S5a). In kinds of literature, the surfactant was reported to successfully promote the migration of anions [39]. Accordingly, cationic surfactant CTAB, anionic surfactant SDS and non-ionic polymer surfactant PVP were added to the membrane solvent of alcohols (Fig. 2f). For all the alcohols, the presence of CTAB showed much higher extraction recovery of the target than the presence of SDS and PVP. The reason is given as follows: HS⁻ and S²⁻ are negatively charged, which is easier to form ion pairs with cation surfactant (CTAB). Moreover, the recovery of hydrogen sulfide raised with the increase of the carbon content of alcohols. With a further increase in the carbon content of alcohols, the recovery of hydrogen sulfide decreased. This might be due to the subsequent increase in the density and viscosity of alcohols, which had a resistance to the migration of the target analyte [40]. Therefore, 1-octanol mixed with CTAB was selected as the membrane solvent for the paper-based EME. The concentration of CTAB in the 1-octanol was also optimized (Fig. S5a). The recovery of the target was almost 0 when no surfactant was used in 1-octanol. With the increase of CTAB, the recovery of the target was enhanced. When the CTAB concentration was higher than 12.5 mg mL⁻¹, the recovery of the target reached a plateau (∼60%). For membrane solvent, the organic solvent should be used as little as possible under the guarantee of ideal recovery, so 5 μL of membrane solvent was added every time according to the experimental results (Fig. S5b).

As the driving force for the transfer of charged ions, the voltage has a great influence on the extraction of the target during the EME process [41]. To meet the requirement of portable detection for ordinary residents, a low and safe voltage was preferred, which could be provided by the commercial battery. As seen in Fig. 3 a, the recovery of the target increased with the increase of the voltage and reached a plateau until the voltage was higher than 9 V. So the battery of 9 V was selected as the power supply of the platform.

Fig. 3.

a) Effect of EME voltage on target recovery (12 min, pH value of the acceptor solution was 12, pH value of the donor solution was 5); b) Effect of pH value of donor solutions on target recovery (9 V, 12 min, pH value of the acceptor solution was 12); c) Effect of extraction time on target recovery (9 V, pH value of the acceptor solution was 12, pH value of the donor solution was 5); d) Color changes in different pH values of chromogen and sulfur ion; e) Color changes between different concentrations of chromogen and sulfur ion; f) Relationship between the time of color reaction and r. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

For transferring the biomarker from the donor phase to the acceptor phase driven by the electric force, the biomarker should be ionized in both two phases [42]. Moreover, once the pH value of human plasma was higher than the physiological pH (7.4), sulfur-containing compounds such as cysteine and glutathione would be decomposed. The generation of hydrogen sulfide during the decomposition would affect detection results. Due to the above two criteria, the donor solution needed to be adjusted to neutral or weakly acidic [43]. The pH effect of samples on the recovery was investigated, it could be seen in Fig. 3b that the recovery of hydrogen sulfide obviously changed with altering the pH of the sample. When the pH value of the sample was near 5, the EME system showed the highest recovery of 60%. Thus, the pH of the plasma was adjusted to 5 with the hydrogen hydrochloric acid solution before the EME process. Besides, the pH of the acceptor solution was adjusted to 12 (conducive to color reaction) to ensure that the sulfur ion would not escape in the form of gas (hydrogen sulfide).

The extraction duration is also an important parameter for EME. Generally, the recovery increases with the increase of the extraction duration, and then reaches a plateau period in a typical EME process [44]. For the present paper-based EME system, the recovery of hydrogen sulfide also showed a similar tendency. Briefly, the recovery gradually increased during the initial extraction duration of 12 min. After 12 min, the recovery reached the plateau (Fig. 3c). Based on this experimental result, 12 min was chosen as the extraction duration. Moreover, EME extraction time accounts for most of the detection time of the biosensor, so this biosensor can complete a round of the screening and response within 15 min (containing the time of other processes, like color reaction and smartphone information processing).

3.4. Quantification of the biomarker

To obtain a sensitive and stable color reaction for biomarker detection, the conditions for color reaction should be optimized. The principle of the color reaction was the chemical reaction between sulfur ion and sodium nitroprusside which resulted in a clear purplish red. Among all the parameters that would affect the quantification of the endogenous hydrogen sulfide in this reaction, the pH value of the reaction system was the most important factor [45]. The influence of the pH value of the solution on the color reaction was shown in Fig. 3d. When the pH value was lower than 12, the reaction color was not obvious (compared to that with a pH value of 12). When the pH value was higher than 12, the reaction color displayed a dark yellow. This dark yellow was not selected for the quantitation, because it originated from the chromogen in the highly alkaline environment. Therefore, the pH value of the solution was selected to be 12 for the color reaction.

