pubs.acs.org/ac Article Low-Cost, User-Friendly, All-Integrated Smartphone-Based Microplate Reader for Optical-Based Biological and Chemical Analyses José Francisco Bergua,⊥ Ruslán Á lvarez-Diduk,⊥ Andrea Idili, Claudio Parolo, Marc Maymó, Liming Hu, and Arben Merkoçi* Cite This: Anal. Chem. 2022, 94, 1271−1285 Read Online See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles. ACCESS Metrics & More Article Recommendations * sı Supporting Information ABSTRACT: The quantitative detection of different molecular targets is of utmost importance for a variety of human activities, Downloaded via 158.109.55.75 on October 21, 2022 at 15:01:58 (UTC). ranging from healthcare to environmental studies. Bioanalytical methods have been developed to solve this and to achieve the quantification of multiple targets from small volume samples. Generally, they can be divided into two different classes: point of care (PoC) and laboratory-based approaches. The former is rapid, low-cost, and user-friendly; however, the majority of the tests are semiquantitative, lacking in specificity and sensitivity. On the contrary, laboratory-based approaches provide high sensitivity and specificity, but the bulkiness of experimental instruments and complicated protocols hamper their use in resource-limited settings. In response, here we propose a smartphone-based device able to support laboratory-based optical techniques directly at the point of care. Specifically, we designed and fabricated a portable microplate reader that supports colorimetric, fluorescence, luminescence, and turbidity analyses. To demonstrate the potential of the device, we characterized its analytical performance by detecting a variety of relevant molecular targets (ranging from antibodies, toxins, drugs, and classic fluorophore dyes) and we showed how the estimated results are comparable to those obtained from a commercial microplate reader. Thanks to its low cost (<$300), portability (27 cm [length] × 18 cm [width] × 7 cm [height]), commercially available components, and open-source-based system, we believe it represents a valid approach to bring high-precision laboratory-based analysis at the point of care. ■ INTRODUCTION The recent COVID-19 pandemic shows how, in our globalized often inappropriate to detect low concentrations of targets.15−17 On the contrary, laboratory-based methodologies world, health and environmental issues can quickly spread produce more precise data that allow better clinical accuracy. across the globe in a matter of days or weeks.1,2 The realization However, they require expensive instruments, multistep of this new status quo is dramatically changing our perspective processes, and trained personnel, which strongly hamper on how we are diagnosing diseases, shifting the focus from a their use in the field.18 Despite the better analytical country-focused surveillance system to a globalized one.3−5 performance displayed by laboratory-testing methodologies, This new horizon forces the research community to develop little effort has been devoted to translating them into portable, more accessible and cost-effective sensing technologies. The cheap, and easy-to-use analytical devices.19 This is mainly due goal is to supply every country with the best available clinical to the technical hurdles related to the integration of all of the resources and know-how to diagnose and contain diseases.6−8 hardware and software into a small device, keeping costs To succeed in this task, the currently available and established low.20,21 Therefore, the development of a low-cost, portable, sensing technologies allow us to pursue two different and easy-to-use device that is able to support laboratory-testing approaches. The first one is based on the use of classical point of care (PoC) sensors (i.e., lateral flow assays and electrochemical sensors such as a glucometer).9−11 The second Received: October 16, 2021 approach is based on the re-engineering of laboratory-based Accepted: December 9, 2021 methodologies for their use at the point of care.12−14 Published: January 3, 2022 Undoubtedly, the former approach includes invaluable tools to carry out extensive, low-cost, and rapid testing directly in the field;9 however, their sensitivity and limits of detection are © 2022 The Authors. Published by American Chemical Society https://doi.org/10.1021/acs.analchem.1c04491 1271 Anal. Chem. 2022, 94, 1271−1285 Analytical Chemistry pubs.acs.org/ac Article Figure 1. Schematic illustration describing the portable device. (A) When the portable platform is closed, the smartphone can be placed on top of the lid for image acquisition (bottom). At the same time, the temperature control module is placed on the top, specifically in the gap found on the top-left corner of the portable platform (top). (B) When the portable platform is open, it is possible to see the different components integrated into the case, such as the microplate (moved by the stepper motor through the y-axis), Bluetooth module, Arduino board, the stepper motor, the temperature, and humidity sensors. (C) Optical system of the portable platform consists of one UV SMD LED and four white LEDs, an excitation filter, an emission filter, and a concave lens to optimize the focal distance between the samples and the smartphone. methodologies at the point of care could strongly impact our software based on open-source systems (i.e., Arduino31,32) method of diagnosing and treating diseases. are valuable tools to achieve the democratization of the Microplates are one of the most used instruments to healthcare system.18 For example, smartphones are low-cost perform laboratory-testing analysis (i.e., biochemical and cell- and portable devices capable of capturing high-resolution based assays) for clinical and biological applications.22−24 photos, displaying enough computing power to analyze Their success comes from their ability to process multiple complex data through a user-friendly interface. 3D printing samples (i.e., commonly 96 individual wells, but some can have technologies allow the low-cost fabrication of almost any kind up to 3456 wells25) and to support different readout strategies of object within minutes or hours using a wide range of (i.e., colorimetric, fluorescence, and luminescence).25 For materials (i.e., plastics, metals).33−36 LEDs represent cheap, example, microplates are commonly used as support in an low-consumption, and tunable light sources that can be easily enzyme-linked immunosorbent assay (ELISA), which is the miniaturized and integrated into a portable device. Finally, gold standard test for the detection of clinically relevant open-source systems allow the control and automation of antigens and antibodies.24,26 They are also routinely used in complex mechanical movements and calculus procedures molecular tests [i.e., polymerase chain reaction (PCR)], cell without the need for expensive and delicate components. In counting, and chemical and biological tests (i.e., drug- this work, we synergically combined these tools to fabricate a screening tests with cells, biochemical tests with bacteria, low-cost (<$300), portable, plate reader capable of carrying chemical characterization of fluorophores, etc.).27−29 Contrary out colorimetric, fluorescence, bioluminescence, and turbidity to the practicality of the multiwell microplate itself, microplate measurements through a commercial smartphone (Figure 1). readers are generally bulky (i.e., not portable), delicate, and We demonstrate how our portable reader can achieve expensive (i.e., ranging between tens to hundreds of thousands analytical performance comparable to the classic commercial of dollars) equipment.18 These features render most of them microplate readers and perform all of the experimental steps nondeployable at the point of care, limiting their adoption in required by bioanalytical assays without the need for further resource-limited environments, such as in low-income instrumentations. ■ countries and teaching laboratories.22 The reason behind their bulkiness and high cost is that they are meant to provide high-quality multiplex measurements in a fully automated MATERIALS AND METHODS laboratory using expensive and delicate components such as Optimization of the Optical System. The resolution and lasers, filters, and computers.30 Fortunately, nowadays, we can clarity of the pictures captured with the portable platform are find analogue components that are cheaper and can provide keystones of our device. Therefore, we used a convergent lens performance similar to their expensive counterparts. to improve the focus capability of the smartphone camera since Smartphones, cheap fabrication methods (i.e., three-dimen- the distance between the smartphone and the ELISA plate is sional, 3D printing), light-emitting diodes (LEDs), and very short (2.0 cm or 20 mm), which limits the focus capability 1272 https://doi.org/10.1021/acs.analchem.1c04491 Anal. Chem. 2022, 94, 1271−1285 Analytical Chemistry pubs.acs.org/ac Article Figure 2. Fluorescence spectra measurements. (A) Microplate is located in the device and the lid is closed, and the fluorescent signal can be easily acquired through a smartphone. The image can be converted to an RGB scale, which allows the quantitative estimation of the fluorescence signal. (B) Emission spectra of the different QDs recorded with the commercial reader (top) and the corresponding pictures taken with the smartphone (bottom). (C) Combinations of different QDs yield different emission colors. (D) Emission spectra of the new combination of different color QDs read with the commercial reader (top) and the corresponding pictures taken using our portable platform (bottom). (the shortest distance the smartphone camera can focus) of the their combinations (to produce cyan, yellow, and pink smartphone cameras and the proper resolution of the images. emission spectra). To perform the measurement, we used In this regard, we optimized the focal