Artificial Intelligence in Point-of-Care Biosensing: Challenges and Opportunities.

The integration of artificial intelligence (AI) into point-of-care (POC) biosensing has the potential to revolutionize diagnostic methodologies by offering rapid, accurate, and accessible health assessment directly at the patient level. This review paper explores the transformative impact of AI technologies on POC biosensing, emphasizing recent computational advancements, ongoing challenges, and future prospects in the field. We provide an overview of core biosensing technologies and their use at the POC, highlighting ongoing issues and challenges that may be solved with AI. We follow with an overview of AI methodologies that can be applied to biosensing, including machine learning algorithms, neural networks, and data processing frameworks that facilitate real-time analytical decision-making. We explore the applications of AI at each stage of the biosensor development process, highlighting the diverse opportunities beyond simple data analysis procedures. We include a thorough analysis of outstanding challenges in the field of AI-assisted biosensing, focusing on the technical and ethical challenges regarding the widespread adoption of these technologies, such as data security, algorithmic bias, and regulatory compliance. Through this review, we aim to emphasize the role of AI in advancing POC biosensing and inform researchers, clinicians, and policymakers about the potential of these technologies in reshaping global healthcare landscapes.

Apta FastZ: An Algorithm for the Rapid Identification of Aptamers with Defined Binding Affinities.

Real-time biomolecular monitoring requires biosensors based on affinity reagents, such as aptamers, with moderate to low affinities for the best binding dynamics and signal gain. We recently reported Pro-SELEX, an approach that utilizes parallelized SELEX and high-content bioinformatics for the selection of aptamers with predefined binding affinities. The Pro-SELEX pipeline relies on an algorithm, termed AptaZ, that can predict the binding affinities of selected aptamers. The original AptaZ algorithm is computationally complex and slows the overall throughput of Pro-SELEX. Here, we present Apta FastZ, a rapid equivalent of AptaZ. The Apta FastZ algorithm considers the spare nature of the sequences from SELEX and is coded to avoid unnecessary comparison between sequences. As a result, Apta FastZ achieved a 10 to 40-fold faster computing speed compared to the original AptaZ algorithm while maintaining identical outcomes, allowing the bioinformatics to be completed within 1-10 h for large-scale data sets. We further validated the affinity of myeloperoxidase aptamers predicted by Apta FastZ by experiments and observed a high level of linear correlation between predicted scores and measured affinities. Taken together, the implementation of Apta FastZ could greatly accelerate the current Pro-SELEX workflow, allowing customized aptamers to be discovered within 3 days using preselected DNA libraries.

A Colorimetric Biosensing Platform with Aptamers, Rolling Circle Amplification and Urease-Mediated Litmus Test.

Here we report on an ultra-sensitive colorimetric sensing platform that takes advantage of both the strong amplification power of rolling circle amplification (RCA) and the high efficiency of a simple urease-mediated litmus test. The presence of a target triggers the RCA reaction, and urease-labelled DNA can hybridize to the biotinylated RCA products and be immobilized onto streptavidin-coated magnetic beads. The urease-laden beads are then used to hydrolyze urea, leading to an increase in pH that can be detected by a simple litmus test. We show this sensing platform can be easily integrated with aptamers for sensing diverse targets via the detection of human thrombin and platelet-derived growth factor (PDGF) utilizing structure-switching aptamers as well as SARS-CoV-2 in human saliva using a spike-binding trimeric DNA aptamer. Furthermore, we demonstrate that this colorimetric sensing platform can be integrated into a simple paper-based device for sensing applications.

An RNA-Cleaving DNAzyme That Requires an Organic Solvent to Function.