Besides, the concentration of chromogen was another important factor affecting the detection. Generally, the chromogen should be excessive compared to the targets in the samples. According to relevant literature, the concentration of hydrogen sulfide in blood after infection with VRIDs was not higher than 400 μmol L⁻¹ [15]. The concentration of chromogen was then optimized with different concentrations of hydrogen sulfide. As seen in Fig. 3e, with the same hydrogen sulfide concentration, the reaction system tended to show the color of chromogen decomposition in the highly alkaline condition when increasing the chromogen concentration. This tendency was much clear when the hydrogen sulfide concentration was ample lower. There was an obvious color change gradient when the concentration of chromogen was 40 mM, and this concentration was then selected as the optimized chromogen concentration in this study.

To verify the stability of the color reaction, the reaction time was investigated here. As shown in Fig. 3f, the color reaction occurred quickly and the reaction equilibrium was achieved in a short time. Moreover, the color was relatively stable in 1 min, so it was important to capture the images within 1 min. In the present study, a reaction duration of 30 s was selected.

3.5. Design of the APP

In this work, the sensor APP was developed based on the Android Studio (Google) integrated development tool. To reduce the impact of the difference in the photographing acquisition on the detection, the long-short sides of the platform completely coincided with the photographing border, and the phone lens was placed horizontally above the device, and the phone was held upright in hand (Fig. 4 ). The height between the lens and the color rendering area was selected at about 8 cm (the whole equipment was just photographed). The distance should not be too far to increase the impact of the surrounding environment, or too close to make it difficult for the lens to focus.

Fig. 4.

Schematic illustration of the smartphone-controlled biosensor for VRIDs screening and its response to result (combination with disinfection equipment).

For the rendering area of detection by APP, it should stably reflect the overall color level. The whole rendering area was 6 mm × 10 mm, it was easily affected by the edge blank area when the whole rendering area was calculated. Besides, there also was an error caused by uneven color when selecting a point for detection. So, it is necessary to optimize the area of calculation. Three different concentrations of hydrogen sulfide (low, medium, and high concentration) reacted with chromogen on the paper chip (all conditions were the same with the reaction of the acceptor and the chromogen), and then three rendering areas were obtained. After that, the four corners and centers of rendering areas were detected by selecting the point and the area of 1 mm × 1 mm, 2 mm × 2 mm, 3 mm × 3 mm, and 4 mm × 4 mm, respectively. As shown in Table S1, the standard deviation was the largest by clicking the point, and the area of 3 mm × 3 mm can stably reflect the overall color level. In short, if the selected detection area was small, it would increase the error by uneven color. On the contrary, the area at the edge of the rendering area would have an impact if the detection area was large. So, the ROI was set as 3 mm × 3 mm in the middle of the rendering area (6 mm × 10 mm).

3.6. Method validation

3.6.1. Detection method of the biomarker

After the successful design of the biosensor, the screening system was utilized to measure the concentration of endogenous hydrogen sulfide in plasma. By spiking or diluting a mixed plasma (the endogenous hydrogen sulfide in mixed plasma was 33.2 ± 1.2 μmol L⁻¹ detected with the MBSP), plasma samples with different concentrations of the hydrogen sulfide (1–300 μmol L⁻¹) were prepared. Using these samples, the feasibility of the present biosensor was evaluated. As shown in Fig. 5 a, the biosensor showed a linear detection range of 10–300 μmol L⁻¹ with a good linear correlation coefficient (R² = 0.9951). In this work, the limit of detection (LOD) was calculated as LOD = 3 λ_b/slope, and the limit of quantification (LOQ) was calculated as LOQ = 10 λ_b/slope, where λ_b was the standard deviation of very low concentration samples. The LOD and LOQ were calculated to be 2.5 μmol L⁻¹ and 8.3 μmol L⁻¹, respectively. To compare the present detection method with the conventional approaches for endogenous hydrogen sulfide measurement, the MBSP method was also investigated here. As seen in Fig. S2, the LOD for the MBSP method was 1.2 μmol L⁻¹, indicating that our system had a detection sensitivity comparable to the general MBSP method. More importantly, in the early stage of VRIDs infection, the level of the endogenous hydrogen sulfide (biomarker) in plasma would be significantly reduced (less than 30 μmol L⁻¹), therefore the detection method by the present biosensor met the demand of VRIDs screening.

Fig. 5.

a) Standard curve of r and hydrogen sulfide in plasma; b) Comparation of biosensor detection and MBSP method using 11 samples; c) Application of the biosensor for plasma from a non-communicable respiratory diseases patient; d) Application of the biosensor for plasma from a VRIDs patient.