length required by the Nunc 96-well black plates with transparent bottom wells optical lens using the following formula because the black plates hinder the cross-contamination of 1 1 1 light between wells while the transparent bottom wells allow = + the samples to be excited from the light source. Then, we f d d′ (1) selected the optimal concentration for each dye using a where “f” is the desired focal length, “d” is the real distance commercial reader. Specifically, we tested different concen- between the smartphone camera and the surface of the liquid trations of QDs diluted in Milli-Q water and we selected those sample to be analyzed, and “d′” is the real focal length of the that provided a similar emission intensity for the three dyes smartphone. We consider using 100 μL of sample per well in (i.e., 25 nM for red QDs, 200 nM for green QDs, and 50 nM the ELISA plate, indicating that the real distance between the for blue QDs) (Figure 2B). After that, we used 50 μL of each smartphone camera and the sample is 17 mm. Therefore, the solution containing the QDs as the working volume and required focal length of the convergent lens is around 11 mm, recorded their fluorescence signal through our portable providing an optimal focus and the best resolution of the platform. Next, to obtain different emission colors, we mixed images captured by the smartphone. and combined 25 μL of each of the red (25 nM), green (200 Chemical Reagents and Materials. Provided in the nM), and blue (50 nM) QDs (Figure 2C,D). Specifically, the Supporting Information. mixing and combination of red QDs and blue QDs yielded Fluorescence Tests. We used the portable reader to magenta color, the combination of red QDs and green QDs measure the signal produced by three different fluorescent yielded yellow color, and the combination of blue QDs and dyes. More specifically, we used Q21341MP red quantum dots green QDs yielded cyan color. (QDs) (excitation 365 nm, emission 615 nm), blue QDs To estimate the limit of detection (LoD) of the portable and (excitation 365 nm, emission 420 nm), and Q21321MP green the commercial microplate reader for a dye, we measured the QDs (excitation 365 nm, emission 530 nm). We demonstrate minimum concentration of the selected fluorescent label able that our device can detect different emission wavelengths to generate an instrumental signal of three times the standard (QDs can all be excited at 365 nm independently of the deviation of the background signal. Specifically, the limit of emission spectra). Specifically, we measured both individual detection was estimated using the following equation in QDs (producing red, green, and blue emissions) as well as SigmaPlot (v 14.5) 1273 https://doi.org/10.1021/acs.analchem.1c04491 Anal. Chem. 2022, 94, 1271−1285 Analytical Chemistry pubs.acs.org/ac Article 3·σBLANK HRP−streptavidin (0.05 μg/mL) to the wells and incubated LOD = [Dye]· (FDYE − FBLANK ) (2) for another 30 min at 37 °C. Of note, it is crucial to perform three to five repetitions of washing steps after every incubation where [Dye] is the concentration of the dye, σBLANK is the process to avoid the unspecific binding of biomolecules, which standard deviation of the fluorescent signal in the absence of could lead to false-positive signals. Finally, we added 100 μL of the dye, FDYE is the fluorescence signal emitted by the dye, and TMB (substrate solution) to the wells and incubated for 20 FBLANK is the fluorescence background signal. min at 37 °C, followed by the addition of 50 μL of 1 M H2SO4 Colorimetric ELISA Tests. To perform ELISA assay, we (stop solution). The plates were immediately read either with used 96-well white clear bottom microplates (Fisher Scientific) the commercial reader (using dual-wavelength detection at 450 to avoid the well-to-well cross-contamination of light of fully nm to quantify the biochemical oxidation of TMB and at 620 transparent plates. We followed a standard immune-sandwich nm to lessen unspecific absorptions from the media) or with ELISA protocol for the detection of the SARS-CoV-2 the portable reader. nucleoprotein antigen, which was recently developed by our Data Analysis of the Colorimetric ELISA Tests. group (Figure 4). First, we coated the wells using 100 μL of a Although the commercial reader can selectively and specifically solution of a capture antibody (polyclonal, 40588-T62, read the yellow optical signal from the wells using the SinoBiological) at a concentration of 5 μg/mL prepared in absorbance value at 450 nm, the smartphone camera only 0.05 M carbonate-bicarbonate buffer (CBS), pH 9.6, for 16 h captures the image but cannot directly quantify the color at 4 °C. Next, we performed a washing step using 100 μL of intensity. To overcome this, we used ImageJ to analyze the washing buffer per well (0.01 M phosphate buffer saline at pH pictures collected after the ELISA tests. Specifically, we 7.4, 0.05% Tween-20), and then we added 3% bovine serum converted the yellow color to a pale gradient into a white to albumin (BSA) (in phosphate-buffered saline, PBS buffer) as a black gradient, which can be easily analyzed by ImageJ. To blocking agent for 2 h at 37 °C. Of note, the blocking agent achieve this, we exported the pictures obtained within the allows coating the uncovered area (without the primary portable platform using the smartphone camera to the software antibodies) on the ELISA wells to prevent further unspecific ImageJ. Then, we set the hue threshold (range of hue values in adsorption of other biomolecules present in the sample during the HSL encodings of the RGB system) between 35 and 45, the immunoassay. We then performed another washing step corresponding to the yellow color in ImageJ, which turns the using the washing buffer. After this step, we added different yellow color to black. Once the black to white images were concentrations of nucleoprotein (from 1 ng/mL to 1 μg/mL) obtained, we quantified the color intensity of the area dissolved in PBS (0.01 M, pH 7.4) to the wells. We let the corresponding to the microplate wells as follows: the whiter solution incubate for 1 h at 25 °C, washed the wells three times the images, the higher the values, whereas the darker the (using the washing buffer), added the detection antibodies images, the lower the values. Afterward, we plotted these data (monoclonal, MA1-7403, ThermoFischer) and let them on a scatter diagram and compared the results obtained by the incubate for 1 h at 25 °C. After this step, we washed the commercial reader and the portable platform using the plate three times with the washing buffer and added the anti- concentration of antigen (i.e., nucleoprotein or human IgG) mouse IgG (modified with horseradish peroxidase, HRP) for 1 as the x-axis and the colorimetric signals recorded as the y-axis. h at 25 °C at a concentration of 0.05 μg/mL prepared in PBS The resultant data were fitted using a four-logistic equation in (0.01 M, pH 7.4). Finally, after washing the plate three times SigmaPlot (v 14.5) ÄÅ É bÑ with the washing buffer, we proceeded with the addition of 100 ÅÅ Ñ (S[0Target] − S[MAX Target]) ÅÅ1 + [Target] ÑÑÑ μL of the colorimetric substrate (tetramethylbenzidine, TMB) S[Target] = S[MAX ÅÅ ÑÑ Target] + ÅÅÇ ÑÑÖ at a concentration of 0.1 M; then, after 20 min, we added 50 μL of 1 M H2SO4 to stop the enzymatic reaction. After this last ( C50% ) (3) step, we collected the colorimetric signal using both a commercial reader (using dual-wavelength detection at 450 where S[Target] is the colorimetric signal in the presence of nm to quantify the biochemical oxidation of TMB and at 620 different concentrations of the target, [Target] is the target MAX nm to lessen unspecific absorptions from the media) and our concentration, S[Target] is the colorimetric signal seen at 0 portable reader using the smartphone. saturating concentrations of the target, S[Target] is the For the second ELISA test, we followed a standard immune- background signal seen in the absence of the target, C50% is sandwich ELISA protocol for the detection of human the midrange concentration of target (inflection point), and b immunoglobulin G (human IgG) (Figure S11). First, we is the slope factor. added 100 μL of anti-HIgG antibody at 2 μg/mL diluted in Bioluminescent Assays. To grow Aliivibrio fischeri for the carbonate-bicarbonate buffer (CBS; 0.05 M, pH 9.6) to coat toxicity measurements, we thawed at room temperature a stock the ELISA wells overnight at 4 °C. Then, we removed the culture of A. fischeri frozen at −80 °C for 10 min. Then, we solution and washed the wells three times using 250 μL of the added 2.5 μL of this stock culture to 25 mL of marine broth washing buffer (0.01 M phosphate buffer saline at pH 7.4, 0.1% (MB medium), and the culture was left to grow at 25 °C and Tween-20), followed by the addition of 200 μL of 3% BSA (w/ 135 rpm (orbital shaking) for 20 h (SSM1 mini orbital shaker v) (in PBS buffer) as the blocking agent for 1 h at 37 °C and from Stuart). Our MB medium contains 0.5% tryptone (w/v), repeated the washing step afterward. Next, we added 100 μL of 2% NaCl (w/v), 0.3% yeast extract (w/v), and 0.3% glycerol human IgG diluted in PBS (0.01 M, pH 7.4) at different (v/v) dissolved in Milli-Q water.37 We monitored the bacterial concentrations to the wells precoated with anti-HIgG antibody growth by measuring the bioluminescence produced in the and incubated for 45 min at 37 °C. After this step, we removed commercial reader resulting always in an optimal growth the solutions and added 100 μL of a solution containing the between 20 and 24 h after inoculation in the raw media. Once anti-HIgG antibody at 0.05 μg/mL in PBS to the wells and the cellular density reached at least 108 CFU/mL and incubated for 30 min at 37 °C. Then, we added 100 μL of bioluminescence was higher than 106 dimensionless units 1274 https://doi.org/10.1021/acs.analchem.1c04491 Anal. Chem. 2022, 94, 1271−1285 Analytical Chemistry pubs.acs.org/ac