Engineering functional nucleic acids that are active under unusual conditions will not only reveal their hidden abilities but also lay the groundwork for pursuing them for unique applications. Although many DNAzymes have been derived to catalyze diverse chemical reactions in aqueous solutions, no prior study has been set up to purposely derive DNAzymes that require an organic solvent to function. Herein, we utilized in vitro selection to isolate RNA-cleaving DNAzymes from a random-sequence DNA pool that were "compelled" to accept 35 % dimethyl sulfoxide (DMSO) as a cosolvent, via counter selection in a purely aqueous solution followed by positive selection in the same solution containing 35 % DMSO. This experiment led to the discovery of a new DNAzyme that requires 35 % DMSO for its catalytic activity and exhibits drastically reduced activity without DMSO. This DNAzyme also requires divalent metal ions for catalysis, and its activity is enhanced by monovalent ions. A minimized, more efficient DNAzyme was also derived. This work demonstrates that highly functional, organic solvent-dependent DNAzymes can be isolated from random-sequence DNA libraries via forced in vitro selection, thus expanding the capability and potential utility of catalytic DNA.

Biomolecular sensors for advanced physiological monitoring.

Body-based biomolecular sensing systems, including wearable, implantable and consumable sensors allow comprehensive health-related monitoring. Glucose sensors have long dominated wearable bioanalysis applications owing to their robust continuous detection of glucose, which has not yet been achieved for other biomarkers. However, access to diverse biological fluids and the development of reagentless sensing approaches may enable the design of body-based sensing systems for various analytes. Importantly, enhancing the selectivity and sensitivity of biomolecular sensors is essential for biomarker detection in complex physiological conditions. In this Review, we discuss approaches for the signal amplification of biomolecular sensors, including techniques to overcome Debye and mass transport limitations, and selectivity improvement, such as the integration of artificial affinity recognition elements. We highlight reagentless sensing approaches that can enable sequential real-time measurements, for example, the implementation of thin-film transistors in wearable devices. In addition to sensor construction, careful consideration of physical, psychological and security concerns related to body-based sensor integration is required to ensure that the transition from the laboratory to the human body is as seamless as possible.

A high-dimensional microfluidic approach for selection of aptamers with programmable binding affinities.

Aptamers are being applied as affinity reagents in analytical applications owing to their high stability, compact size and amenability to chemical modification. Generating aptamers with different binding affinities is desirable, but systematic evolution of ligands by exponential enrichment (SELEX), the standard for aptamer generation, is unable to quantitatively produce aptamers with desired binding affinities and requires multiple rounds of selection to eliminate false-positive hits. Here we introduce Pro-SELEX, an approach for the rapid discovery of aptamers with precisely defined binding affinities that combines efficient particle display, high-performance microfluidic sorting and high-content bioinformatics. Using the Pro-SELEX workflow, we were able to investigate the binding performance of individual aptamer candidates under different selective pressures in a single round of selection. Using human myeloperoxidase as a target, we demonstrate that aptamers with dissociation constants spanning a 20-fold range of affinities can be identified within one round of Pro-SELEX.

Nucleic Acid Enzyme-Activated CRISPR-Cas12a With Circular CRISPR RNA for Biosensing.

clustered regularly interspaced short palindromic repeats (CRISPR)-Cas systems are increasingly used in biosensor development. However, directly translating recognition events for non-nucleic acid targets by CRISPR into effective measurable signals represents an important ongoing challenge. Herein, it is hypothesized and confirmed that CRISPR RNAs (crRNAs) in a circular topology efficiently render Cas12a incapable of both site-specific double-stranded DNA cutting and nonspecific single-stranded DNA trans cleavage. Importantly, it is shown that nucleic acid enzymes (NAzymes) with RNA-cleaving activity can linearize the circular crRNAs, activating CRISPR-Cas12a functions. Using ligand-responsive ribozymes and DNAzymes as molecular recognition elements, it is demonstrated that target-triggered linearization of circular crRNAs offers great versatility for biosensing. This strategy is termed as "NAzyme-Activated CRISPR-Cas12a with Circular CRISPR RNA (NA3C)." Use of NA3C for clinical evaluation of urinary tract infections using an Escherichia coli-responsive RNA-cleaving DNAzyme to test 40 patient urine samples, providing a diagnostic sensitivity of 100% and specificity of 90%, is further demonstrated.

