Fluorescence-Based Portable Assays for Detection of Biological and Chemical Analytes

Fluorescence-based detection techniques are part of an ever-expanding field and are widely used in biomedical and environmental research as a biosensing tool. These techniques have high sensitivity, selectivity, and a short response time, making them a valuable tool for developing bio-chemical assays. The endpoint of these assays is defined by changes in fluorescence signal, in terms of its intensity, lifetime, and/or shift in spectrum, which is monitored using readout devices such as microscopes, fluorometers, and cytometers. However, these devices are often bulky, expensive, and require supervision to operate, which makes them inaccessible in resource-limited settings. To address these issues, significant effort has been directed towards integrating fluorescence-based assays into miniature platforms based on papers, hydrogels, and microfluidic devices, and to couple these assays with portable readout devices like smartphones and wearable optical sensors, thereby enabling point-of-care detection of bio-chemical analytes. This review highlights some of the recently developed portable fluorescence-based assays by discussing the design of fluorescent sensor molecules, their sensing strategy, and the fabrication of point-of-care devices.


Introduction
Fluorescence-based assays are among the most widely applied sensing techniques, and they have been used for a variety of applications, including biomedical research, environmental monitoring, and in the food industry, among others [1][2][3][4][5][6]. The popularity of fluorescence assays is due to their high sensitivity, specificity, fast operation, and the availability of diverse types of fluorophores that absorb and emit light covering a broad spectrum of wavelengths from ultraviolet to infrared. There are a variety of fluorescence assays-which include immunoassays, foster resonance energy transfer (FRET), fluorescence correlation spectroscopy (FCS), and polarization-that are used for sensing applications [7][8][9][10][11][12][13]. Most of these assays involve a target recognition element, commonly an antibody or a protein or chemical receptor attached to a fluorophore or fluorescence sensing agent (small molecule or nanoparticles), that bind to the specific analyte [6,14]. Upon binding to the analyte, the fluorophore often undergoes photophysical changes, resulting in variation in the fluorescence signal or the upper state lifetime, or a shift in emission wavelength, which can be detected using readout devices such as fluorometers, microscopes, or cytometers, equipped with photomultiplier tubes or CCD/sCMOS cameras [15][16][17][18][19]. These assays are highly sensitive and efficient in detecting biological and chemical analytes, but most of them are lab-based techniques that require a trained operator, dedicated workspace, involve complex steps and are time intensive. Additionally, fluorescence readout devices are bulky and expensive. Therefore, most of these assays have limited applicability in resource limited settings, as well as in the field or point-of-care (POC).
To address this issue, a significant emphasis has been placed on designing and developing low-cost, easy-to-use, portable fluorescence-based technologies, with the aim and Au NCs, which were separated by hydrophobic barrier created using an ink printing method ( Figure 1). A serum sample with or without bacterial contamination was taken on a clean glass substrate and pressed on the paper sensor for interaction. The response of the sensors to each specified bacteria was monitored at an optimum pH close to serum, using a smartphone-based portable fluorometer. In the fluorometer, the paper assay device (PAD) was placed in a sample holder and irradiated with a UV light at 365 nm, as shown in Figure 1. The cabinet was equipped with a smartphone to record the fluorescence intensity change. Analyses of "before" and "after" images of each detection zone were performed using ImageJ software. The CuNCs under UV light showed a blue emission, and AuNCs showed a red fluorescence emission. After interaction with the bacterial contaminated sample, the NCs underwent fluorescence quenching. This was due to the aggregation of the NCs caused by the functional groups present on the protein shell of the NCs and the active sites of the peptidoglycan in the cell wall of the bacteria, which subsequently caused dynamic or static quenching of the NCs. Each of the CuNCs and AuNCs have a different affinity towards different functional moiety, further distinguishing between "Gram positive" (thick peptidoglycan cell wall, less lipopolysaccharide layer) and "Gram negative" bacteria (thin peptidoglycan layer, thick lipopolysaccharide layer). The limit of detection using this fluorometric PAD device was <50 Cfu/mL for S. aureus, S. pyogenes, E. coli, and P. aeruginosa (RSD < 5%). They further tested a spiked serum sample and reported a recovery percentage of 93-107%.
Sensors 2023, 23, x FOR PEER REVIEW 3 of 22 possible to monitor and differentiate pathogenic bacterial species in a complex mixture. The sensors were based on copper (CuNCs) and gold nanoclusters (AuNCs) stabilized with proteins-pepsin, trypsin, ovalbumin, and glutathione. The detection assay, composed of six rectangular detection zones made of paper substrate (1.5 cm × 1.0 cm), was impregnated with Cu and Au NCs, which were separated by hydrophobic barrier created using an ink printing method ( Figure 1). A serum sample with or without bacterial contamination was taken on a clean glass substrate and pressed on the paper sensor for interaction. The response of the sensors to each specified bacteria was monitored at an optimum pH close to serum, using a smartphone-based portable fluorometer. In the fluorometer, the paper assay device (PAD) was placed in a sample holder and irradiated with a UV light at 365 nm, as shown in Figure 1. The cabinet was equipped with a smartphone to record the fluorescence intensity change. Analyses of "before" and "after" images of each detection zone were performed using ImageJ software. The CuNCs under UV light showed a blue emission, and AuNCs showed a red fluorescence emission. After interaction with the bacterial contaminated sample, the NCs underwent fluorescence quenching. This was due to the aggregation of the NCs caused by the functional groups present on the protein shell of the NCs and the active sites of the peptidoglycan in the cell wall of the bacteria, which subsequently caused dynamic or static quenching of the NCs. Each of the CuNCs and AuNCs have a different affinity towards different functional moiety, further distinguishing between "Gram positive" (thick peptidoglycan cell wall, less lipopolysaccharide layer) and "Gram negative" bacteria (thin peptidoglycan layer, thick lipopolysaccharide layer). The limit of detection using this fluorometric PAD device was <50 Cfu/mL for S. aureus, S. pyogenes, E. coli, and P. aeruginosa (RSD < 5%). They further tested a spiked serum sample and reported a recovery percentage of 93-107%. Stepwise illustration of a paper-based assay system for sensing bacteria using a smartphone as a readout device for point-of-care application [54]. Stepwise illustration of a paper-based assay system for sensing bacteria using a smartphone as a readout device for point-of-care application [54].
