Fast detection of bacterial gut pathogens on miniaturized devices: an overview

ABSTRACT

Introduction

Gut microbes pose challenges like colon inflammation, deadly diarrhea, antimicrobial resistance dissemination, and chronic disease onset. Development of early, rapid and specific diagnosis tools is essential for improving infection control. Point-of-care testing (POCT) systems offer rapid, sensitive, low-cost and sample-to-answer methods for microbe detection from various clinical and environmental samples, bringing the advantages of portability, automation, and simple operation.

Areas covered

Rapid detection of gut microbes can be done using a wide array of techniques including biosensors, immunological assays, electrochemical impedance spectroscopy, mass spectrometry and molecular biology. Inclusion of Internet of Things, machine learning, and smartphone-based point-of-care applications is an important aspect of POCT. In this review, the authors discuss various fast diagnostic platforms for gut pathogens and their main challenges.

Expert opinion

Developing effective assays for microbe detection can be complex. Assay design must consider factors like target selection, real-time and multiplex detection, sample type, reagent stability and storage, primer/probe design, and optimizing reaction conditions for accuracy and sensitivity. Mitigating these challenges requires interdisciplinary collaboration among scientists, clinicians, engineers, and industry partners. Future efforts are essential to enhance sensitivity, specificity, and versatility of POCT systems for gut microbe detection and quantification, advancing infectious disease diagnostics and management.

KEYWORDS:

• Point-of-care
• gut pathogens
• POCT
• microfluidics
• isothermal amplification

1. Introduction

Diarrheal illness is a global health concern causing over 1.6 million deaths annually worldwide [1]. Diarrhea is a nonspecific symptom that can be triggered by a wide range of factors, including gastrointestinal tract pathogens (Clostridium botulinum, Salmonella spp., Shigella spp., Escherichia coli, Campylobacter jejuni, Vibrio spp., Listeria monocytogenes, etc), other illnesses (inflammatory bowel disease), and medications. This makes it challenging to pinpoint the exact cause of diarrhea without thorough diagnostic testing. Due to the potential for infectious causes of diarrhea, patients are often placed in single-occupancy rooms with infection control measures in place to prevent the spread of illness. Moreover, the COVID-19 pandemic has added pressure on the already limited resources, exacerbating the challenges related to managing patients with diarrhea in hospital settings.

The ‘gold standard method’ for diagnosing enteric infections relies on the combination of bacterial culture and confirmatory biochemical (triple sugar iron test, urease, citrate, etc) and serological tests [2]. While traditional methods have been widely used and accepted, they can be time-consuming, often require trained personnel and may not provide rapid results. Centralized, classical laboratory testing of patient stool samples has a long turnaround time for results, leading to delays in diagnosing the underlying cause of diarrhea in patients.

In recent years, advances in molecular diagnostics, such as polymerase chain reaction (PCR) and other nucleic acid (NA)-based techniques, have emerged as alternative methods for diagnosing enteric infections [3], providing quicker results and greater sensitivity in detecting specific bacterial pathogens. However, current techniques such as PCR and enzyme-linked immunosorbent assay (ELISA) for pathogen detection have limitations due to multiple sample preparation steps and time consumption. There is a need to miniaturize and integrate these techniques into a single point-of-care device (POC). POC devices offer advantages in terms of size, cost, sensitivity, accuracy, and the ability to detect multiple analytes in a single sample.

The main purpose of POC devices is to provide diagnostic capabilities directly at the point where care is being administered, eliminating the need for patients to visit a clinical laboratory for testing. An ideal POC device should allow patients to conduct tests at home with minimal supervision, providing quick results. These devices are expected to be affordable, portable, and user-friendly. POC devices are often designed by integrating different technologies, such as biosensors, plasmonic and/or microfluidic systems, along with electrochemical or optical readout mechanisms [4]. These technologies are combined into a compact platform to enable real-time detection of pathogens. POC systems are particularly promising in resource-limited settings. They adhere to the REASSURED criteria, which stands for Real-time connectivity, Ease of specimen collection, affordable, specific, sensitive, user-friendly, rapid, equipment-free analysis, and delivery to remote areas. These criteria were established by the World Health Organization (WHO) to guide the development of diagnostic tools suitable for economically underdeveloped regions, ultimately improving global healthcare quality [5,6].

This review will comprehensively and critically discuss the various types of gut pathogens’ POC diagnostic platforms (Figure 1) underlining the potential impact of micro- and nanotechnology on developing such platforms for the simultaneous screening of multiple pathogens.

Figure 1. Point-of-care detection methods. Traditional methods for detection of gut pathogens include bacterial cultivation, PCR, biochemistry tests and microscopy analysis and generally take from 24 to 72 hours from sample collection to diagnostic. Point-of-care detection uses immunological tests, molecular biology methods as well as biosensors (colorimetric, electrochemical, fluorometric) leading to a fast diagnostic (generally ranging from 10 minutes to 3 hours). Created using Biorender.com.