In Vitro Selection and Characterization of a DNAzyme Probe for Diverse Pathogenic Strains of Clostridium difficile.

Clostridium difficile frequently causes an infectious disease known as Clostridium difficile infection (CDI), and there is an urgent need for the development of more effective rapid diagnostic tests for CDI. Previously we have developed an RNA-cleaving fluorogenic DNAzyme (RFD) probe, named RFD-CD1, that is capable of detecting a specific strain of C. difficile but is too specific to recognize other pathogenic C. difficile strains. To overcome this issue, herein we report RFD-CD2, another RFD that is not only highly specific to C. difficile but also capable of recognizing diverse pathogenic C. difficile strains. Extensive sequence and structure characterization establishes a pseudoknot structure and a significantly minimized sequence for RFD-CD2. As a fluorescent sensor, RFD-CD2 can detect C. difficile at a concentration as low as 100 CFU/mL, thus making this DNAzyme an attractive molecular probe for rapid diagnosis of CDI caused by diverse strains of C. difficile.

Monitoring Cardiac Biomarkers with Aptamer-Based Molecular Pendulum Sensors.

Reagent-free electronic biosensors capable of analyzing disease markers directly in unprocessed body fluids will enable the development of simple & affordable devices for personalized healthcare monitoring. Here we report a powerful and versatile nucleic acid-based reagent-free electronic sensing system. The signal transduction is based on the kinetics of an electrode-tethered molecular pendulum-a rigid double stranded DNA with one of the strands displaying an analyte-binding aptamer and the other featuring a redox probe-that exhibits field-induced transport modulated by receptor occupancy. Using chronoamperometry, which enables the sensor to circumvent the conventional Debye length limitation, the binding of an analyte can be monitored as these species increase the hydrodynamic drag. The sensing platform demonstrates a low femtomolar quantification limit and minimal cross-reactivity in analyzing cardiac biomarkers in whole blood collected from patients with chronic heart failure.

Engineering a Ligase Binding DNA Aptamer into a Templating DNA Scaffold to Guide the Selective Synthesis of Circular DNAzymes and DNA Aptamers.

Functional nucleic acids (FNAs), such as DNAzymes and DNA aptamers, can be engineered into circular forms for improved performance. Circular FNAs are promising candidates for bioanalytical and biomedical applications due to their intriguing properties of enhanced biological stability and compatibility with rolling circle amplification. They are typically made from linear single-stranded (ss) DNA molecules via ligase-mediated ligation. However, it remains a great challenge to synthesize circular ssDNA molecules in high yield due to inherent side reactions where two or more of the same ssDNA molecules are ligated. Herein, we present a strategy to overcome this issue by first using in vitro selection to search from a random-sequence DNA library a ligatable DNA aptamer that binds a DNA ligase and then by engineering this aptamer into a general-purpose templating DNA scaffold to guide the ligase to execute selective intramolecular circularization. We demonstrate the broad utility of this approach via the creation of several species of circular DNA molecules, including a circular DNAzyme sensor for a bacterium and a circular DNA aptamer sensor for a protein target with excellent detection sensitivity and specificity.

A DNA Barcode-Based Aptasensor Enables Rapid Testing of Porcine Epidemic Diarrhea Viruses in Swine Saliva Using Electrochemical Readout.