In another study, Xie et al., designed a point-of-care test for Mycobacterium tuberculosis (Mtb), using a naturally secreted enzyme by tubercle bacilli, BlaC, as a marker and BlaCspecific fluorogenic substrate, a chemically modified cephalosporin-based fluorescence molecule, as probe [55]. The detection mechanism was based on the enzymatic hydrolysis of the fluorogenic substrate by the BlaC enzyme releasing the fluorophore, thus enhancing the fluorescence signal intensity. A green-fluorescent probe-Tokyo green-was modified and coupled with cephalosporin to develop the highly stable CDG-OMe, fluorogenic substrate. In the presence of BlaC, the enzyme reacted selectively with the substrate, and there was a significant increase, more than~200-fold, in the fluorescence intensity at 520 nm. For the practical point-of-care application, they further fabricated a simple device integrated with an LED light source, excitation and emission filters, and a smartphone. Images captured with a smartphone through the box hole made it possible to detect~10 c.f.u live Mtb in the human sputum in less than 10 min. Therefore, this rapid TB diagnostic tool can be used for the sensitive and selective detection of low quantities of Mycobacterium present in the human sputum and other bodily fluids, without the need for prior sample preparation and additional complicated steps.

Detection of Anions and Cations
Metal ions are essential for maintaining a healthy life, as they are responsible for the regulation of cell-to-cell interaction; the proper functioning of nerve cells, the brain, and heart; muscle cells; DNA regulation; transporting oxygen; maintaining osmotic pressure; and many other biological processes [56,57]. Change in their concentrations can cause severe malfunctions in the body, e.g., growth disorders, carcinogenesis, or even death [57]. Therefore, the precise quantification of anions and cations serves as a major tool for understanding any underlying disease condition, and for maintaining a person's overall well-being. For example, fluoride ion is an essential element that provides strength to bones and teeth, preventing dental cavities, osteofluorosis, etc. [58,59]. Common detection techniques include ion chromatography, spectroscopy, and ion-selective electrodes, all of which require complicated equipment and sample pre-treatment, making the detection process exhaustive [60,61]. Fluorescent probes, on the other hand, provide sensitive simple detection approach [62][63][64][65][66][67][68][69][70][71][72]. For "on-site" detection, Yu et al. fabricated a multicolor ratiometric fluorescent test paper for the point-of-care detection of fluoride ions (Fions) in water using filter paper [73]. The test paper (3.5 × 1 cm 2 ) was prepared by an ink-printing method using "ink" made by mixing a fluoride-sensitive organic probe (C-TIPS) and Fion inert red CdTe quantum dots in optimal proportions. The fluorescent sensor showed a fluorescence "turn-on" effect upon binding to fluoride ions. The test paper exhibited a distinguishable fluorescence color change from red to purple to blue under a UV lamp. The probe C-TIPS was fabricated using 7-hydroxycoumarin, and the blue fluorescence of the compounds was chemically quenched by covalently conjugating it with triisopropylsilyl chloride (TIPS). Upon binding to the Fion, the recovery of the highly bright blue fluorescence suggested the removal of the TIPS moiety from the compound by cleaving the Si-O bond, thus releasing very bright blue, fluorescent coumarin in the system. The overlapping broad absorbance and excitation wavelength of the CdTe QDs, 7-hydroxycoumarin, and C-TIPS allowed the ratiometric analysis using same excitation wavelength at 365 nm. This sensor showed high sensitivity and selectivity toward Fions. In the presence of Fions, the emission peak of the sensor, at 455 nm, gradually increased, which was proportional to the concentration of Fions, while the emission peak 630 nm remained unchanged. The limit of detection was calculated to be~0.285 µM. The practicality of the sensor was tested using lake water, tap water, well water, and human urine spiked with Fions, and showed the excellent potential of the sensor to detect Fions in a real sample, with a recovery percentage of 96.5%. The change in fluorescence color could be distinguished by the naked eye, thus allowing on-site visual assay of Fions in environmental water and biological fluids without any pre-treatment.
In another study, Zhang et al. established a point-of-care sweat diagnostics technique for the detection of chloride ions. They achieved this by designing a low-cost citratederived fluorescence sensor-based chloridometer, operated using a smartphone [74]. Citric acid (CA)-modified cysteine acted as a potential fluorescent chloride sensor to measure the chloride level present in sweat, which is an essential marker for cystic fibrosis diagnosis [74]. Apart from its cost effectiveness and high chloride selectivity, CA-cysteine was also selected as a fluorescent sensor for its exceptional photostability, high quantum yield (81%), and longer lifetime (10 ns) compared to other chloride sensors. An ultraviolet LED at 365 nm was used to excite CA-cysteine, which generated a sharp blue fluorescence signal with an emission maximum of 441 nm. The fluorescent signal was subsequently captured and measured by the smartphone camera, as shown in Figure 2. In the presence of chloride ions, the blue fluorescence signal of the CA-cysteine solution underwent fluorescence quenching, which can be attributed to the non-radiative relaxation of the excited fluorophore due to the presence of chloride ions. The detection limit was reported to be~0.8 mM. Furthermore, successful clinical validation of the device was performed using sweat from both healthy subjects and cystic fibrosis patients.