Article (recorded with the commercial reader SpectraMax iD3), we 50 μL of antibiotics at different concentrations (from 0 to 125 used our plate reader to carry out the bioluminescent assay μg/mL diluted in appropriate growth media) to the ELISA (please refer to Figure S18 in the Supporting Information, SI wells containing 50 μL of bacteria. This step was followed by for a detailed explanation of the bacterial growth protocol). We an incubation step performed at 25 °C with shaking for A. mixed 50 μL of bacteria and 50 μL of the sample fischeri and at 37 °C without shaking for E. coli. Eventually, we [pentachlorophenol (PCP) at different concentrations diluted determined the bacterial growth using the commercial reader in 2% NaCl water] in a well and incubated them under to measure the OD600 and the smartphone together with the agitation using the stepper motor (using the agitator portable platform to analyze the turbidity of the media based functionality of the reader) for 5 min. Eventually, individual on the dark wax-patterned circles. Briefly, to increase the pictures from each well were captured with the smartphone contrast between the growth media and the media containing a using ISO 1000 and different integration times. The pictures high concentration of bacteria, a dark spot of wax printed over captured with the smartphone were later analyzed using the a white paper was placed at the center below the ELISA wells. software ImageJ (version 1.53h). This new method allows for increasing the contrast between To quantify the toxicity of PCP over A. fischeri, we those samples containing low concentrations of bacteria determined the EC50 (half-maximal effective concentration) (darker spot, light not reflected) and high concentrations of for both the portable and the commercial microplate readers. bacteria (whiter spot, light reflected by the presence of We measured the bioluminescence intensity of the bacteria particles [bacterial cells] within the media), boosting the after the exposure to different concentrations of PCP for 5 min sensitivity of the detection of bacterial growth within the and subtracted the value of the blank (growth media without portable platform. bacteria) from all of the measurements. Afterward, we divided We determined the % of bacterial growth for both the all of the values by the bioluminescence of the control culture portable and the commercial microplate readers to quantify the (bacteria without antibiotic) and plotted all of the results on a antimicrobial capacity of the antibiotics against the different scatter diagram to calculate the EC50 values for each reader, strains of bacteria. We measured the turbidity of the media with the x-axis being the concentration of PCP and the y-axis according to the method previously explained, where the the normalized bioluminescence. Specifically, the EC50 was absence of bacterial growth due to a high concentration of estimated using the following equation in GraphPad Prism (v antibiotics yields a darker wax circle than the growth of 8.0.1.244) bacteria that yields a wither wax circle. First, we subtracted the value of the blank (growth media without bacteria) from all of (top − bottom) the measurements. Afterward, we divided all of the values by y = bottom + (1 + 10(x − log EC50)) (4) the turbidity value obtained with the control culture (bacteria without antibiotic) and plotted all of the results on a scatter where x is the concentration of PCP, y is the bioluminescence diagram to calculate the corresponding EC50 values, with the x- output of the bacteria, Top and Bottom are parameters axis being the concentration of antibiotic in μg/mL and the y- calculated by the software after the mathematical adjustment axis the growth percentage (growth %, au) corresponding to and fitting of all of the data, and EC50 is the concentration of the normalized turbidity of the media. Specifically, the % of PCP that inhibits 50% of the bacterial bioluminescence. inhibition was estimated using the following equation in Description of the Production of Wax-Patterned SigmaPlot (v 14.5) Circles. We created a matrix (12 × 8) of circles of 0.108 cm The equation for drug screening using the commercial in diameter aligned at a distance of 0.905 cm from each other reader (distance between each well of the 96 ELISA plate) so that % growth inhibition every circle is in the center of each well. The design was printed on a poly(ethylene terephthalate) (PET) substrate of (OD600sample − OD600 blank ) 12.5 × 8.2 cm using a wax (solid-state) printer (Xerox = × 100 OD600noantibiotic (5) ColorQube 8570DN) in black color. It should be noted that other types of printers (consumer inkjet or laser printer) can The equation for drug screening using the portable platform be used; however, wax printing has a transmittance of 0% and produces a pure black color. In this way, maximum efficiency (Isample − Iblank ) % growth inhibition = × 100 in terms of contrast is achieved. Inoantibiotic (6) Bacterial Culture and Drug Screening. We grew A. fischeri in marine broth (MB) at 25 °C either in an orbital OD600sample refers to the optical density at 600 nm of the agitator or within the portable platform under shaking bacterial growth with a certain concentration of antibiotic, conditions (20 h cultivation). We grew the two strains of whereas OD600blank refers to the optical density of the raw Escherichia coli and Salmonella typhimurium in tryptic soy broth media without bacteria and OD600noantibiotic refers to the (TSB) at 37 °C either in the incubator or within the portable optical density of the bacterial growth without any antibiotic in platform (18 h cultivation) without any shaking (Figure S20). the equation used for the commercial reader. In this sense, the The growth was monitored by including blank samples (media subscripts used for intensity (i.e., Isample) in the equation used for the portable platform refer to the same parameters. ■ without bacteria) and comparing the optical density at 600 nm of the cultures with and without bacteria after the incubation process. RESULTS AND DISCUSSION We diluted the bacterial cultures to normalize the bacterial Optical and Mechanical Platform Design. Our portable concentration to 2·108 CFU/mL (corresponding to a value of reader is a rectangular-shaped box (27 cm [length] × 18 cm optical density of 0.2 au) by adding the appropriate volume of [width] × 7 cm [height]) made of polyoxymethylene, which is growth media to reach a final volume of 50 μL. Next, we added designed to contain all of the hardware components (i.e., 1275 https://doi.org/10.1021/acs.analchem.1c04491 Anal. Chem. 2022, 94, 1271−1285 Analytical Chemistry pubs.acs.org/ac Article electronic, mechanical, and optical) required to perform the samples to be analyzed). The emission filter (long-pass filter different bioanalytical tests (Figure 1). As an optical reader, we 400 nm) allows avoiding the collection of the light coming used a commercial smartphone, which also allows controlling from the excitation source, thereby reducing the background the different components of the device through an integrated signal observed in the pictures captured using the smartphone. app (Figures 1A and S3A). To make the overall device more The light source is located on the bottom (within the portable versatile and easier to use, we incorporated a sliding lid into the platform) and exploits five LEDs (Figure 1C): one UV LED device to allow easy loading of the plate and correct acquisition with an excitation wavelength at 365 nm to support of the image by the smartphone. Specifically, the optical lens fluorescence assays and four white LEDs to support different on the lid indicates where the smartphone should be placed to colorimetric assays. To carefully select the correct excitation keep its camera aligned to the selected well (Figure 1A, wavelength, we combined the UV LED with an excitation filter bottom). At the same time, the lid guarantees stable lighting (bandpass 370 nm) to perform fluorescence-based assays and conditions to obtain reproducible images that can be improve the acquisition of the optical signal. quantitatively analyzed. We installed several functional Finally, the smartphone plays a key role because it allows us components inside the device: an Arduino microcontroller, to acquire the image directly from the well plate and collect the the thermoelectric module, a linear actuator coupled with a optical signal. In this work, we used a smartphone (Huawei stepper motor, the optical system, humidity, and temperature P20 Lite) introduced into the market (in the midrange) almost sensors, and the complementary electronic elements (i.e., 4 years ago, which displays a camera of 16 MegaPixels, a motor driver, resistors, LEDs, etc.) (Figure 1B). All of these diaphragm aperture of f/2.2, a sensor size of 1/3.1″, and a components are commercially available, and we selected them pitch size of 1.1 μm. These represent the minimum to keep the overall cost of the device below $300 (Table S1). requirements that the smartphone must display to obtain the Of note, we did not include the price of the phone (Huawei same results we obtained and described throughout this work. P20 Lite) in the overall cost of the device because it may vary Moreover, taking into account that manufacturers are currently from country to country and brand to brand and because producing smartphones using better sensors and optics in the smartphones are widespread and used all over the world. In low-/midrange of the market, the use of a new and different 2021, for example, the number of smartphone users in the smartphone can only positively affect the acquisition of the world stood at 6.378 billion, which is over 80% of the world image from the well plate, improving the overall sensitivity of population.38 the device. To monitor and precisely select the temperature inside the Automation, Interface Design, and Device Operation. device, we installed