Pen-side testing of farm animals for infectious diseases is critical for preventing transmission in herds and providing timely intervention. However, most existing pathogen tests have to be conducted in centralized labs with sample-to-result times of 2-4 days. Herein we introduce a test that uses a dual-electrode electrochemical chip (DEE-Chip) and a barcode-releasing electroactive aptamer for rapid on-farm detection of porcine epidemic diarrhea viruses (PEDv). The sensor exploits inter-electrode spacing reduction and active field mediated transport to accelerate barcode movement from electroactive aptamers to the detection electrode, thus expediting assay operation. The test yielded a clinically relevant limit-of-detection of 6 nM (0.37 µg mL-1 ) in saliva-spiked PEDv samples. Clinical evaluation of this biosensor with 12 porcine saliva samples demonstrated a diagnostic sensitivity of 83 % and specificity of 100 % with a concordance value of 92 % at an analysis time of one hour.

Aptamers from random sequence space: Accomplishments, gaps and future considerations.

Aptamers are molecular recognition elements made of nucleic acids. Diverse synthetic aptamers have been discovered in a large number of SELEX experiments since 1990. This review begins with the analysis of these SELEX experiments by examining the range of targets as well as the affinity and specificity for these targets by DNA, RNA or modified nucleic acid aptamers generated from these experiments. This is followed by comparisons of synthetic aptamers with natural RNA aptamers from riboswitches and with some of the best naturally occurring protein based recognition elements for proteins and small molecules. These comparisons reveal the gaps between man-made aptamers and natural recognition elements. We then put forward a series of ideas for consideration by the aptamer community towards developing better aptamers to solve real world problems. These include performing aptamer selections with libraries containing larger random sequence domains to derive larger aptamers with more intricate structures, conducting more reselection experiments using mutagenized DNA libraries based on initially selected aptamer sequences to search for better members of an aptamer family, selecting aptamers that are programmed to recognize two or more different epitopes of the same target in order to build multivalent aptamers for increased affinity, expanding the effort of selecting aptamers using modified nucleic acids with enhanced chemical functionalities, innovating SELEX methods to drive for the selection of aptamers with truly outstanding affinity and specificity, and having terminal applications in mind when creating new aptamers so that they are highly functional in the environment where the applications are planned.

Functional Nucleic Acids for Pathogenic Bacteria Detection.

Pathogens have long presented a significant threat to human lives, and hence the rapid detection of infectious pathogens is vital for improving human health. Current detection methods lack the means to detect infectious pathogens in a simple, rapid, and reliable manner at the time and point of need. Functional nucleic acids (FNAs) have the potential to overcome these limitations by acting as key components for point-of-care (POC) biosensors due to their distinctive advantages that include high binding affinities and specificities, excellent chemical stability, ease of synthesis and modification, and compatibility with a variety of signal-amplification and signal-transduction mechanisms.This Account summarizes the work completed in our groups toward developing FNA-based biosensors for detecting bacteria. In vitro selection has led to the isolation of many RNA-cleaving fluorogenic DNAzymes (RFDs) and DNA aptamers that can recognize infectious pathogens, including Escherichia coli, Clostridium difficile, Helicobacter pylori, and Legionella pneumophila. In most cases, a "many-against-many" approach was employed using a DNA library against a crude cellular mixture of an infectious pathogen containing diverse biomarkers as the target to isolate RFDs, with combined counter and positive selections ensuring high specificity toward the desired target. This procedure allows for the isolation of pathogen-specific FNAs without first identifying a suitable biomarker. Multiple target-specific DNA aptamers, including anti-glutamate dehydrogenase (GDH) circular aptamers, anti-degraded toxin B aptamers, and anti-RNase HII aptamers, have also been isolated for the detection of bacteria such as Clostridium difficile. The isolated FNAs have been integrated into fluorescent, colorimetric, and electrochemical biosensors using various signal transduction mechanisms. Both simple-to-use paper-based analytical devices and hand-held electrical devices with integrated FNAs have been developed for POC applications. In addition, signal-amplification strategies, including DNA catenane enabled rolling circle amplification (RCA), DNAzyme feedback RCA, and an all-DNA amplification system using a four-way junction and catalytic hairpin assembly (CHA), have been designed and applied to these systems to further increase their detection sensitivity. The use of these FNA-based biosensors to detect pathogens directly in clinical samples, such as urine, blood, and stool, has now been demonstrated with an outstanding sensitivity of as low as 10 cells per milliliter, highlighting the tremendous potential of using FNA-based sensors in clinical applications. We further describe strategies to overcome the challenges of using FNA-based biosensors in clinical applications, including strategies to improve the stability of FNAs in biological samples and prevent their nonspecific degradation from nucleases and strategies to deal with issues such as signal loss caused by nonspecific binding and biofouling. Finally, the remaining roadblocks for employing FNA-based biosensors in clinical applications are discussed.