In another study, Zhang et al. established a point-of-care sweat diagnostics techniqu for the detection of chloride ions. They achieved this by designing a low-cost citrate-de rived fluorescence sensor-based chloridometer, operated using a smartphone [74]. Citri acid (CA)-modified cysteine acted as a potential fluorescent chloride sensor to measur the chloride level present in sweat, which is an essential marker for cystic fibrosis diagno sis [74]. Apart from its cost effectiveness and high chloride selectivity, CA-cysteine wa also selected as a fluorescent sensor for its exceptional photostability, high quantum yield (81%), and longer lifetime (10 ns) compared to other chloride sensors. An ultraviolet LED at 365 nm was used to excite CA-cysteine, which generated a sharp blue fluorescence sig nal with an emission maximum of 441 nm. The fluorescent signal was subsequently cap tured and measured by the smartphone camera, as shown in Figure 2. In the presence o chloride ions, the blue fluorescence signal of the CA-cysteine solution underwent fluores cence quenching, which can be attributed to the non-radiative relaxation of the excited fluorophore due to the presence of chloride ions. The detection limit was reported to b ~0.8 mM. Furthermore, successful clinical validation of the device was performed using sweat from both healthy subjects and cystic fibrosis patients. Chromium (Cr 3+ ) is another important cation in the body, and its imbalance (exces or deficiency) has been correlated to diabetes, kidney toxicity, and cardiovascular disease [75]. Zhang et al., fabricated a hydrogel digital assay based on metal-AIEgen framework (MAFs) for the ultrasensitive detection of Cr 3+ ions in water, food, and biological fluids by successfully integrating the hydrogel assay with a smartphone readout device [75]. They developed a white emissive MAFs@QDs-PVP hydrogel sensor complex by incorporating blue emissive MAFs (Ex./Em ~310 nm/465 nm) with high quantum yield ~99% and red emissive QDs (Em. 604 nm) into a polyvinylpyrrolidone (PVP) hydrogel matrix, and th combined fluorescent sensor was freeze dried to form a sponge-like hydrogel sensor. Th hydrogel swells once the analyte solution comes into contact with it. The analytical per formance of the sensor was tested using Cr 3+ ions, and blue fluorescence-quenching of th MAFs was observed almost instantaneously. The limit of detection was calculated to b ~0.1 nM. In the presence of Cr 3+ ions, the color of the white emissive sensor transitioned to red fluorescence, due to the quenching of blue emissive MAFs. Cr 3+ ions selectively bind to the surface of the MAFs, thereby quenching the blue fluorescence of the hydrogel sen sor due to the energy transfer between the Cr 3+ ions and the MAFs, owing to thei Chromium (Cr 3+ ) is another important cation in the body, and its imbalance (excess or deficiency) has been correlated to diabetes, kidney toxicity, and cardiovascular diseases [75]. Zhang et al., fabricated a hydrogel digital assay based on metal-AIEgen frameworks (MAFs) for the ultrasensitive detection of Cr 3+ ions in water, food, and biological fluids by successfully integrating the hydrogel assay with a smartphone readout device [75]. They developed a white emissive MAFs@QDs-PVP hydrogel sensor complex by incorporating blue emissive MAFs (Ex./Em~310 nm/465 nm) with high quantum yield~99% and red emissive QDs (Em. 604 nm) into a polyvinylpyrrolidone (PVP) hydrogel matrix, and the combined fluorescent sensor was freeze dried to form a sponge-like hydrogel sensor. The hydrogel swells once the analyte solution comes into contact with it. The analytical performance of the sensor was tested using Cr 3+ ions, and blue fluorescence-quenching of the MAFs was observed almost instantaneously. The limit of detection was calculated to bẽ 0.1 nM. In the presence of Cr 3+ ions, the color of the white emissive sensor transitioned to red fluorescence, due to the quenching of blue emissive MAFs. Cr 3+ ions selectively bind to the surface of the MAFs, thereby quenching the blue fluorescence of the hydrogel sensor due to the energy transfer between the Cr 3+ ions and the MAFs, owing to their overlapping absorption and fluorescence emission wavelength. Furthermore, the sensor was tested with biological fluids such as saliva, urine, and serum, and a recovery percentage of~103-110% was observed in spiked samples. For the point-of-care application, the images of the hydrogel sensor before and after Cr 3+ ion interaction were captured using a smartphone and quantitatively analyzed using RGB color space. Additionally, they also demonstrated a paper-based lateral flow fluorescence immunoassay (LFIA), using MAFs for the effective detection of alpha-fetoprotein (AFP), a serum biomarker for hepatocellular carcinoma. The LFIA comprised of AFP-antibody1 (Ab1)-conjugated MAFs, a secondary Ab (Ab2) for capturing AFP, and an anti-Ab on a paper platform. Once the AFP was captured and deposited on the test line, a bright fluorescent light was observed under UV light. The estimated LOD of 0.6 pg/mL was achieved with this method. The reliability of the LFIA was evaluated using AFP-spiked human serum and with patient urine samples. The sensitivity of this MAFs LFIA was 100-1000 fold higher than traditional QDs-based LFIAs [75].

Detection of Biochemical Analytes
Biochemical compounds are the building blocks of life. For example, carbohydrates, lipids, proteins, nucleic acids, and enzymes are responsible for different functions such as energy storage, acting as messengers, helping in the maintenance of cell structure, and carrying out important chemical reactions, amongst others [76,77]. An imbalance of these biomolecules can cause several life-threatening diseases, such as kidney dysfunction, jaundice, sickle cell anemia, cardiovascular diseases, hormonal imbalance, diabetes etc. [78,79]. Therefore, the precise monitoring of biochemical analytes is crucial for disease management and to maintain a healthy life. Guo et al., developed a dual emissive ratiometric fluorescent sensor molecule integrated with a smartphone for the accurate sensing of glucose (Glu) and cholesterol (Chol) in patients with metabolic syndrome [80]. The amount of Glu and Cho was determined by the quantification of H 2 O 2 , which is the principal product produced in the presence of glucose oxidase (GOx) and cholesterol oxidase (ChOx). The dual emissive sensor molecule was composed of AgNPs/UiO-66-NH 2 , where silver nanoparticles (Ag-NPs) acted as sensing molecules. The AgNPs were absorbed on a metal organic framework (MOF) UiO-66-NH 2 , with sodium borohydride and OPD (o-phenylenediamine) acting as a chromogenic substrate. In the presence of GOx and ChOx, the analyte glucose and cholesterol are converted to H 2 O 2 which, in turn, etches AgNPs to release Ag + ions, and further oxidizes OPD to fluorescent 2,3-diaminophenazine (DAP) in situ. Due to the overlapping emission profile of AgNPs/UiO-66-NH 2 (Ex. 360 nm, Em. 425 nm) and the absorbance of DAP (emission 555 nm), the fluorescence intensity ratio of F 555nm /F 425nm showed gradual enhancement, which was proportional to the target analyte, inducing inner-filter effectbased ratiometric sensing. The sensor showed excellent sensitivity and selectivity toward H 2 O 2 , with a detection limit of 0.2 µmol/L. As both GOx and ChOx could convert O 2 to H 2 O 2 , the H 2 O 2 dependent signal was translated to quantify the amount of glucose and cholesterol in the human serum samples. The quantified Glu and Cho were determined to be~0.5 and~0.7 µmol/L, respectively. Additionally, the fluorescence color changes from blue to yellow green are easily distinguishable by the naked eye as well. For on-site application, the sensor AgNPs/UiO-66-NH 2 , along with OPD, was immobilized on a paper substrate, which was integrated with a 3D printed smartphone attachment device, where the test strip was placed in a holder and excited by an in-built UV lamp at 365 nm, as shown in Figure 3. The captured images were processed in the RGB color space with a smartphone app. The recovery rates were found to be~90% to 101%, using a spiked human serum sample, justifying the potential of these test strips for the quantitative analysis of glucose (Glu) and cholesterol (Chol) present in the human serum sample without any pre-treatment.  