3.6.2. Selectivity of the detection method

Some other anions also co-existed in the plasma, which would affect the detection of the biomarker (hydrogen sulfide). To confirm the high selectivity of this smartphone-controlled biosensor, the common anions (HCO₃ ⁻, SO₄ ²⁻ and SCN⁻) were added to the plasma. As known, the HCO₃ ⁻, SO₄ ²⁻, and SCN⁻ in normal human circulation were approximately 24 mmol L⁻¹, 0.3 mmol L⁻¹, and 30 μmol L⁻¹, respectively [[46], [47], [48]]. The concentration gradient (an extra spiking) of three anions was set according to their normal content in plasma (the highest concentration was 5 folds to the normal level). After the addition of the interfering anions, the biomarker in the plasmas was detected by the smartphone-controlled biosensor. As shown in Fig. S6, compared to the detection result (35.9 ± 1.8 μmol L⁻¹) before the extra spiking of the interferences, the concentrations of hydrogen sulfide detected by the biosensor remained constant (33.6 ± 2.3–37.4 ± 2.2 μmol L⁻¹) even when the inferences were 5 times higher to their normal ranges. These results showed that anions in plasma did not affect the separation and detection of the hydrogen sulfide by this biosensor. The high detection selectivity of the biosensor provided the screening of biomarkers in plasma with high applicability.

3.6.3. Comparison with the conventional approach

Traditional detection methods of endogenous hydrogen sulfide primarily relied on large and expensive instruments (such as ICP-MS or UV spectrophotometer), and the pretreatment of biological samples before detection was very tedious (such as the method of MBSP). To verify the accuracy and reliability of the quantitative system of the biosensor in this study, the detection results of the endogenous hydrogen sulfide by our method were compared with the detection results by MBSP. Here, a total of 11 plasma samples were evaluated. As shown in Fig. 5b, there was a strong positive correlation between our method and MBSP, with a slope of 1.04 ± 0.01 and a correlation coefficient of 0.998, suggesting that the results from the two methods matched within the experimental error. These results strongly indicated that the accuracy of our method was as good as MBSP, and the inexpensive, rapid, portable, and user-friendly detection method developed in this study was suitable and reliable.

3.6.4. Evaluation of the response system

In this study, the response system mainly relied on the APP of the smartphone. To verify that the response system developed can operate normally, an air purifier as disinfection equipment was chosen to connect with the APP in advance. Three samples were prepared for testing: the first sample was provided by a health volunteer, the endogenous hydrogen sulfide in this plasma was 37.2 ± 2.5 μmol L⁻¹ detected by MBSP; the second sample was obtained from the volunteer who had non-communicable respiratory disease with some symptoms (sneezing), and the endogenous hydrogen sulfide in this plasma was 50.7 ± 3.5 μmol L⁻¹ detected by MBSP; the third sample was a VRIDs patient sample, in which the endogenous hydrogen sulfide was 13.5 ± 2.1 μmol L⁻¹ detected by MBSP. The biosensor was used to detect the hydrogen sulfide in these three samples, and the detection results were then transferred to APP (instructing the air purifier to be switched on or not). As shown in Table S2, for the first sample, the concentration of the hydrogen sulfide was detected to be 39.4 ± 2.4 μmol L⁻¹ (within the normal range), and the air purifier was therefore not switched on according to the instructions of APP. For the volunteer who had a non-communicable respiratory disease, the concentration of the hydrogen sulfide in his plasma was detected to be 52.6 ± 2.9 μmol L⁻¹, the air purifier was therefore not switched on either (Fig. 5c). For the third sample, the concentration of the hydrogen sulfide was detected to be 14.3 ± 1.8 μmol L⁻¹ (among the risk range of VRIDs), and the air purifier was therefore switched on (Fig. 5d). These results indicated that the response system developed could operate rationally according to the concentrations of endogenous hydrogen sulfide. In addition, the hydrogen sulfide concentration of the non-communicable respiratory diseases sample was within the normal range, so the biosensor had good specificity for the screening of VRIDs. It is noted that except for the air purifier, the APP designed in this work could also realize the remote control of the other equipment. With the era of medical practice supported by mobile phones coming [49], the detection and response systems controlled by the smartphone could continue to be upgraded to make greater contributions to disease prevention and health maintenance.