2. POCT detection methods

2.1. Immunological detection

Immunological detection is based the binding of specific antibodies (polyclonal antibody or monoclonal antibody) to their specific antigens. For the detection of gut pathogens, several immunological methods can be used including enzyme-linked immunosorbent assay (ELISA), lateral flow immunoassay (LFIA), latex agglutination, immunofluorescence, Western blot, immunomagnetic separation [7]. The limit of detection, techniques, specificity, and detection speed for immunological diagnostic of foodborne bacterial pathogens are presented in Table 1.

Table 1. Examples of immunological methods used for detecting foodborne pathogens.

Pathogen	Detection technique	Sample	LOD	Assay duration	Ref.
Salmonella enteritidis	Nanobody based-ELISA	Spiked milk samples	1.4×10^5 CFU/mL 6 CFU/mL in pre-enriched samples	10 h pre-enrichment 3.5 h ELISA detection	[8]
Salmonella spp.	Label-free Immunochromatographic strips	Poultry samples	10^2 CFU/mL of S.pullorum 10^3 CFU/mL of S.gallinarum 10^4 CFU/mL of S.enterica		[9]
Escherichia coli O157:H7	Paper-based ELISA	Spiked beef sample	10^4 CFU/mL	<3 h	[10]
	Ag enhanced LFIA	Spiked orange juice, milk, & tap water	2 × 10^3 CFU/mL	30 min	[11]
	Antibody coated microspheres	Bacterial suspension	49.1 CFU/μL	1.5 h	[12]
Vibrio parahaemolyticus	magnetic nanobead-based immunoseparation and quantum dot immunofluorescence	Spiked clams	100 CFU/mL	4 h	[13]
	Immunochromatographic strips labelled with nanomaterials	Spiked seafood	5.5 × 105 CFU/mL		[14]

ELISA has certain limitations including relatively low sensitivity and time consumption. LFIA holds immense potential for on-site instant diagnosis due to its convenience, speed, and cost-effectiveness. However, there are challenges when applying LFIA to bacterial detection, including issues related to sensitivity, quantitative information, and compatibility with complex matrices [15]. These limitations stem from the use of gold nanoparticles (AuNPs) as reporters in traditional LFIA setups. To address this caveat, Wen et al [16] recently integrated Au-Fe₃O₄ multifunctional NPs with LFIA as promising alternative reporters for colorimetric and photothermal dual detection of Salmonella typhimurium. The dual-mode LFIA method represents an advancement over traditional Au-based LFIA, tackling the limitations of traditional LFIA, particularly by significantly enhancing the detection sensitivity by two orders of magnitude. Recently, a silver enhancement strategy was developed by Bazsefidpar et al (2023) as an accessible, fast and cost-effective approach for LFIA signal amplification improve the limit of detection (LOD) of E. coli O157:H7 from 2 × 10⁶ CFU/mL to 2 × 10³ CFU/mL [11]. Traditionally built commercial LFA devices exhibit several limitations, including lower specificity and poorer sensitivity compared to laboratory tests, thus maintaining the requirement for diagnostic confirmation by more complex laboratory tests.

2.2. Nucleic acid testing

2.2.1. Nucleic acid amplification

Immunological methods are based on the detection of specific antibodies that interact with antigens present on the microbial cells and are relatively simple to operate but might not provide the same level of sensitivity and specificity as nucleic acid (NA) testing. The process of NA (DNA or RNA) detection for microorganisms involves several steps: pathogen capture, cellular lysis, nucleic acid extraction and purification as well as NA amplification and detection.

Classic methods, such as polymerase chain reaction (PCR), multiplex polymerase chain reaction (mPCR), quantitative real time PCR (qPCR), having great specificity with reliable and reproducible results, proved to be laborious with a need for expensive equipment. A stand-alone PCR reaction requires careful temperature consideration for each step that takes place during the process (denaturation, primer annealing, amplification), trained technicians, while duration that can vary depending on the type of protocol used [7]. Taking this drawback into consideration, mPCR has been adopted as a go to technique in detection of multiple pathogens simultaneously which proved more efficient considering the time and cost needed to develop and optimize the assay with the desired targets. Additionally, other benefits of using mPCR are noted alongside the capacity of coupling multiple pairs of primers in one reaction thus shortening the number of steps required to detect multiple pathogens and reducing the reagent costs making this process more affordable in the long run [17]. qPCR provides real time data on the state of amplification and by coupling this with the possibility of adding multiple probes which can be differentiated by the absorption and emission spectrum, the same reaction can provide more information regarding the desired target or multiple targets [18].

Since they rapidly and efficiently amplify specific DNA or RNA sequences at a constant temperature, typically between 37°C and 65°C, eliminating the need for complex thermal cycling, isothermal amplification methods are particularly suitable for POCT. The most common isothermal amplification methods include: Loop-mediated isothermal amplification (LAMP), Recombinase Polymerase Amplification (RPA), rolling circle amplification (RCA), helicase-dependent amplification (HDA), nucleic acid sequence-based amplification (NASBA), cross priming amplification (CPA), strand swat amplification (SDA) and recombinant enzyme assisted amplification assay (RAA) [19].