In another study, Wang et al. developed a hybrid ratiometric fluorescence sensor molecule based on an aggregation induced emission (AIE) of hyperbranched polymer nanoaggregate of tetraphenylethylene (HPA-TPE) and Rhodamine B (RhB) for the selective and sensitive testing of free bilirubin in water and urine samples [81]. In patients with liver dysfunction, free bilirubin tends to get accumulated in the body, causing gallbladder disease, anemia, and neurotoxicity [82,83]. The HPA-TPENA/RhB hybrid fluorescence sensor molecule was fabricated by dissolving and mixing HPA-TPE polymer nanoaggregate and Rhodamine B individually in phosphate buffer (pH = 7.4). When the system was irradiated with UV-light at 355 nm, both RhB and HPA-TPENA were excited, thereby producing dual emission peaks at 577 nm and 465 nm [81]. HPA-TPENA/RhB together showed weaker emission intensity at 465 nm, compared to HPA-TPENA alone. This is due to the overlap between emissions of HPA-TPENA (465 nm) and the broad absorption peak of RhB at ~400-550 nm, causing an energy transfer between them. The addition of bilirubin exhibited a linear reduction of the fluorescence intensity ratio I465/I577 with the increment of bilirubin concentration. The blue emission, with a peak at 465 nm, was significantly suppressed due to the Förster resonance energy transfer (FRET) quenching between the bilirubin and polymer nanoaggregate, whereas the Rhodamine B emission peaked at 577 nm and did not show any sensitivity to bilirubin. This eventually shifted the fluorescence peak from blue to orange, which was observable. This mechanism resulted in a detection limit of ~25 nM in the solution phase, with recovery values in the range of ~92.5-103.3%, and relative standard deviations being less than 6.1%. They further built a portable and affordable point-of-care bilirubin detection device by incorporating xanthum gum hydrogel with this hybrid fluorescence sensor, coupled with a handheld UV light and smartphone for capturing images, as shown in Figure 4. The hydrogel showed a bluewhite emission, which gradually changed to a yellow color upon increasing the concentration of bilirubin. This device was also successfully tested on human urine for quantifying bilirubin. In another study, Wang et al. developed a hybrid ratiometric fluorescence sensor molecule based on an aggregation induced emission (AIE) of hyperbranched polymer nanoaggregate of tetraphenylethylene (HPA-TPE) and Rhodamine B (RhB) for the selective and sensitive testing of free bilirubin in water and urine samples [81]. In patients with liver dysfunction, free bilirubin tends to get accumulated in the body, causing gallbladder disease, anemia, and neurotoxicity [82,83]. The HPA-TPE NA /RhB hybrid fluorescence sensor molecule was fabricated by dissolving and mixing HPA-TPE polymer nanoaggregate and Rhodamine B individually in phosphate buffer (pH = 7.4). When the system was irradiated with UV-light at 355 nm, both RhB and HPA-TPE NA were excited, thereby producing dual emission peaks at 577 nm and 465 nm [81]. HPA-TPE NA /RhB together showed weaker emission intensity at 465 nm, compared to HPA-TPE NA alone. This is due to the overlap between emissions of HPA-TPE NA (465 nm) and the broad absorption peak of RhB at~400-550 nm, causing an energy transfer between them. The addition of bilirubin exhibited a linear reduction of the fluorescence intensity ratio I 465 /I 577 with the increment of bilirubin concentration. The blue emission, with a peak at 465 nm, was significantly suppressed due to the Förster resonance energy transfer (FRET) quenching between the bilirubin and polymer nanoaggregate, whereas the Rhodamine B emission peaked at 577 nm and did not show any sensitivity to bilirubin. This eventually shifted the fluorescence peak from blue to orange, which was observable. This mechanism resulted in a detection limit of~25 nM in the solution phase, with recovery values in the range of 92.5-103.3%, and relative standard deviations being less than 6.1%. They further built a portable and affordable point-of-care bilirubin detection device by incorporating xanthum gum hydrogel with this hybrid fluorescence sensor, coupled with a handheld UV light and smartphone for capturing images, as shown in Figure 4. The hydrogel showed a blue-white emission, which gradually changed to a yellow color upon increasing the concentration of bilirubin. This device was also successfully tested on human urine for quantifying bilirubin.  Enzymes are essential elements for metabolic digestion in humans. They also serve as a specific biomarker, for example, of elevated levels of trypsin or amylase, which are associated with patients suffering from pancreatitis. The normal range of trypsin in human serum is 115-350 ng/mL, an excess of which can lead to pancreatic cancer and cystic fibrosis [84][85][86]. Li et al. developed a fluorescence nano-sensing platform for the detection of trypsin (TRY) by utilizing glutathione-capped gold nanoclusters (AuNCs) [87]. Glutathione (GSH) was used as a template and stabilizer during the AuNC synthesis process. AuNCs typically have an excitation peak at 418 nm, with an emission peak observed at 625 nm. The fluorescence intensity of the AuNC was quenched by conjugating positively charged cytochrome C (Cyt C) through an electron transfer (ET) process. The catalyzing effect of TRY hydrolyzed the Cyt C, causing it to break down into negatively charged heme-peptide particles, thus weakening the ET transfer, which subsequently induced the fluorescence recovery of AuNCs. The sensor reportedly has a detection limit of ~0.08 μg mL −1 in a phosphate buffer solution. The practical implementation of the AuNCs-Cyt Cbased platform was successfully demonstrated by analyzing TRY in spiked human serum and urine samples, with a reported recovery range between 95% and 109%, with RSD less than 4.06%. Furthermore, paper-based test strips were developed by immobilizing the AuNCs-Cyt C on absorbent papers for point-of-care detection application. The fluorescent images were captured using a smartphone and split into RGB values using ImageJ software. With the help of an image processing algorithm, a relationship between the RGB values and the logarithmic TRY concentration was established, in order to quantify the concentration.