a thermoelectric module (also known as a The functional core of our device is shared by the Arduino Peltier module), which works in combination with the microcontroller and the smartphone through a dedicated app, temperature and humidity sensors (Figure 1A, top). Indeed, which allows them to work in synergy, making the device through the smartphone app, we can precisely and quickly intuitive and easy to use. Specifically, we created the app using (minutes) set specific temperature conditions. The module MIT App Inventor 2, an open-source web application that includes two heat sinks and two fans that allow the automatic allows newcomers to computer programming to create and rapid change of the temperature gradient without the need application software.38,39 The graphic user interface (GUI) of for user intervention (Figure S1). Specifically, we can lower the our app is shown in Figure S3, together with all of the blocks temperature up to 7 °C and increase it by more than 30 °C; containing the logical instructions for programming it. The app therefore, if the room temperature is 25 °C, we can cover the is connected to the Arduino device through a Bluetooth temperature range from 18 to 57 °C (Figure S2). At the same connection. The GUI is designed to contain several buttons time, to precisely control the moving and the shaking of the and sliding bars to select and control the operation of the plate inside the device, we integrated a stepper motor in the portable reader (i.e., read temperature and humidity; type of supporting frame (Figure 1B). Specifically, this component is assay: colorimetric, fluorescence, bioluminescence; incubation, integrated into the mechanical frame (linear actuator) and is alignment, temperature adjustment; and the number of designed to house a 96-well plate. Thanks to its spatial fine- measurements) (Figure S3A). Of note, there are commercially tuning, it is possible to correctly align all of the optical parts available apps to read 96-well microplates. However, they have with every single well of the plate. To minimize the size of the three main limitations compared to our system. First, they unit, the movement of the plate is automatized (again using the generally perform just only one kind of measurement, while smartphone) on the z-axis (from well 1 to 12), while the our system can perform four different measurements movement on the x-axis can be performed only manually by (fluorescence, colorimetric, luminescence, and turbidity). placing the desired microplate row aligned with the Second, since they are independent of hardware, their results smartphone camera on top. Additionally, the activation of may be affected by different lighting conditions. Finally, being the motor can be programmed to shake (linear movement up commercially available, the user has limited possibilities to to 160 rpm) the plate, supporting all of the conjugation and adapt the app to their specific needs. incubation steps that are critical in many bioanalytical assays For a better understanding, most of the applications (i.e., ELISA or cell cultures). (concerning their interaction with the Arduino) are designed The core of the device is the optical system, which is to send a number via Bluetooth to the microcontroller that composed of an emission filter, a microlens (to decrease the interprets it as an action to be performed (Figure S3B). For focal distance; for more details, see the Materials and Methods instance, to perform a colorimetric test, we need only to press section), an excitation filter, and light-emitting diodes as the associated button of the app, which sends the number excitation sources (Figure 1C). Specifically, the first two (13). On pressing the associated button, the Arduino powers components are integrated into the lid of the device, and they up the digital pin defined as pinWhite (white light), the app are located between the smartphone (used to acquire the opens the camera to capture a picture of the first well, and images) and the 96-well plate (i.e., containing the different next, send the number (41) to the Arduino to move the 1276 https://doi.org/10.1021/acs.analchem.1c04491 Anal. Chem. 2022, 94, 1271−1285 Analytical Chemistry pubs.acs.org/ac Article Figure 3. Fluorescence measurements. (A) Images of different concentrations of red QDs captured using different smartphone camera settings (same ISO value [1000] and varying integration time values [from 5 to 0.033 s]). (B) Calibration curves obtained to estimate the sensitivity and the limit of detection (LoD) of our platform. We used red QDs as a testbed and modulated both the LoD and sensitivity of our optical platform using different integration times with the smartphone camera (absolute values). (C) Comparison of the normalized calibration curves obtained with the commercial reader (black curve) and the portable platform. The “error bars” we report for fluorescent measurements reflect standard deviations derived using three independent replicates. stepper motor to the next well (710 steps is the distance from Specifically, two are commercially available (i.e., Q21341MP well to well). In this process, there is a variable that counts the red QDs and Q21321MP green QDs from Fisher Scientific) number of times that this cycle is repeated and can be modified and the third one (blue graphene QDs) was synthesized to achieve a specific number of measurements. The rest of the following a previous protocol.43 We selected these dyes assays work similarly. Fluorescence turns on the UV LED while because they display similar absorption spectra, and so they Bioluminescence keeps the device dark. To better understand can be excited with similar efficiency by the same optical this part, see the Arduino Sketch file and the connection source, a UV LED at 365 nm (Figure S5A). Next, we collected diagram Figure S4. Finally, a demonstration of the device in a the images from wells containing different solutions of the real working environment can be seen in Movie S1. three dyes at nanomolar concentrations (Figure 2A). We find Fluorescence Test. Our portable reader allows measuring that the optical filters used for the excitation (bandpass at 370 the fluorescence signal emitted from standard dyes showing its nm) and emission (long-pass at 400 nm) allow the smartphone ability to support multiplexed analysis like commercial reader to detect only the fluorescence signal emitted by each microplate readers. Fluorescence is widely exploited as a signal single dye (Figure 2B). More specifically, the conversion of the readout for the detection of many clinically relevant optical signal to the RGB scale allows obtaining blue, green, biomarkers, thanks to its high sensitivity and rapid and red colors, which correspond to the expected fluoresce detection.40,41 Despite its advantages, the application of signals emitted from three dyes (Figure 2B; for image fluorescence-based diagnostic techniques in low-resource acquisition and analysis procedure, see Figure S6). scenarios has been hampered by the use of bulky, fragile, The images acquired by the smartphone demonstrate the and expensive instruments.33,42 To demonstrate the ability of ability of our system to detect all of the different RGB our device to selectively detect different fluorescence signals, secondary colors associated with the corresponding emission we used a set of three fluorophore molecules as a testbed. spectra. To challenge our device, we tested it against solutions 1277 https://doi.org/10.1021/acs.analchem.1c04491 Anal. Chem. 2022, 94, 1271−1285 Analytical Chemistry pubs.acs.org/ac Article Figure 4. Colorimetric measurements. (A) Schematic of the colorimetric detection of the N-protein from SARS-CoV-2. In step 1, the first capture antibody is attached to the microwell, and then the microwells are washed and blocked with BSA (See Colorimetric ELISA Tests in Materials and Methods). In step 2, the N-protein (i.e., the target analyte) is added to the microwells, and the unbound proteins are removed using a washing buffer. In step 3, a second capture antibody is added and incubated to allow the binding with the N-protein previously captured by the first capture antibody. In step 4, after an additional washing step, an antibody labeled with the HRP enzyme can recognize the second capture antibody. After a final washing step, the enzymatic substrate TMB is added to the solution, and the intensity of the generated yellow color is correlated to the concentration of N-protein in the sample. (B) Raw colorimetric signal (top) collected from the wells using the smartphone and their conversion to black/white scale (bottom) for the estimation of the different concentrations of the N-protein. (C) Calibration curves for SARS-CoV-2 nucleoprotein were obtained with the portable reader (left) and a commercial microplate reader (right). The error bars we report for colorimetric measurements reflect standard deviations derived using three independent replicates. that display emission spectra composed of the sum of two optimized the concentrations of the previously characterized optical signals. Specifically, through the precise combination of blue, green, and red QDs so that they can display the same two dyes, we have created emission spectra that can be fluorescence intensity (Figure S7). Then, we mixed red QDs at converted into a secondary color of the RGB scale (i.e., 25 nM with green QDs at 200 nM to obtain a fluorescence magenta, yellow, and cyan) (Figure 2C). Of note, the RGB signal corresponding to the yellow color, red QDs at 25 nM scale is a color model in which primary colors (i.e., red, green, with blue QDs at 50 nM to obtain magenta, and green QDs at and blue) are combined in different proportions to produce a 200 nM with blue QDs at 50 nM to obtain cyan (Figure 2C). myriad of different colors.44 The secondary colors are obtained Through the analysis of the acquired images, we demonstrate when two of the primary colors