3.6.5. Practical application in the screening of VRIDs

Generally, there is an incubation period in which the body has no clinical symptoms during the infection of diseases. For example, the incubation period of COVID-19 is generally 3–7 days [50]. Obviously, early diagnosis of the disease can significantly inhibit the spread of the virus. Biomarkers can be observed in the early stage of disease, which can be used for virus screening [51]. To verify that the present biosensor could be used for screening of VRIDs, four volunteers (two men and two women) were selected to test the applicability of the biosensors. First, the on-site detections of the endogenous hydrogen sulfide were accomplished by the biosensor. Within the next five days, the volunteers were continuously observed whether they had respiratory symptoms. If related symptoms were monitored, the volunteer went to the hospital for respiratory disease examinations immediately. All volunteers carried out nucleic acid testing for COVID-19 on the first and fourth days. As illustrated in Table 1 , among the four volunteers, only one volunteer showed a low hydrogen sulfide concentration (16.2 ± 3.3 μmol L⁻¹, less than 30 μmol L⁻¹). This volunteer developed mild respiratory infection symptoms (cough and headache) a day later and then had been diagnosed with influenza by etiological examination. This result of the hydrogen sulfide reduction was consistent with the decrease in patients with early COVID-19 [16]. The remaining three volunteers did not have any adverse respiratory symptoms during the five-day observation period, and had no respiratory infectious diseases after hospital examination. For two times nucleic acid testing of COVID-19, all results of volunteers were negative. Therefore, from the testing results of the four volunteers, the biosensor can be used to screen VRIDs, though it still needs a lot of data to obtain a more accurate reference value range.

Table 1.

Screening results of four volunteers using the developed biosensor.

Volunteer	Biomarker (μM)	Risk level	COVID-19 testing		Other VRIDs examination	Equipment connection
			1	2
1	34.7 ± 2.4	Low	-a	-	-	No
2	39.8 ± 2.6	Low	-	-	-	No
3	46.5 ± 2.8	Low	-	-	-	No
4	16.2 ± 3.3	High	-	-	Influenza	Yes

^a“-” indicates that the result of VRIDs testing is negative.

3.6.6. Facile operation by ordinary residents

Moreover, five ordinary residents (who had no experience related to this experiment) were randomly selected to prove the serviceability of the biosensor. The detection of three samples from each person was performed according to the experimental protocols. As a control, we also conducted the detection in a normal way. All detection results were summarized in Table S3. It is seen that the detection results from the ordinary residents and the professionals were matched within the experimental errors, indicating that the smartphone-controlled biosensor for screening and response to VRIDs provided great serviceability for ordinary residents. It is emphasized here that the concentration of hydrogen sulfide in blood will also change in other diseases [21,22]. Therefore, the screening method developed in this study is only suitable for ordinary residents without basic diseases, and is not applicable to people with serious cardiovascular disease, chronic kidney disease, Alzheimer's disease, etc.

4. Conclusion

The spread of VRIDs makes the shortage of medical resources more prominent in a short time. One of the great difficulties is to complete the VRIDs detection in time for common populations. Therefore, it is a vital trend to develop low-cost portable screening platforms with easy operation and remote emergency response systems for reducing the risk of infection. In this work, to overcome the limitations of nucleic acid detection (requiring professionals and larger instruments) and antigen-antibody detection (low sensitivity and the window period), a smartphone-controlled biosensor was developed for screening and response to VRIDs. The biosensor developed for screening and making the response to VRIDs only needed 15 min for one test, making this smartphone-controlled biosensor suitable for ordinary residents. Using this biosensor, non-communicable respiratory diseases could be successfully differentiated from VRIDs. To the best of our knowledge, this is the first time that a smartphone-controlled biosensor integrating sample preparation, biomarker detection and remote response technology for VRIDs was reported. However, there are still some shortcomings remained in the present work: i) more experimental data are needed to upgrade the range of hydrogen sulfide for the early detection of VRIDs. This related work is undergoing in our lab with the help of Tongji Hospital affiliated with Tongji Medical College, Huazhong University of Science & Technology; ii) More positive samples (including samples from the COVID-19 patients) are needed to further test the applicability of the biosensor. This will be done with the help of our foreign collaborators. Whatever, we believe that the present biosensor will facilitate ordinary residents to better deal with the challenges brought by COVID-19 or other VRIDs in the future, and it is foreseen that this work can bring a new idea to the initial prevention and instant control of VRIDs.

Credit authors statement

Yaxing Ma: Conceptualization, Methodology, Formal analysis, Validation, Visualization, Writing – original draft, Writing – review & editing. Yaoyu Luo: Conceptualization, Methodology, Investigation, Formal analysis, Validation, Visualization, Writing – review & editing. Xinrui Feng: Conceptualization, Methodology, Validation, Writing – review & editing. Chuixiu Huang: Resources, Supervision, Writing – review & editing. Xiantao Shen: Supervision, Funding acquisition, Writing – review & editing.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix A. Supplementary data

The following is the Supplementary data to this article:

Data availability

Data will be made available on request.