Recently, a novel ultrasensitive fluorescence-based bioanalytical platform for the detection of multiple different biomarkers was developed [88][89][90]. In this study the authors explored nanostructures comprised of silica nanoparticles immobilized with natural NAD + -dependent or chimeric PQQ-dependent enzymes. The nanoconjugate was spotted on a fiberglass support. PQQ and NAD + -dependent enzymes are important in many cellular processes, and their activity can be used as a biomarker for various diseases. The substrate was designed to produce a fluorescent signal upon reaction with PQQ or NAD +dependent enzymes via a substrate-stimulated enzymatic reduction of a fluorescence probe phenazine methosulfate or its derivatives (Ex. 365 nm, Em 465 nm). In the presence of the analyte substrate, the immobilized enzymes reacted with it, changing the color of the spot attributed to the reduced form of the dye. Images of the colored spot were Enzymes are essential elements for metabolic digestion in humans. They also serve as a specific biomarker, for example, of elevated levels of trypsin or amylase, which are associated with patients suffering from pancreatitis. The normal range of trypsin in human serum is 115-350 ng/mL, an excess of which can lead to pancreatic cancer and cystic fibrosis [84][85][86]. Li et al. developed a fluorescence nano-sensing platform for the detection of trypsin (TRY) by utilizing glutathione-capped gold nanoclusters (AuNCs) [87]. Glutathione (GSH) was used as a template and stabilizer during the AuNC synthesis process. AuNCs typically have an excitation peak at 418 nm, with an emission peak observed at 625 nm. The fluorescence intensity of the AuNC was quenched by conjugating positively charged cytochrome C (Cyt C) through an electron transfer (ET) process. The catalyzing effect of TRY hydrolyzed the Cyt C, causing it to break down into negatively charged heme-peptide particles, thus weakening the ET transfer, which subsequently induced the fluorescence recovery of AuNCs. The sensor reportedly has a detection limit of~0.08 µg mL −1 in a phosphate buffer solution. The practical implementation of the AuNCs-Cyt C-based platform was successfully demonstrated by analyzing TRY in spiked human serum and urine samples, with a reported recovery range between 95% and 109%, with RSD less than 4.06%. Furthermore, paper-based test strips were developed by immobilizing the AuNCs-Cyt C on absorbent papers for point-of-care detection application. The fluorescent images were captured using a smartphone and split into RGB values using ImageJ software. With the help of an image processing algorithm, a relationship between the RGB values and the logarithmic TRY concentration was established, in order to quantify the concentration.
Recently, a novel ultrasensitive fluorescence-based bioanalytical platform for the detection of multiple different biomarkers was developed [88][89][90]. In this study the authors explored nanostructures comprised of silica nanoparticles immobilized with natural NAD +dependent or chimeric PQQ-dependent enzymes. The nanoconjugate was spotted on a fiberglass support. PQQ and NAD + -dependent enzymes are important in many cellular processes, and their activity can be used as a biomarker for various diseases. The substrate was designed to produce a fluorescent signal upon reaction with PQQ or NAD + -dependent enzymes via a substrate-stimulated enzymatic reduction of a fluorescence probe phenazine methosulfate or its derivatives (Ex. 365 nm, Em 465 nm). In the presence of the analyte substrate, the immobilized enzymes reacted with it, changing the color of the spot attributed to the reduced form of the dye. Images of the colored spot were captured using a smartphone camera in grayscale and processed via ImageJ software. The sensitivity was enhanced Sensors 2023, 23, 5053 9 of 22 by using the multilayer formation of enzyme-functionalized on the silica nanoparticles, resulting in the amplification of the fluorescence intensity of the reduced dye in the sensor spot. The sensor platform was successfully tested on different biological fluids, such as serum, urine, saliva, and sweat, for the detection of different biomarkers, such as glucose and alcohol (LOD 10 µM), α-amylase (LOD 2 pM), and some immunosuppressive drugs (cyclosporin A and tacrolimus with LOD 2 pM).
Albumin is a protein abundantly present in the human body. Abnormal levels of albumin, especially in urine, was observed in patients with kidney disease, indicating acute kidney failure, polycystic kidney disease, vascular disease, and neoplasia, and it therefore serves as an essential biomarker [91,92]. Most of the fluorescent probes are incapable of detecting urinary albumin with precise accuracy, as urine itself displays strong fluorescence, causing significant interference with high background noise [93,94]. Zhao et al. developed a novel olefin-extended chalcone fluorescent probe DNC, with an enlarged transition dipole moment (∆µ) and an emission wavelength >600 nm, for the point-of-care detection of albuminuria in urine [95]. The red emissive fluorescent probe, DNC, far from the excitation and emission range of the urine, helped in avoiding the spectral interference from urine fluorescence. Upon excitation at 480 nm, the fluorescence from urine was negligible at 610 nm, thereby ensuring a low background signal compared to the DNC fluorescence emission in response to albumin. This eventually provided an improved and amplified signal response against albumin. The fluorescence of DNC was significantly enhanced due to binding with the specific sites (drug sites) of albumin, and it demonstrated a highly sensitive and selective determination of albumin with a detection limit of 23 nM in solution. The sensor was tested using a urine sample spiked with albumin, and it was able to detect albumin concentration of~61 nM with RSD < 12%. An easy-to-use, smartphone-based POC system was developed for testing clinical samples collected from the patient. The mobile phone was used to capture images of the sensor, DNC, which was then analyzed using ImageJ software. The negative samples exhibited a blue color emission, while the sample spiked with albumin gradually changed from blue to a blue-violet to carmine color, which was easily distinguished by the naked eye. These results showed that the DNC probe can be successfully used to diagnose clinical A2-level and A3-level albuminuria in urine samples without pre-treatment.

Devices for Environmental Monitoring
Environmental monitoring is mainly aimed at air quality, water quality, and agriculture, with the ultimate goal of improving human health. The purpose of environmental monitoring is to identify potentially harmful microorganisms, chemicals, and other contaminants that can directly enter and pollute the ecosystem through food, air, or water [96]. For example, every year, approximately 48 million people in the US alone fall sick due to foodborne illness, among which~3000 people die. Similarly, it has been estimated that about 35 million die due to contaminated drinking water globally, including 7.15 million from related illnesses in the US, with children being most vulnerable. Therefore, there has been a drive to developing portable testing devices for the cost-effective, real-time, on-site detection of both organic and inorganic contaminants, as well as micro-organisms, in air, water, and solid samples. This will hasten the process of detection, allowing authorities to intervene quickly in the rapid elimination of contaminants from the environment [97][98][99]. Here, we highlight some of the applications related to the detection of organic compounds, as well as parasites in drinking water.