at the same intensity are combined. The software ImageJ (i.e., a free software employed the ability of our system to detect the different fluorescence for the analysis of the images)45,46 allows us to measure the signals which can be converted to the expected secondary color intensity of individual pixels or a selected area of the image, (Figure 2D). The possibility to detect an optical signal providing quantitative data based on the RGB system by giving composed of the combination of two fluorophores demon- three numbers that correspond to the intensity of the red, strates the ability of the platform to fully exploit the RGB scale green, and blue colors, respectively.45 Furthermore, we for multiplexing detection.47,48 1278 https://doi.org/10.1021/acs.analchem.1c04491 Anal. Chem. 2022, 94, 1271−1285 Analytical Chemistry pubs.acs.org/ac Article Figure 5. Bioluminescent measurements. (A) Schematic of the performance of the bioluminescent toxicity assays with A. fischeri. The bacteria A. fischeri are grown in seawater (≈2% NaCl), yielding microbial bioluminescence (on the left). Afterward, the bacteria are placed in contact with a sample containing a specific concentration of pentachlorophenol (PCP, a pesticide) after 20 h of growth (in the center). Eventually, after 5 min of incubation, the presence of a toxic substance induces a decay in the bioluminescence intensity (on the right). (B) Bioluminescence inhibition profile of A. fischeri obtained after 5 min of exposure to different concentrations of pentachlorophenol using a constant ISO 1600 and different integration times (10 s on top and 15 s on the bottom). (C) Presence of PCP in the samples has toxic effects on A. fischeri through oxidative stress and DNA mutations. Therefore, the higher the concentration of PCP, the greater harm to A. fischeri, and thus the stepper bioluminescence decay. On the left, bioluminescence spectra of A. fischeri collected at different concentrations of pentachlorophenol; on the right, normalized calibration curves describing the bioluminescence inhibition profiles of A. fischeri obtained with our portable platform using different smartphone camera settings (blue curves) and the commercial reader (black curve). The corresponding EC50 values are 8.67 mg/L for the microplate (commercial) reader, 8.76 mg/L for the conditions ISO 1000 and 10 s with the portable platform, and 8.80 mg/L for the conditions ISO 1600 and 15 s with the portable platform. The error bars we report for bioluminescent measurements reflect standard deviations derived using three independent replicates. Finally, we demonstrate that our smartphone-based optical image acquisition and analysis, see Materials and Methods and reader displays analytical performance comparable to commer- Figure S6). Furthermore, to improve the detection of the cial microplate readers. Indeed, after demonstrating the fluorescence signal, we explored and characterized two possibility to selectively detect different types of QDs and parameters involved in the image acquisition: the acquisition then different fluorescent signals (Figure 3), we decided to time and the ISO of the phone camera (Figures 3A and S8). characterize our optical system through the estimation of its Through the rational tuning of these parameters, the device limit of detection (LoD) for a selected dye.49 To do this, we can be programmed to achieve different sensitivities (estimated used the red QDs due to their slightly higher absorbance at by LoD from 0.1 to 3 nM) and linear ranges (between 300 and 365 nm, which leads to a higher fluorescence signal. 0.1 nM; Figure 3B). To better characterize the analytical Specifically, we measured the fluorescence at different performance of the device, we compared it with a commercial concentrations of the dye, ranging from 0.1 to 300 nM. We reader. The estimated LoD for the commercial reader (0.14 ± observe that the intensity of the red color is perfectly 0.07 nM) is comparable to the value observed for our optical correlated to the dye concentration (Figure 3A). Again, reader, such as the useful linear range (Figure 3C). Finally, to using ImageJ, we can easily extract the relative RGB value, demonstrate the versatility of our device, we detected different which allows us to perform a quantitative measurement (for concentrations of a second commercially available fluorophore. 1279 https://doi.org/10.1021/acs.analchem.1c04491 Anal. Chem. 2022, 94, 1271−1285 Analytical Chemistry pubs.acs.org/ac Article To do this, we used fluorescein (FAM), and using the same magnitude (from 30 to 1000 ng/mL). Such results are approach, we can track its fluorescent signal estimating a LoD comparable to those obtained using a commercial plate reader of 0.32 ± 0.05 μM (Figure S9). The obtained LoD, in this (Figure 4C, right), where we observe a slightly lower LoD case, is higher than the value estimated for the commercial (8.74 ± 0.03 ng/mL) and a linear range spanning from 10 to reader (0.14 ± 0.07 nM) because the excitation wavelength of 1000 ng/mL. Finally, to further demonstrate the diagnostic fluorescein is around 470 nm, while the UV LED installed properties of our device, we used the same approach to detect within the portable platform emits at 365 nm. In this sense, the IgG antibodies as a second biomarker observing similar LoD could be further improved by installing an excitation UV analytical performance (Figure S12). LED at 470 nm for fluorescein-related applications. This result Bioluminescent Assays. The portable smartphone-based highlights the ability of our system to perform quantitative reader can also support the reading of bioluminescence, detection of commercial fluorescent organic dyes. extending the possible real-world applications of the device. Colorimetric ELISA Test. Our device can support Over the last three decades, bioluminescent bacteria have been colorimetric-based diagnostic techniques allowing the direct exploited to develop bioanalytical platforms to achieve quantification of clinically relevant biomarkers. To evaluate its environmental monitoring.54−59 Their sensing mechanisms analytical performance as a testbed, we carried out the ELISA rely on the metabolic capabilities of such bacteria to produce assay, which is recognized as one of the clinical gold standard light in response to an enzymatic reaction (carried out by a techniques for the detection of protein biomarkers.50 More luciferase enzyme).56,60 Because this signal is a direct indicator specifically, ELISA tests are immunoassays used to detect of their metabolic status (which is in communication with the analytes (i.e., antibodies, antigens, proteins, and small media), its changes or variations can be used to track the molecules) employing enzymatic-tagged antibodies that presence of toxic and dangerous compounds61−63 (Figure 5A). provide amplification of the optical output, which is directly Here, as a testbed, we employ the bacteria A. fischeri (a marine proportional to the concentration of the analyte. Their Gram-negative bacteria) as a bioluminescent probe to detect popularity comes from their high throughput (i.e., when the presence of the pesticide pentachlorophenol (PCP) in combined with microplates) and sensitivity (i.e., due to the water.61,64 enzymatic reaction) that allow measuring nanomolar and First, we characterized and optimized the signal acquisition picomolar ranges of biomarkers in hundreds of patients in a process of the smartphone camera. To do this, we explored the single day.51−53 Despite its clinical relevance, the long protocol effect of different ISO values and integration time with respect (hours) based on batch processes (i.e., washing, dilutions, and to (1) the intensity of the bioluminescence and (2) the incubations steps) and expensive readouts have prevented their contrast of the pictures acquired. To do this, we created a use at the point of care or in resource-limited settings.22 Here, matrix of images by selecting five different ISO values (400, the goal is not only to demonstrate the ability of our 640, 800, 1000, and 1600) and five different integration times smartphone-based platform to achieve diagnostic data but also (4, 6, 8, 10, and 15 s). Using these parameters, we collected the to make the overall assay more user-friendly through the use of bioluminescent signal from different wells provided by bacteria a single device. Specifically, the versatility of our system helped (Figures 5B and S13). Comparing the different pictures, we us to perform all of the critical steps required for the can observe how the use of low ISO values and short preparation of the assay. For example, the temperature sensor integration time provides low reflection and light contami- and the mechanical motion (i.e., shaking of the plate) allow nation; however, this leads to very low intensity values and fully performing all of the incubation steps without the need contrast of the bioluminescence signal captured with the for further instruments. smartphone. On the contrary, high ISO values and integration To test our smartphone-based plate reader, we challenged it time provide brighter images, but the overexposure times often with an ELISA test for the detection of the clinically relevant lead to burned images that become tricky to analyze (Figure SARS-CoV-2 N-protein (Figure 4A). To achieve this, we S12). Hence, to achieve a good compromise among bio- developed an immune-sandwich assay based on the use of two luminescence intensity, light reflection, and optimal contrast, primary antibodies, which can recognize the presence of the N- we selected an ISO value between 1000 and 1600 combined protein through the formation of a surface-adsorbed immune- with a long integration time (10 and 15 s; Figure 5B). sandwich complex (Figure 4A). The presence of the secondary Next, we used the portable plate reader to measure the HRP-labeled antibody (streptavidin−horseradish peroxidase) bacteria’s bioluminescence in