Rapid and Specific Imaging of Extracellular Signaling Molecule Adenosine Triphosphate with a Self-Phosphorylating DNAzyme.

Adenosine 5'-triphosphate (ATP) is a central extracellular signaling agent involved in various physiological and pathological processes. However, precise measurements of the temporal and spatial components of ATP dynamics are lacking due primarily to the limitations of available methods for ATP detection. Here, we report on the first effort to design a self-phosphorylating DNAzyme (SPDz) sensor for fluorescence imaging of ATP. In response to ATP, SPDz sensors exhibit subsecond response kinetics, extremely high specificity, and micromolar affinities. In particular, we demonstrate cell-surface-anchored SPDz sensors for fluorescence imaging of both stress-induced endogenous ATP release in astrocytes and mechanical stimulation-evoked ATP release at the single-cell level. We also validated their utility for visualizing the rapid dynamic properties of ATP signaling upon electrical stimulation in astrocytes. Thus, SPDz sensors are robust tools for monitoring ATP signaling underlying diverse cellular processes.

A DNA Nanoflower-Assisted Separation-Free Nucleic Acid Detection Platform with a Commercial Pregnancy Test Strip.

There is a constant drive for affordable point-of-care testing (POCT) technologies for the detection of infectious human diseases. Herein, we report a simple platform for DNA detection that takes advantage of four techniques: commercially available pregnancy test strips (PTS), amplicon generation via loop-mediated isothermal amplification (LAMP), toehold-mediated strand displacement, and noncovalent immobilization of DNA on paper surface with DNA nanoflowers. This simple, separation-free platform is highly specific, as demonstrated with the detection of rtL180M, a single-nucleotide polymorphism observed in hepatitis B virus (HBV) associated with antiviral drug resistance. It is very sensitive, capable of detecting the targeted mutation at 2 copies µL-1 . It is able to correctly identify the unmutated and rtL180M genome types of HBV in clinical samples. Given its wide adaptability, we expect this platform can be easily modified for the detection of genetic variations associated with various pathogens and human diseases.

Integrating programmable DNAzymes with electrical readout for rapid and culture-free bacterial detection using a handheld platform.

The detection and identification of bacteria currently rely on enrichment steps such as bacterial culture and nucleic acid amplification to increase the concentration of target analytes. These steps increase assay time, cost and complexity, making it difficult to realize a truly rapid point-of-care test. Here we report the development of an electrical assay that uses electroactive RNA-cleaving DNAzymes (e-RCDs) to identify specific bacterial targets and subsequently release a DNA barcode for transducing a signal onto an electrical chip. Integrating e-RCDs into a two-channel electrical chip with nanostructured electrodes provides the analytical sensitivity and specificity needed for clinical analysis. The e-RCD assay is capable of detecting 10 CFU (equivalent to 1,000 CFU ml-1) of Escherichia coli selectively from a panel containing multiple non-specific bacterial species. Clinical evaluation of this assay using 41 patient urine samples demonstrated a diagnostic sensitivity of 100% and specificity of 78% at an analysis time of less than one hour compared with the several hours needed for currently used culture-based methods.

Detection of SARS-CoV-2 Viral Particles Using Direct, Reagent-Free Electrochemical Sensing.