Detection of Volatile Organic Compounds (VOCs)
VOCs such as ketones, alkenes, esters, alcohols, and benzene derivatives are quite prevalent in nature, and their presence in the environment has been shown to increase health risks [34,100].  [101]. A film was developed by soaking the fluorescent sensor in a solid substrate such as cellulose paper, which displayed a "turn-on" emission upon exposure to different VOCs, such as ethyl acetate, toluene, and dichloromethane. Initially, the luminogen exhibited weak emissions due to AIE in the solid substrate, which is highly susceptible to VOCs. Upon exposure of the sensor to different VOCs, it diffuses into the molecular packing, causing intramolecular motion, exhibiting changes in their emissions via ICT. The mechanism established was the change in the disordered molecular arrangement of the sensor to a VOC-driven optimized conformation in the gas phase. The sensor exhibited either an increase or shift in emissions around~490-523 nm at an excitation wavelength of 370 nm, depending on the polarity of the solvent. The fluorescent response to each VOC was imaged and processed using RGB and lab color space to quantify the level of VOCs. The limit of detection (LOD) of the HNPMO sensor in the presence of ethyl acetate was observed to be~7.04 mg/m 3 (1.8 ppm). A wearable sensor was further fabricated by drop-casting or soaking the sensor HNPMO, and an inert reference luminogen on a paper-substrate that could be easily incorporated into labor suits, gloves, and even nails, etc., as shown in Figure 5. The inert luminogen exhibited an orange emission, while the HNPMO exhibited a bright green emission upon VOC exposure for 2 min. This color emission can be easily visualized by the naked eye, and therefore it can be used as predictive tool for early warning of VOC contamination in air.
HNPMO, [(E)-(3-(((2-hydroxynaphthalen-1-yl)methylene)amino) phenyl)(phenyl) anone]-based wearable fluorescent sensor for the detection of volatile organic comp (VOC), applying both aggregation-induced emission (AIE) and intramolecular c transfer (ICT) strategies [101]. A film was developed by soaking the fluorescent sen a solid substrate such as cellulose paper, which displayed a "turn-on" emission up posure to different VOCs, such as ethyl acetate, toluene, and dichloromethane. In the luminogen exhibited weak emissions due to AIE in the solid substrate, which is susceptible to VOCs. Upon exposure of the sensor to different VOCs, it diffuses in molecular packing, causing intramolecular motion, exhibiting changes in their emi via ICT. The mechanism established was the change in the disordered molecular ar ment of the sensor to a VOC-driven optimized conformation in the gas phase. The exhibited either an increase or shift in emissions around ~490-523 nm at an exc wavelength of 370 nm, depending on the polarity of the solvent. The fluorescent res to each VOC was imaged and processed using RGB and lab color space to quant level of VOCs. The limit of detection (LOD) of the HNPMO sensor in the presence o acetate was observed to be ~7.04 mg/m 3 (1.8 ppm). A wearable sensor was further cated by drop-casting or soaking the sensor HNPMO, and an inert reference lumi on a paper-substrate that could be easily incorporated into labor suits, gloves, and nails, etc., as shown in Figure 5. The inert luminogen exhibited an orange emission, the HNPMO exhibited a bright green emission upon VOC exposure for 2 min. This emission can be easily visualized by the naked eye, and therefore it can be used as p tive tool for early warning of VOC contamination in air.  In another study, Hyungi et al. developed a multiplex analysis-based chemical sensing system, consisting of a indolizine core chemical structure called Kaleidolizine (KIz), from which a library of 75 various fluorescent KIz compounds were designed and synthesized [102]. The KIz system has been proven to possess tunable, sensitive, and predictable photophysical properties in a solid state, and it was used to build a fluorescent array on cellulose paper printed by wax. These KIz derived compounds could be excited by a single excitation wavelength of 365 nm, and lights of varying colors and intensities covering the full spectral range of visible wavelengths (416-620 nm) were emitted. Upon interaction with different chemical analytes, each Klz compound exhibited significant changes in their emission profile, influenced by the ICT process, in their molecular environment, showing intensity variations, bathochromic shifts etc. This eventually contributed to representing a unique ID for different individual analytes. A smartphone-based sensing system was developed with a phone and LED to illuminate the sensor array, which captured and processed the images using a lab-designed automated color extraction application, including various computer vision algorithms. A random forest machine-learning algorithm was trained and implemented to analyze the pattern of color changes and differentiate the detected chemicals in the sensing arrays. As a proof of concept, the system was able to detect and discriminate 35 different volatile organic compounds (VOCs) with a detection accuracy of more than 97%.

Water Quality Monitoring
Despite several efforts over the last few decades, more than two billion people lack access to safe drinking water. In US alone, there are 2 million people who do not have access to clean water, which makes them prone to waterborne diseases caused by bacteria, viruses, and parasites, such as giardiasis, cholera, dysentery, hepatitis etc. [103][104][105]. Koydemir et al. reported a field portable and cost-effective rapid detection platform for Giardia lamblia cyst, a waterborne parasite [106]. The battery-operated microscope platform consisted of a smartphone coupled with a 3D-printed opto-mechanical attachment, including a custom designed disposable cassette that holds up to~20 mL of water. The disposable sample cassette was composed of absorbent pads, two different membrane filters (one with 8 µm pores to prevent the back flow of the particles, and one with 5 µm pores to trap the parasites). Prior to analysis, the water sample spiked with Giardia cyst was incubated with an anti-Giardia fluorescein-labelled stain and was dispensed on the capture membrane. After trapping the parasites, the cassette was then attached to the back of the fluorescent microscope, and the captured cysts in the filter membrane were then illuminated using eight blue LEDs. The images were then captured and transmitted to a server for image analysis, using a smartphone app. Machine learning was used for rapid and automatic counting, as well as differentiating of the Giardia cysts from other unwanted auto-fluorescence microobjects. The limit of detection was~12 cysts per 10 mL of water detected within 2 min after imaging with~84% cyst-counting sensitivity. The overall turnaround time was 1 h. The group further developed a fluorescence-based portable device for the detection of microplastics [35], E. coli [107], live bacteria, and microorganisms in water [36,[108][109][110][111].