the presence of increasing and the enzymatic substrate [i.e., tetramethylbenzidine concentrations of PCP, a broad-spectrum pesticide generally (TMB)] allows the recognition of the sandwich complex and found in water samples.65,66 More specifically, PCP is the quantification of the N-protein through (Figure 4B, top). metabolized into electrophilic compounds that bind to DNA Specifically, after the completion of the enzymatic reaction (15 and proteins, causing double and single DNA strand breaks, min), the colorimetric signal is collected from each well of the base oxidation, formation of reactive oxygen species (ROS), plate using the smartphone. Then, the acquired images are and protein adducts.67 We tested the bacterial population processed and analyzed by ImageJ, and they are converted to a against six different, environmentally relevant concentrations of grayscale color to improve the contrast with the background PCP spanning the concentrations range from 1.5 to 100 mg/ color (Figure S10; see Materials and Methods for further L61,64,68 (Figure S14). As expected, the toxicity of PCP inhibits details). Next, we have plotted the obtained values in function the cellular fitness of the bacteria affecting the luciferase of the N-protein concentrations used during the assay, and we activity, inducing a decrease in the overall bioluminescent observe the expected sigmoidal curve (Figure 4C, left and signal (Figure 5B).64 This behavior is further confirmed by the Figure S11). Using a four-logistic equation fit, we estimated a bioluminescence spectra, which do not display a change in the limit of detection of 31.4 ± 0.3 ng/mL (here, the errors emission wavelength distribution but only in the intensity represent the relative standard deviation among three different (Figure 5C, left). Plotting the bioluminescent signal as a wells) and a linear dynamic range that spans 2 orders of function of the concentration of PCP, we can build toxicity 1280 https://doi.org/10.1021/acs.analchem.1c04491 Anal. Chem. 2022, 94, 1271−1285 Analytical Chemistry pubs.acs.org/ac Article Figure 6. Turbidimetric measurements. (A) On the left: schematic of the principle underlying low bacterial concentration (=low turbidity) and high bacterial concentration (=high turbidity); the presence of an active antimicrobial agent (i.e., antibiotic) hinders bacterial growth, promoting bacterial death, and thus low turbidity media as a result of low bacterial growth. Conversely, the absence of an active antimicrobial agent does not prevent the growth of bacteria, which reach high concentrations and thus result in high turbidity media. On the right: schematic of the performance of the turbidimetric measurements using the portable platform. A dark circle made of wax is stuck on the bottom of the microplate wells, and the bacterial samples are illuminated from below with a white LED light. In the presence of many bacteria, the bacterial cells scatter the white light rays in many directions, which makes the dark wax circle appear whiter from the top, where the smartphone camera captures the images. Conversely, in the absence of bacteria, the white light rays are not scattered in such a high frequency, and thus the dark wax circle appears darker. (B) Images acquired with the smartphone and the portable platform for the growth inhibition of different bacteria using different antibiotics. On the left: samples of A. fischeri grown in the presence of different concentrations of kanamycin (from 0 to 64 μg/mL). The higher the concentration of kanamycin, the lower the turbidity and darker the wax circle because of the absence of bacterial growth. On the right: samples of E. coli ATCC25922 grown in the presence of different concentrations of amoxicillin (from 0 to 64 μg/mL). The higher the concentration of amoxicillin, the lower the turbidity and darker the wax circle because of the absence of bacterial growth. (C) Growth inhibition profiles of E. coli ATCC11303 corresponding to the turbidity changes under different antibiotic concentrations in the portable platform (left) and the commercial reader (right). For the commercial reader: EC50Pen = 29.15 μg/mL, EC50Amp = 0.04 μg/mL, EC50Amo = 1.42 μg/mL, and EC50Kan = 2.32 μg/mL. The error bars we report for turbidimetric measurements reflect standard deviations derived using three independent replicates. profiles.69,70 Then, using a sigmoidal fit, we estimated EC50 commercial reader. Despite this, the device can detect the PCP values71 (i.e., PCP concentration that causes 50% bio- in its relevant range, demonstrating again its versatility and luminescence inhibition), which correspond to 9 ± 2 and 16 ability to support sensing assays based on bioluminescent.72 ± 5 mg/L using ISO 1000/10 s and ISO 1600/15 s, Drug Screening Test. Finally, we demonstrate the use of respectively. These data are slightly higher than those obtained our portable reader to perform a conventional growth from the commercial reader (4.0 ± 0.3 mg/L) (Figure 5C, inhibition assay. Generally, agar or broth dilution methods right). We believe this difference is due to two factors: the are used to determine the minimal inhibitory concentration faster shaking intensity of the commercial reader that promotes (MIC) of antimicrobial substances73 (Figure 6A, left). This aeration and thus bioluminescence and the lower sensitivity of value describes the lowest concentration of the antimicrobial the smartphone camera compared to the optical sensor of the agent able to inhibit the growth of a bacterial population. 1281 https://doi.org/10.1021/acs.analchem.1c04491 Anal. Chem. 2022, 94, 1271−1285 Analytical Chemistry pubs.acs.org/ac Article Therefore, through its estimation, it is possible to characterize screening device. In this regard, we tested four different strains the activity of antimicrobial drugs against different bacteria, of bacteria (A. fischeri, E. coli ATCC11303, E. coli ATCC25922, providing important information for clinicians, for example, and Salmonella) with different concentrations of four anti- during the selection of the pharmacological treatment.74 Broth biotics (kanamycin, ampicillin, penicillin, and amoxicillin). dilution assay is performed in test tubes, and it requires only a After 20 h of growth (at 25 °C for A. fischeri and at 37 °C for E. naked-eye reading of the visible inhibition growth.75 During coli), we evaluated the turbidity of the media [marine broth the last few decades, its miniaturization and mechanization (MB) for A. fischeri and tryptic soy broth (TSB) for E. coli], through the use of small and disposable microdilution trays estimating the optical density (OD) at 600 nm using the have made this method practical and very popular.76 Using this commercial reader (Figures S17 and S18). Through this first setup, for example, the bacterial growth over time is estimated screening test, we selected only the couples of bacteria and through the measurement of the turbidity, which corresponds antibiotics that display an inhibiting effect to repeat the to the absorbance value of the liquid microbial culture measurements using our portable platform (Figure 6C). The collected at 600 nm. So, the higher the bacterial concentration results demonstrate the ability of our device to perform an in the solution, the greater will be the scattering of light from inhibition test, and we find a correlation between the two the solution, inducing a higher absorbance value. methods, observing a slightly higher sensitivity for the Having this in mind, we adapted our portable platform to commercial reader. support the turbidity measurements. Of note, it is extremely difficult to determine and quantify changes in turbidity with a conventional smartphone camera due to the lack of contrast ■ CONCLUSIONS Here, we showed that by purchasing and assembling between different images under the same illumination commercially available, low-cost hardware components, and conditions.77,78 To overcome this, we placed a dark spot using open-source software with a smartphone, it is possible to below the ELISA wells to introduce a contrast between the create a portable plate reader that supports optical-based growth media and the media containing a high concentration analysis with analytical performance like those found in bulky, of bacteria (Figure S15). Specifically, we used as a template the expensive commercial readers (Table S2). Specifically, we dimensions of a 96-well plate to print a transparent PET sheet demonstrated how the portable plate reader can successfully with a pattern of black wax circles (radius of 0.108 cm) measure fluorescent, colorimetric, bioluminescent, and turbid- centered at each well. The area of black circles is smaller than ity signals. Subsequently, we employed it to carry out gold the area of the well, so this ensures the passage of light through standard analytical assays such as an ELISA for the detection of the wells from the excitation source. Next, we placed the wax a SARS-CoV-2 N-protein, a toxicity test measuring the effect of printed sheet on the bottom of the 96-well plate (stuck on the a pesticide on bioluminescent bacteria, and a drug-screening metallic plate, see Figure S15), and we tested how the presence test checking the antimicrobial activity of four different of bacteria and its growth can affect the black color of the antibiotics. Taking full advantage of the smartphone circular background. capabilities, in addition to employing it to control the device Specifically, we tested solutions containing different and measure the signal, we have also used it to perform concentrations of bacteria, and we acquired the image using multistep assays that may be easy to follow also by the smartphone (Figure S16). To improve the analysis of the nonspecialized personnel. Despite the advances achieved in picture, we converted it to the white/black scale using ImageJ, terms of functionality, versatility, ease of use, and portability, and we sampled