The development of new methods for direct viral detection using streamlined and ideally reagent-free assays is a timely and important, but challenging, problem. The challenge of combatting the COVID-19 pandemic has been exacerbated by the lack of rapid and effective methods to identify viral pathogens like SARS-CoV-2 on-demand. Existing gold standard nucleic acid-based approaches require enzymatic amplification to achieve clinically relevant levels of sensitivity and are not typically used outside of a laboratory setting. Here, we report reagent-free viral sensing that directly reads out the presence of viral particles in 5 minutes using only a sensor-modified electrode chip. The approach relies on a class of electrode-tethered sensors bearing an analyte-binding antibody displayed on a negatively charged DNA linker that also features a tethered redox probe. When a positive potential is applied, the sensor is transported to the electrode surface. Using chronoamperometry, the presence of viral particles and proteins can be detected as these species increase the hydrodynamic drag on the sensor. This report is the first virus-detecting assay that uses the kinetic response of a probe/virus complex to analyze the complexation state of the antibody. We demonstrate the performance of this sensing approach as a means to detect, within 5 min, the presence of the SARS-CoV-2 virus and its associated spike protein in test samples and in unprocessed patient saliva.

A Highly Specific DNA Aptamer for RNase H2 from Clostridium difficile.

Molecular recognition elements with high specificity are of great importance for the study of molecular interactions, accurate diagnostics, drug design, and personalized medicine. Herein, a highly specific DNA aptamer for RNase H2 from Clostridium difficile (C. difficile) was generated by SELEX and minimized to 40 nucleotides. The aptamer exhibits a dissociation constant (Kd) of 1.8 ± 0.5 nM and an inhibition constant (IC50) of 7.1 ± 0.6 nM for C. difficile RNase H2, both of which are 2 orders of magnitude better for the same enzyme from other control bacteria. The fluorescent version of the aptamer can distinguish C. difficile from several other control bacteria in a cell lysate assay. This work demonstrates that a ubiquitous protein like RNase H2 can still be used as the target for the development of highly specific aptamers and the combination of the protein and the aptamer can achieve the recognition specificity needed for a diagnostic test and drug development.

Evolution of a highly functional circular DNA aptamer in serum.

Circular DNA aptamers are powerful candidates for therapeutic applications given their dramatically enhanced biostability. Herein we report the first effort to evolve circular DNA aptamers that bind a human protein directly in serum, a complex biofluid. Targeting human thrombin, this strategy has led to the discovery of a circular aptamer, named CTBA4T-B1, that exhibits very high binding affinity (with a dissociation constant of 19 pM), excellent anticoagulation activity (with the half maximal inhibitory concentration of 90 pM) and high stability (with a half-life of 8 h) in human serum, highlighting the advantage of performing aptamer selection directly in the environment where the application is intended. CTBA4T-B1 is predicted to adopt a unique structural fold with a central two-tiered guanine quadruplex capped by two long stem-loops. This structural arrangement differs from all known thrombin binding linear DNA aptamers, demonstrating the added advantage of evolving aptamers from circular DNA libraries. The method described here permits the derivation of circular DNA aptamers directly in biological fluids and could potentially be adapted to generate other types of aptamers for therapeutic applications.

Ribbon of DNA Lattice on Gold Nanoparticles for Selective Drug Delivery to Cancer Cells.

Herein, we report on the design of a programmable DNA ribbon using long-chain DNA molecules with a user-defined repetitive padlock sequence. The DNA ribbon can be further combined with gold nanoparticles (AuNPs) to create a composite nanomaterial that contains an AuNP core and a high-density DNA crown carrying a cancer-cell-targeting DNA aptamer, a fluorescent tag for location tracking, and a cell-killing drug. This composite material can be efficiently internalized by cancer cells and its cellular location can be tracked by fluorescence imaging. The system offers several attractive characteristics, including simple design, tunable DNA crown, high drug-loading capacity, selective cell targeting, and pH-sensitive drug release. These features make such a material a promising therapeutic agent.