Groundwater pollution by toxic chemicals such as antibiotic is another major problem posing a serious threat to human health [112,113]. Tetracycline is one such antibiotic overly used by humans, which is hard to degrade in living organisms and is converted into toxic compounds, causing serious groundwater pollution [114,115]. Zhang et al., constructed a ratiometric fluorescent sensor by combining molybdenum disulphide quantum dots (MoS 2 QDs) and europium ions (Eu 3+ ) to visually detect antibiotic tetracycline (TC) residues in the environment, with the help of a portable smartphone-based detection system [116]. The MoS 2 QDs exhibited an emission peak at 470 nm, while Eu 3+ showed a peak at 620 nm, upon excitation at 400 nm. In the presence of TC, the Eu 3+ ion bound to TC and formed Eu-TC, with no significant shift in the emission peak, but exhibiting weak fluorescence at 620 nm through the antenna effect. However, when MoS 2 was added, the fluorescence emission at 470 nm decreased with a significant enhancement of the fluorescence emission peak at 620 nm, due to the fluorescence resonance energy transfer between MoS 2 and Eu 3+ ion through the TC linear molecules, which served as a transport channel between them. Thus, MoS 2 QDs demonstrated their potential to serve as an indicator, as well as an enhancer, in the MoS 2 -Eu 3+ ratiometric fluorescent probe, to acquire dual signal output for TC detection. By taking the fluorescence intensity ratio at 620 nm and 470 nm (F 620 /F 470 ), highly sensitive detection of TC was performed successfully in water and biological fluids with a detection limit (LOD) of 2 nM. For validation, the system was tested with a tap water sample and mouse serum spiked with TC and a recovery of 94.4-108.4% was obtained with an RSD of less than 5.36%. The transition of fluorescence color change from blue to red can be observed by the naked eye, which was translated in the development of paper-based sensing strips for "on-site" sensing. A portable readout device was developed, which was composed of a dark box, miniature UV flashlight, LED lamp, movable multifunctional filter board, and a smartphone for imaging the paper strips immobilized with a fluorescence sensor. The quantitative analysis of TC was executed by the RGB color recognition software installed in a smartphone, and a detection limit of 0.05 µM with reproducibility within a 1-40 µM TC range was exhibited.
While water quality monitoring is important, water itself can act as an impurity in specific cases. Examples include chemical production, food processing, biomedicine, and the liquor industry, where water can lead to the hydrolysis of organic reagents, thereby degrading its quality [117][118][119][120]. Therefore, it is mandatory to monitor the water content for these applications. For this purpose, Liang et al. constructed a multicolor fluorescent sensor solution (G/R-CDs) by mixing green carbon dots (G-CDs) and red carbon dots (R-CDs) [121]. They used a solvothermal method and a hydrothermal method to synthesize R-CDs and G-CDs, respectively. Under the excitation wavelength of 390 nm, the R-CDs fluorescence spectrum exhibited an emission peak at 683 nm, with a quantum yield of 2.7%, where the fluorescence intensity gradually decreased with the increment of water content via an aggregation-caused quenching (ACQ) effect. However, the emission peak of the G-CDs experienced a red shift from 500 nm to 535 nm with a 7.9% quantum yield due to the solvent effect, while the overall fluorescence intensity remained unaffected with the addition of water content. As a result, G-CDs were used as the reference signal and R-CDs as the response. A linear relationship was established between the fluorescence intensity ratio of F 535 /F 683 and the water content in two different ranges of 3-30% and 40-60%, and a detection limit of 1.2% was obtained. In this system, G-CDs served as toners to enhance the visualization effect of G/R-CDs by complementing their fluorescence color with an exhibition of different colors, such as orange, pink, green, and grey, with individual water contents. The color variation was easily distinguishable with the naked eye; however, to eliminate human error, a smartphone was used to capture images and split the images into RGB values of the fluorescence color [121], as shown in Figure 6. This method was successful in detecting water in commercial liquor, medical alcohol etc., with an RSD value of less than 5%, and it thus manifested its potency to serve as a visual portable detector in the field of analytical chemistry.
x FOR PEER REVIEW 13 of 22

Typical POC Fluorescence Readout Devices
Fluorescence readout devices are a crucial aspect of POC assays, as they enable quantitative analysis of the biochemical analyte of interest. Particular emphasis needs to be placed on device design, in order to make them lightweight, compact, and low-cost. The readout devices generally consist of light sources to excite the sample; optical bandpass filters to spectrally separate the excitation and emission wavelengths; lenses for focusing the exciting light and for collecting the fluorescence photons; and, finally, a detector to record the fluorescence signal [22]. Of the many types of light sources available, diode lasers are more commonly used in POC devices. Other types of light sources that have been explored include flashlights and high-power LED lights [122]. Depending on the applications, it may be necessary to include multiple light sources for multi-spectral imaging. Incorporating multiple light sources can provide flexibility and versatility for measuring different types of fluorescent molecules, which is needed for multiplexed or ratiometric assays [123]. Sometimes heat sinks are also used to control the temperature of the light sources, in order to maintain stable performance and minimize fluctuations in output power. For detecting fluorescent photons, most of the POC devices use complementary metal oxide semiconductors (CMOS) cameras, such as smartphones, photodiodes etc. Other detectors, such as low-cost charge-coupled devices (CCDs), photodiodes (PDs), and avalanche photodiodes (APDs), have also been explored [22,[124][125][126]. Advanced detectors, such as superconducting nanowire single-photon detectors, which have extremely high sensitivity and low noise levels, can be explored for even more precise and sensitive fluorescence detection [127]. Additionally, modified CMOS sensors for imaging at near infra-red wavelengths can also be beneficial, as they can avoid issues related to background fluorescence, including autofluorescence [128]. All these components need to be housed together in a compact lightproof box, which can be easily constructed using 3D printing. With the use of computer-aided design (CAD) software, the 3D model of the device can be created and then printed cheaply. Furthermore, 3D printing can also be used to create a variety of components, such as the sample holder, microfluidic channels, and the holder of optical components, as the 3D printer can be customized to specific requirements [122,129]. The design of the sample holder is important in order to facilitate easy integration of the assay platform, e.g., the microfluidic channels, to enable precise control

Typical POC Fluorescence Readout Devices
Fluorescence readout devices are a crucial aspect of POC assays, as they enable quantitative analysis of the biochemical analyte of interest. Particular emphasis needs to be placed on device design, in order to make them lightweight, compact, and low-cost. The readout devices generally consist of light sources to excite the sample; optical bandpass filters to spectrally separate the excitation and emission wavelengths; lenses for focusing the exciting light and for collecting the fluorescence photons; and, finally, a detector to record the fluorescence signal [22]. Of the many types of light sources available, diode lasers are more commonly used in POC devices. Other types of light sources that have been explored include flashlights and high-power LED lights [122]. Depending on the applications, it may be necessary to include multiple light sources for multi-spectral imaging. Incorporating multiple light sources can provide flexibility and versatility for measuring different types of fluorescent molecules, which is needed for multiplexed or ratiometric assays [123]. Sometimes heat sinks are also used to control the temperature of the light sources, in order to maintain stable performance and minimize fluctuations in output power. For detecting fluorescent photons, most of the POC devices use complementary metal oxide semiconductors (CMOS) cameras, such as smartphones, photodiodes etc. Other detectors, such as low-cost charge-coupled devices (CCDs), photodiodes (PDs), and avalanche photodiodes (APDs), have also been explored [22,[124][125][126]. Advanced detectors, such as superconducting nanowire single-photon detectors, which have extremely high sensitivity and low noise levels, can be explored for even more precise and sensitive fluorescence detection [127]. Additionally, modified CMOS sensors for imaging at near infra-red wavelengths can also be beneficial, as they can avoid issues related to background fluorescence, including autofluorescence [128]. All these components need to be housed together in a compact lightproof box, which can be easily constructed using 3D printing. With the use of computer-aided design (CAD) software, the 3D model of the device can be created and then printed cheaply. Furthermore, 3D printing can also be used to create a variety of components, such as the sample holder, microfluidic channels, and the holder of optical components, as the 3D printer can be customized to specific requirements [122,129]. The design of the sample holder is important in order to facilitate easy integration of the assay platform, e.g., the microfluidic channels, to enable precise control of sample delivery and mixing.