only the area of the black circles. We found we believe this platform could be improved further. More that the black circle (not in contact with the solution) becomes specifically, sensitivity, accurate image acquisition, and data whiter due to the presence of bacterial cells, which induces the processing remain the main challenges. These can be easily scattering of the light (Figure 6A, right). Accordingly, in the improved by exploiting the modularity of the platform through presence of a higher concentration of bacteria, we observe a the use of more expensive and better components (i.e., whiter wax circle in the smartphone images (Figure S16). hardware and software), which will allow improving the Therefore, this new method allows to quantify and differentiate performance but sacrifice the overall cost. samples containing low concentrations of bacteria (darker spot, Although there are commercially available 96-well portable light not reflected) and high concentrations of bacteria (whiter readers in the market, their cost is generally higher than 3000€ spot, light reflected by the presence of particles [bacteria cells] and they are designed to carry out single detection methods.80 within the media), boosting the sensitivity of the detection of We believe that while they may represent a valid tool for bacterial growth within the portable platform (Figures 6 and average-funded groups that want to perform colorimetric S17). Finally, as demonstrated for the previous systems, to measurements in a space-limited laboratory, they fail to increase the sensitivity during the detection of bacterial provide a versatile and low-cost platform. More specifically, growth, we optimized this system by exploring the smartphone the portable reader we present in our manuscript is not only camera settings (Figure S16). The best results are obtained considerably cheaper but can also carry out fluorescence, using a single sheet of wax-patterned PET, the maximum ISO luminescence, and turbidity measurements. Furthermore, since value (3200), and 1/8 s of integration time with the it is built by the user itself, it can also be easily repaired/ smartphone camera, providing a relative change of 42% customized using low-cost materials, whereas commercial between the media without bacteria (MB) and the media portable readers must be sent to the factory, implying the with bacteria (≈ 109 CFU/mL). impossibility to carry out measurements for several weeks, and, Next, we envisaged a direct application for these turbidity in the case the warranty is expired, it would also imply shipping measurements. Since antibiotic resistance has become one of and repairing costs. We hope that this work will lead the way to the major problems the healthcare system is facing nowa- the adaptation of laboratory-based methods for their use at the days,79 the use of a portable platform to detect bacterial point of care, giving healthcare workers new and effective tools resistance to antibiotics provides a powerful functionality as a to control dangerous situations and to provide the best 1282 https://doi.org/10.1021/acs.analchem.1c04491 Anal. Chem. 2022, 94, 1271−1285 Analytical Chemistry ■ pubs.acs.org/ac Article possible care independently of the resources available in each ACKNOWLEDGMENTS region of the world. ■ ICN2 is funded by CERCA programme, Generalitat de Catalunya. Grant SEV-2017-0706 funded by MCIN/AEI/ ASSOCIATED CONTENT 10.13039/501100011033. The authors acknowledge Consejo * Supporting Information sı Superior de Investigaciones Científicas (CSIC) for the project The Supporting Information is available free of charge at “COVID19-122” granted in the call “Nuevas ayudas https://pubs.acs.org/doi/10.1021/acs.analchem.1c04491. extraordinarias a proyectos de investigación en el marco de las medidas urgentes extraordinarias para hacer frente al Materials and methods; list of the components and their impacto económico y social del COVID-19 (Ayudas CSIC- relative costs; thermoelectric module; temperature COVID-19)” and the research project INTCATCH 2020, gradient; graphic user interface; electronic diagram; Development and application of novel, integrated tools for Arduino sketch; raw absorbance spectra and emission of three dyes; optimization of the dyes’ concentrations for monitoring and managing catchments supported by the the RGB secondary colors; optimization of the image European Union’s Horizon 2020 research and innovation acquisition process; ELISA and detection of IgG program (Grant 689341). The authors also acknowledge the antibodies; workflow for colorimetric, fluorescence, project MAT2017-87202-P funded by MCIN/AEI/10.13039/ bioluminescent, and turbidity measurements; calibration 501100011033/and FEDER Una manera de hacer Europa. curves obtained with the portable reader and the R.A.-D. acknowledges funding from the European Union commercial reader; bacterial growth inside the device; Horizon2020 Programme under Grant No. 881603 (Graphene drug screening using the spectrophotometer (PDF) Flagship Core 3). A.I. was supported by PROBIST postdoctoral fellowship funded by the European Research Demonstration of the portable device (Movie S1) (AVI) Council (Marie Skłodowska-Curie grant agreement no. ■ 754510). AUTHOR INFORMATION Corresponding Author Arben Merkoçi − Institut Català de Nanociència i ■ REFERENCES (1) Gubrium, A.; Gubrium, E. Lancet 2021, 397, 2244−2245. Nanotecnologia (ICN2), 08193 Barcelona, Spain; CSIC and (2) Marshall, M. Nature 2021, 594, 168−170. (3) Al-Aly, Z.; Xie, Y.; Bowe, B. Nature 2021, 594, 259−264. the Barcelona Institute of Science and Technology (BIST), (4) Maher, B.; Van Noorden, R. Nature 2021, 594, 316−319. 08036 Barcelona, Spain; Institucio′ Catalana de Recerca i (5) Nsubuga, P.; White, M. E.; Thacker, S. B.; Anderson, M. A.; Estudis Avançats (ICREA), 08010 Barcelona, Spain; Blount, S. B.; Broome, C. V.; Chiller, T. M.; Espitia, V.; Rubina, I.; orcid.org/0000-0003-2486-8085; Sosin, D.; Stroup, D. F.; Tauxe, R. V.; Vijayaraghavan, M.; Trostle, M. Email: arben.merkoci@icn2.cat Publich Health Surveillance: A Tool for Targeting and Monitoring Interventions; Disease Control Priorities in Developing Countries; The Authors International Bank for Reconstruction and Development/The World José Francisco Bergua − Institut Català de Nanociència i Bank: Washington, 2006. Nanotecnologia (ICN2), 08193 Barcelona, Spain; (6) Luo, Y. T.; Wang, J. H.; Zhang, M. M.; Wang, Q. Z.; Chen, R.; orcid.org/0000-0002-5969-193X Wang, X. L.; Wang, H. L. J. Clin. Lab. Anal. 2021, 35, No. 1241506. Ruslán Á lvarez-Diduk − Institut Català de Nanociència i (7) Garg, A.; Ghoshal, U.; Patel, S. S.; Singh, D. V.; Arya, A. K.; Nanotecnologia (ICN2), 08193 Barcelona, Spain; Vasanth, S.; Pandey, A.; Srivastava, N. J. Med. Virol. 2021, 93, 2281− orcid.org/0000-0002-9876-1574 2286. Andrea Idili − Institut Català de Nanociència i (8) Yelagandula, R.; Bykov, A.; Vogt, A.; Heinen, R.; Ö zkan, E.; Strobl, M. M.; Baar, J. C.; Uzunova, K.; Hajdusits, B.; Kordic, D.; Nanotecnologia (ICN2), 08193 Barcelona, Spain; Suljic, E.; Kurtovic-Kozaric, A.; Izetbegovic, S.; Schaeffer, J.; Hufnagl, orcid.org/0000-0002-6004-270X P.; Zoufaly, A.; Seitz, T.; Al-Rawi, M.; Ameres, S.; Baar, J.; Bauer, B.; Claudio Parolo − Institut Català de Nanociència i Beer, N.; Bergauer, K.; Binder, W.; Blaukopf, C.; Bochev, B.; Nanotecnologia (ICN2), 08193 Barcelona, Spain; Barcelona Brennecke, J.; Brinnich, S.; Bundalo, A.; Busslinger, M.; Clausen, T.; Institute for Global Health, 08036 Barcelona, Spain; de Vries, G.; Dekens, M.; Drechsel, D.; Dzupinkova, Z.; Eckmann- orcid.org/0000-0001-9481-4408 Mader, M.; Fellner, M.; Fellner, T.; Fin, L.; Gapp, B. V.; Grabmann, Marc Maymó − Institut Català de Nanociència i G.; Grishkovskaya, I.; Hagelkruys, A.; Handler, D.; Haselbach, D.; Nanotecnologia (ICN2), 08193 Barcelona, Spain Hempel, L.; Hill, L.; Hoffmann, D.; Horer, S.; Isemann, H.; Kalis, R.; Liming Hu − Institut Català de Nanociència i Nanotecnologia Kellner, M.; Kley, J.; Köcher, T.; Köhler, A.; Krauditsch, C.; Kula, S.; (ICN2), 08193 Barcelona, Spain; orcid.org/0000-0002- Lang, S.; Latham, R.; Leitner, M. C.; Leonard, T.; Lindenhofer, D.; 8666-9287 Manzenreither, R. A.; Matl, M.; Mechtler, K.; Meinhart, A.; Mereiter, S.; Micheler, T.; Moeseneder, P.; Neumann, T.; Nimpf, S.; Nordborg, Complete contact information is available at: M.; Ogris, E.; Pagani, M.; Pauli, A.; Peters, J. M.; Pjevac, P.; Plaschka, https://pubs.acs.org/10.1021/acs.analchem.1c04491 C.; Rath, M.; Reumann, D.; Rieser, S.; Rocha-Hasler, M.; Rodriguez, A.; Ropek, N.; Ross, J. J.; Scheuch, H.; Schindler, K.; Schmidt, C.; Author Contributions Schmidt, H.; Schnabl, J.; Schüchner, S.; Schwickert, T.; Sommer, A.; ⊥ Soldoroni, D.; Stadlmann, J.; Steinlein, P.; Strobl, M.; Strobl, S.; Sun, J.F.B. and R.A.-D. contributed equally to this work. The manuscript was written through the contributions of all Q.; Tang, W.; Trübestein, L.; Trupke, J.; Umkehrer, C.; Urmosi- authors. All authors have given approval to the final version Incze, S.; Versteeg, G.; Vogt, V.; Wagner, M.; Weissenboeck, M.; of the manuscript. Werner, B.; Zuber, J.; Födinger, M.; Allerberger, F.; Stark, A.; Cochella, L.; Elling, U. Nat. Commun. 2021, 12, No. 1243296. Notes (9) Mytton, O. T.; McCarthy, N.; Watson, J.; Whiting, P. Br. Med. J. The authors declare no competing financial interest. 2021, No. n1411. 1283 https://doi.org/10.1021/acs.analchem.1c04491 Anal. Chem. 2022, 94, 1271−1285 Analytical Chemistry pubs.acs.org/ac Article (10) Trombetta, B. A.; Kandigian, S. E.; Kitchen, R. R.; Grauwet, K.; (32) Kirthika, V.; Vecraraghavatr, A. K. In Design and Development of Webb, P. K.; Miller, G. A.; Jennings, C. G.; Jain, S.; Miller, S.; Kuo, Y.; Flexible On-Board Diagnostics and Mobile Communication for Internet of Sweeney, T.; Gilboa, T.; Norman, M.; Simmons, D. P.; Ramirez, C. Vehicles, 2018 International Conference on Computer, Communica- E.; Bedard, M.; Fink, C.; Ko, J.; De León Peralta, E. J.; Watts, G.; tion, and Signal Processing (ICCCSP); IEEE, 2018. Gomez-Rivas, E.; Davis, V.; Barilla, R. M.; Wang, J.; Cunin, P.; Bates, (33) Quesada-González, D.; Merkoçi, A. Biosens. Bioelectron. 