While many portable fluorescence sensing systems still rely on external instruments to aid in their operation, such as a laptops or computers connected to controlling and processing the data, single-board computers like Raspberry Pi or Arduino offer a solution to this problem. Raspberry Pi computers are compact, low-cost, and require low power, making them ideal for portable applications [130,131]. They can be used to handle complicated tasks such as image processing, data analysis, and controlling peripheral devices, simultaneously. Alternatively, the Raspberry Pi can be integrated with a smartphone, controlled, and accessed with a smartphone app. The Raspberry Pi can even control the smartphone itself, sending signals to its built-in sensors, or triggering actions based on data from the smartphone's camera or GPS. Another way to connect Raspberry Pi and smartphones is through wireless communication protocols such as Bluetooth or Wi-Fi. For instance, the Raspberry Pi can be configured as a Wi-Fi hotspot, allowing smartphones to connect to it and access its resources or services, thereby facilitating data transfer and processing in real-time.

Conclusions
The present review paper highlights some of the recently developed fluorescencebased portable assay techniques for the sensitive and effective detection of biochemical analyte, pathogens, anions, and cations, as summarized in Table 1. These fluorescence assays were integrated into paper-based platforms or hydrogels and coupled with smartphonereadout devices or even wearable optical sensors. Although fluorescence-based assays are becoming widely popular for the point-of-care detection of important biological and environmental analytes, some challenges lie ahead, which include improving the signal-to-noise ratio, multiplexing, and expanding its utilization for a wide variety of analytes. One specific way of increasing the fluorescence intensity is by utilizing surface plasmon resonance. This can be achieved by integrating plasmonic (typically gold or silver) nanoparticles or substrates with the assay platform [126,132]. This approach has been shown to enhance the fluorescence signal manyfold, and can also be easily integrated with POC devices [133][134][135][136][137]. Additionally, the advantage of using plasmonic nanoparticles is that they enable the colorimetric sensing of biomarkers. A few of the plasmonic NP-based tests are already available commercially, e.g., pregnancy tests and COVID test kits. Various strategies have been employed for signal amplification in colorimetric assays to achieve nanomolar to picomolar level detection, such as the detection of bilirubin using gold nanocages (LOD~39.35 nM), or trypsin using peptide-immobilized AuNPs (LOD 0.5 nM) [138][139][140]. For further improvement of colorimetric assays to the sub-picomolar detection level, different strategies have been applied [141,142]. For example, it has been shown that the sensitivity of plasmonic NPs can be enhanced by modulating their signal intensity. This method includes the acquisition of multiple images of the test and control lines using two distinct wavelengths, 525 nm (which corresponds to the peak absorption of nanoparticles) and 660 nm (where there is no nanoparticle absorption)-at a frequency of 5 Hz, which resulted in the periodic flashing of the test line. This approach led to~58-fold improvement on conventional LIFA readout devices, enabling the detection of sub-200 fM concentrations [142]. Modulation of signal intensity has been used to enhance the signal of multiple modalities, including fluorescence and holography [143,144]. Similar modulation-based strategies can be employed in the POC fluorescence readout, as well as to improve the limit of detection by increasing the signal-to-noise ratio. Another way to improve sensitivity, is with the use of novel fluorescent nanomaterials, which has excellent photostability and a high fluorescence quantum yield (>90%), e.g., quantum dots, polymer dots, carbon dots etc. [145][146][147][148][149][150][151][152][153][154][155]. Another area of improvement is the assay platform, for example by utilizing novel microfluidic chips that enable testing with ultrasmall sample volumes [129,[156][157][158][159][160]. The microfluidic chips can also be used for filtration and sample purification, which can result in higher accuracy [129,161,162]. Another crucial challenge at present is the miniaturization of these assay platforms and detection systems in order to expand their utility, particularly with regard to wearable devices (band, rings, patches, watches, glasses, shoes) [163][164][165][166][167]. Presently, wearable devices are very popular, as they make it possible to monitor a person's vitals in real time [168,169]. These devices are often embedded with the Internet of Things (IoT) for the easy transfer of real-time data to physicians, thereby helping them with diagnostics and the decision-making process [170][171][172][173]. An IoT-enabled time-resolved fluorescence reader for the POC detection of Plasmodium antigen was recently reported [174]. However, the application of IoT in fluorescence-based POC devices is yet to be properly realized. We believe that IoT-enabled sensitive fluorescence-based POC devices will have a great impact in aiding personalized medicine by enabling testing at home or at the point-of-care. These low-cost devices can be used globally and have the potential to alleviate the huge healthcare cost in both developed and under-developed countries.

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