2017, S.; Morrison-Smith, C.; Nicholson, B.; Wong, E.; El-Mufti, L.; Kann, 92, 549−562. M.; Bolling, A.; Fortin, B.; Ventresca, H.; Zhou, W.; Pardo, S.; Kwock, (34) Cevenini, L.; Calabretta, M. M.; Tarantino, G.; Michelini, E.; M.; Hazra, A.; Cheng, L.; Ahmad, Q. R.; Toombs, J. A.; Larson, R.; Roda, A. Sens. Actuators, B 2016, 225, 249−257. Pleskow, H.; Luo, N. M.; Samaha, C.; Pandya, U. M.; De Silva, P.; (35) Yuk, H.; Lu, B.; Lin, S.; Qu, K.; Xu, J.; Luo, J.; Zhao, X. Nat. Zhou, S.; Ganhadeiro, Z.; Yohannes, S.; Gay, R.; Slavik, J.; Mukerji, S. Commun. 2020, 11, No. 1604. S.; Jarolim, P.; Walt, D. R.; Carlyle, B. C.; Ritterhouse, L. L.; Suliman, (36) Chen, Z.; Li, Z.; Li, J.; Liu, C.; Lao, C.; Fu, Y.; Liu, C.; Li, Y.; S. BMC Infect. Dis. 2021, 21, No. 580. Wang, P.; He, Y. J. Eur. Ceram. Soc. 2019, 39, 661−687. (11) Chaibun, T.; Puenpa, J.; Ngamdee, T.; Boonapatcharoen, N.; (37) Difco & BBL Manual, 2nd ed.; Marine Broth 2216, 2021. Athamanolap, P.; O’Mullane, A. P.; Vongpunsawad, S.; Poovorawan, (38) O’Dea, S. Number of Smartphone Users from 2016 to 2021. Y.; Lee, S. Y.; Lertanantawong, B. Nat. Commun. 2021, 12, Statista GmbH, 2021. https://www.statista.com/statistics/330695/ No. 1065865. number-of-smartphone-users-worldwide/. (12) Stevenson, M.; Metry, A.; Messenger, M. Health Technol. Assess. (39) APP Inventor. MIT App Inventor 2; Technological Institute of 2021, 25, 1−74. Massachusetts, 2021. (13) Alkharsah, K. R. GMS Ger. Med. Sci. 2021, 19, No. 1264310. (40) Patton, E. W.; Tissenbaum, M.; Harunani, F. MIT App (14) Bhadra, S.; Riedel, T. E.; Lakhotia, S.; Tran, N. D.; Ellington, A. Inventor: Objectives, Design, and Development. In Computational D. mSphere 2021, 6, 7−8. Thinking Education; Springer: Singapore, 2019; pp 31−49. https:// (15) Bastos, M. L.; Perlman-Arrow, S.; Menzies, D.; Campbell, J. R. doi.org/10.1007/978-981-13-6528-7_3. Ann. Intern. Med. 2021, 174, 501−510. (41) Zhao, X.; Dai, X.; Zhao, S.; Cui, X.; Gong, T.; Song, Z.; Meng, (16) Khorasani, A.; Chegini, A.; Mirzaei, A. Can. Respir. J. 2020, H.; Zhang, X.; Yu, B. Spectrochim. Acta, Part A 2021, 247, 2020, No. 962332. No. 119038. (17) Silva Júnior, J. V. J.; Merchioratto, I.; de Oliveira, P. S. B.; (42) Gao, L.; Wang, W.; Wang, X.; Yang, F.; Xie, L.; Shen, J.; Rocha Lopes, T. R.; Brites, P. C.; de Oliveira, E. M.; Weiblen, R.; Brimble, M. A.; Xiao, Q.; Yao, S. Q. Chem. Soc. Rev. 2021, 50, 1219− Flores, E. F. J. Virol. Methods 2021, 288, No. 114007. 1250. (18) Balsam, J.; Ossandon, M.; Bruck, H. A.; Lubensky, I.; Rasooly, (43) Bergua, J. F.; Hu, L.; Fuentes-Chust, C.; Á lvarez-Diduk, R.; A. Expert Opin. Med. Diagn. 2013, 7, 243−255. Hassan, A. H. A.; Parolo, C.; Merkoçi, A. Lab Chip 2021, 21, 2417− (19) Stegeman, I.; Ochodo, E. A.; Guleid, F.; Holtman, G. A.; Yang, 2426. (44) Dong, Y.; Shao, J.; Chen, C.; Li, H.; Wang, R.; Chi, Y.; Lin, X.; B.; Davenport, C.; Deeks, J. J.; Dinnes, J.; Dittrich, S.; Emperador, D.; Chen, G. Carbon 2012, 50, 4738−4743. Hooft, L.; Spijker, R.; Takwoingi, Y.; Van den Bruel, A.; Wang, J.; (45) Naren, G.; Hsu, C.-W.; Li, S.; Morimoto, M.; Tang, S.; Langendam, M.; Verbakel, J. Y.; Leeflang, M. M. G. Cochrane Hernando, J.; Guirado, G.; Irie, M.; Raymo, F. M.; Sundén, H.; Database Syst. Rev. 2020, 2020, 2020−2021. Andréasson, J. Nat. Commun. 2019, 10, No. 3996. (20) Hafsiya, T. H.; Rose, B. In An IoT-Cloud Based Health (46) Labno, C. Basic Intensity Quantification with ImageJ; University Monitoring Wearable Device for Covid Patients, 2021 7th International of Chicago, 2021. Conference on Advanced Computing and Communication Systems (47) Rasband, W. ImageJ; U.S. National Institutes of Health: (ICACCS); IEEE, 2021; pp 266−269. Bethesda, Maryland, USA, 2021. (21) Lonini, L.; Shawen, N.; Botonis, O.; Fanton, M.; Jayaraman, C.; (48) Carl, P.; Sarma, D.; Gregório, B. J. R.; Hoffmann, K.; Lehmann, Mummidisetty, C. K.; Shin, S. Y.; Rushin, C.; Jenz, S.; Xu, S.; Rogers, A.; Rurack, K.; Schneider, R. J. Anal. Chem. 2019, 91, 12988−12996. J. A.; Jayaraman, A. IEEE J. Transl. Eng. Health Med. 2021, 9, (49) Zhang, C.; Jiang, Z.; Jin, M.; Du, P.; Chen, G.; Cui, X.; Zhang, No. 4900311. Y.; Qin, G.; Yan, F.; Abd El-Aty, A. M.; Hacimüftüoğlu, A.; Wang, J. (22) Berg, B.; Cortazar, B.; Tseng, D.; Ozkan, H.; Feng, S.; Wei, Q.; Food Chem. 2020, 326, No. 126813. Chan, R. Y. L.; Burbano, J.; Farooqui, Q.; Lewinski, M.; Di Carlo, D.; (50) Shrivastava, A. Chron. Young Sci. 2011, 2, 21−25. Garner, O. B.; Ozcan, A. ACS Nano 2015, 9, 7857−7866. (51) Lenquin, R. M. Clin. Biochem. 2005, 51, 2415−2418. (23) Servais, P.; Garcia-armisen, T.; et al. Can. J. Microbiol. 2007, 53, (52) Shirai, K.; Mawatari, K.; Ohta, R.; Shimizu, H.; Kitamori, T. 798−801. Analyst 2018, 143, 943−948. (24) ThermoFisher Scientific. ELISA (Enzyme-Linked Immuno- (53) Kimoto, M.; Shermane Lim, Y. W.; Hirao, I. Nucleic Acids Res. sorbent Assay), 2021. https://www.thermofisher.com/es/es/home/ 2019, 47, 8362−8374. life-science/protein-biology/protein-biology-learning-center/protein- (54) Guo, S.; Lakshmipriya, T.; Gopinath, S. C. B.; Anbu, P.; Feng, biology-resource-library/pierce-protein-methods/overview-elisa.html. Y. Nanoscale Res. Lett. 2019, 14, No. 222. (25) Jones, E.; Michael, S.; Sittampalam, G. Comput. Sci. 2016, (55) Sharma, P.; Asad, S.; Ali, A. J. Biosci. 2013, 38, 251−258. 1337−1363. (56) Menz, J.; Schneider, M.; Kümmerer, K. Chemosphere 2013, 93, (26) Ciaurriz, P.; Fernández, F.; Tellechea, E.; Moran, J. F.; Asensio, 990−996. A. C. Beilstein J. Nanotechnol. 2017, 8, 244−253. (57) Faria, E. C.; Treves Brown, B. J.; Snook, R. D. J. Environ. Monit. (27) Allkja, J.; Bjarnsholt, T.; Coenye, T.; Cos, P.; Fallarero, A.; 2004, 6, 97−102. Harrison, J. J.; Lopes, S. P.; Oliver, A.; Pereira, M. O.; Ramage, G.; (58) Girotti, S.; Ferri, E. N.; Fumo, M. G.; Maiolini, E. Anal. Chim. Shirtliff, M. E.; Stoodley, P.; Webb, J. S.; Zaat, S. A. J.; Goeres, D. M.; Acta 2008, 608, 2−29. Azevedo, N. F. Biofilm 2020, 2, No. 100010. (59) Roda, A. Chemiluminescence and Bioluminescence; Roda, A., Ed.; (28) Prata, J. C.; Venâncio, C.; da Costa, J. P.; Lopes, I.; Duarte, A. Royal Society of Chemistry: Cambridge, 2010. https://doi.org/10. C.; Rocha-Santos, T. Mar. Pollut. Bull. 2021, 170, No. 112615. 1039/9781849732024. (29) Lage, O. M.; Ramos, M. C.; Calisto, R.; Almeida, E.; (60) Visick, K. L.; Foster, J.; Doino, J.; McFall-Ngai, M.; Ruby, E. G. Vasconcelos, V.; Vicente, F. Mar. Drugs 2018, 16, No. 279. J. Bacteriol. 2000, 182, 4578−4586. (30) Dincer, C.; Bruch, R.; Kling, A.; Dittrich, P. S.; Urban, G. A. (61) Baldwin, T. O.; Devine, J. H.; Heckel, R. C.; Lin, J. W.; Shadel, Trends Biotechnol. 2017, 35, 728−742. G. S. J. Biolumin. Chemilumin. 1989, 4, 326−341. (31) Visconti, P.; de Fazio, R.; Costantini, P.; Miccoli, S.; Cafagna, (62) Bergua, J. F.; Á lvarez-Diduk, R.; Hu, L.; Hassan, A. H. A.; D. J. Commun. Software Syst. 2019, 15, 89−100. Merkoçi, A. J. Hazard. Mater. 2021, 406, No. 124434. 1284 https://doi.org/10.1021/acs.analchem.1c04491 Anal. Chem. 2022, 94, 1271−1285 Analytical Chemistry pubs.acs.org/ac Article (63) Environmental Bio-Detection Products. Aliivibrio Fischeri Toxicity Tests; Environmental Bio-Detection Products Inc., 2021. (64) Jarque, S.; Masner, P.; Klánová, J.; Prokeš, R.; Bláha, L. Front. Microbiol. 2016, 7, No. 1738. (65) Froehner, K.; Backhaus, T.; Grimme, L. H. Chemosphere 2000, 40, 821−828. (66) Pubchem. Pentachlorophenol; Pubchem, 2021. (67) Bettinetti, R.; Kopp-Schneider, A.; Vignati, D. A. L. Sci. Total Environ. 2018, 613−614, 39−45. (68) Igbinosa, E. O.; Odjadjare, E. E.; Chigor, V. N.; Igbinosa, I. H.; Emoghene, A. O.; Ekhaise, F. O.; Igiehon, N. O.; Idemudia, O. G. Sci. World J. 2013, No. 460215. (69) Froehner, K.; Meyer, W.; Grimme, L. H. Chemosphere 2002, 46, 987−997. (70) Mortimer, M.; Kasemets, K.; Heinlaan, M.; Kurvet, I.; Kahru, A. Toxicol. In Vitro 2008, 22, 1412−1417. (71) Liu, J.; Morales-Narváez, E.; Orozco, J.; Vincent, T.; Zhong, G.; Merkoçi, A. Nano Res. 2017, 11, 114−125. (72) Ventura, F. F.; Mendes, L. F.; Oliveira, A. G.; Bazito, R. C.; Bechara, E. J. H.; Freire, R. S.; Stevani, C. V. Environ. Toxicol. Chem. 2020, 39, 1558−1565. (73) World Health Organization. Pentachlorophenol in Drinking- Water; World Health Organization, 2021. (74) Wiegand, I.; Hilpert, K.; Hancock, R. E. W. Nat. Protoc. 2008, 3, 163−175. (75) Reller, L. B.; Weinstein, M.; Jorgensen, J. H.; Ferraro, M. J. Clin. Infect. Dis. 2009, 49, 1749−1755. (76) Balouiri, M.; Sadiki, M.; Ibnsouda, S. K. J. Pharm. Anal. 2016, 6, 71−79. (77) Benkova, M.; Soukup, O.; Marek, J. J. Appl. Microbiol. 2020, 129, 806−822. (78) Ceylan Koydemir, H.; Rajpal, S.; Gumustekin, E.; Karinca, D.; Liang, K.; Göröcs, Z.; Tseng, D.; Ozcan, A. Sci. Rep. 2019, 9, No. 19901. (79) World Health Organization. Antibiotic Resistance; World Health Organization, 2021. (80) Wolflabs. Microplate Readers, 2021. https://www.wolflabs.co. Recommended by ACS uk/laboratory-products/microplate-readers. Characterizing a New Fluorescent Protein for a Low Limit of Detection Sensing in the Cell-Free System Caroline E. Copeland, Yong-Chan Kwon, et al. JULY 19, 2022 ACS SYNTHETIC BIOLOGY READ Easy and Rapid Approach to Obtaining the Binding Affinity of Biomolecular Interactions Based on the Deep Learning Boost Ying-Feng Chang, Chao-Sung Lai, et al. JULY 14, 2022 ANALYTICAL CHEMISTRY READ What Digital Immunoassays Can Learn from Ambient Analyte Theory: A Perspective Hans H. Gorris and Tero Soukka APRIL 11, 2022 ANALYTICAL CHEMISTRY READ Development of an Innovative Quantification Assay Based on Aptamer Sandwich and Isothermal Dumbbell Exponential Amplification Mathilde Aubret, Arnaud Buhot, et al. FEBRUARY 10, 2022 ANALYTICAL CHEMISTRY READ Get More Suggestions > 1285 https://doi.org/10.1021/acs.analchem.1c04491 Anal. Chem. 2022, 94, 1271−1285