So far, LAMP is the most frequently used isothermal amplification technique [20]. The NA amplification takes place at a constant temperature, typically around 60°C to 65°C, eliminating the need for a thermal cycler, making LAMP particularly useful in settings with limited equipment [21]. This method has good sensitivity, and it was reported to identify asymptomatic carriers of Vibrio cholerae, therefore being a powerful tool for public health management [22]. LAMP can be effectively combined with other technological advancements to enhance its capabilities for detecting target organisms [23,24].

Invasive Non-Typhoidal Salmonella (iNTS) is caused by non-typhoidal Salmonella strains that invade the bloodstream and cause severe illness. Symptoms of clinical iNTS blood infections can appear even when the bacterial burden in the blood is as low as 10 CFU/mL. Recently, a combined approach using specific antibody-coupled magnetic bead-based pathogen concentration cand LAMP was employed to detect Salmonella Typhimurium with a LOD of 14 CFU/mL [25].

RPA is an isothermal DNA amplification method, which involves the coupling of recombinase-driven primers with strand-displacement DNA synthesis [26]. RPA can be integrated with various rapid detection methods to create POCT assays for detection of gut pathogens. A study led by Ma et al (2020) developed a multiplex RPA approach combined with colloidal gold-labeled lateral flow dipsticks to simultaneously detect multiple foodborne pathogens (V. parahaemolyticus , Staphylococcus aureus, and Salmonella Enteritidis) in seafood samples with a detection limit around 10¹ CFU/mL [27].

CRISPR-based detection leverage the specificity of Cas effector proteins, such as Cas9, Cas12, and Cas13, to detect specific NA sequences. Cas proteins, originally known for their role in gene editing, are capable of binding to target DNA or RNA sequences with high specificity, thanks to the guiding mechanism provided by guide RNA (gRNA). The binding of the Cas protein to the target sequence triggers a response that can be measured by fluorescence, colorimetry, potentiometry or lateral flow assays [28]. One of the main advantages of CRISPR-based sensing is its ability to detect extremely low concentrations of analytes, surpassing the sensitivity of traditional detection methods. Identifying pathogenic organisms and their genotypic variants can inform tailored treatment strategies, enhancing the efficiency of medical interventions. The CRISPR system’s inherent flexibility allows for easy customization by changing the guide RNA (gRNA) sequence. This adaptability enables the detection of a wide range of NA sequences, making it a versatile tool for identifying a wide range of pathogens.

CRISPR has been used for diagnostic of Helicobacter pylori, a bacterium associated with ulcer and gastric cancer which is routinely identified using methods such as endoscopy, urea breath test, rapid urease test, antigen test, serology and microbial culture [29]. By combining CRISPR-Cas12a technology with RPA and a gold nanoparticle-based lateral flow biosensor, Qiu et al described a sensitive and rapid detection method for H. pylori directly from stool samples [30].

The majority of CRISPR-Cas sensors developed so far have been designed to detect NAs, including DNA and RNA sequences. While NA detection is important, there’s a demand for CRISPR-based sensors that can detect a wider range of targets including proteins, small molecules, and other biomolecules of interest. Examples of molecular methods used for the detection of foodborne pathogens are listed in Table 2.

Table 2. Examples of molecular methods used for detecting foodborne pathogens.

Target pathogen	Detection method	Type of sample	LOD	Assay duration	Ref.
Listeria monocytogenes, Escherichia coli, Staphylococcus aureus, Salmonella enterica	mPCR	spiked lettuce and minced meat	10^3 CFU/mL	120 min - mPCR 80 min - agarose gel	[17]
Vibrio parahaemolyticus, Listeria monocytogenes ATCC 19,115, Cronobacter sakazakii, Shigella flexneri, Pseudomonas putida, Escherichia coli O157:H7, Vibrio vulnificus, Vibrio alginolyticus		spiked aquatic food	10^4 CFU/mL	90 min-PCR	[31]
Enterococcu durans KCTC 13,289, Enterococcu faecalis KACC 11,859, Enterococcus faecium KACC 11,954, Enterococcu hirae KACC 16,328, Enterococcu lactis KACC 15,681, Enterococcus mundtii KACC 13,824	qPCR	spiked food samples	10^2 CFU/mL	80 min - PCR	[18]
Campylobacter jejuni, Campilobacter coli	LAMP	broiler production	50 CFU/mL	30 min for amplification	[32]
E. coli O157:H7	Bioluminescent - based LAMP	various food samples	5 CFU/mL	90–120 min	[33]
Listeria monocytogenes	Colorimetric lamp	Contaminated ready-to-eat-food	55 CFU/mL	30 min	[34]
Staphylococcus aureus, Pseudomonas aeruginosa	RPA-LFD	artificially spiked fresh milk	1 CFU/reaction S. aureus and 5 CFU/reaction P.aeruginosa	25 min	[35]
Yersinia enterocolitica	CRISPR/Cas12a-RPA-based system	artificially spiked raw pork meat	1.7 CFU/mL	45 min	[36]

2.2.2. Nucleic acid sequencing

The use of next generation sequencing (NGS) in POC is a topic of interest currently going in the industry thanks to the capacity of the devices to process the samples in a parallel manner. In spite of having major drawbacks, such as the time needed to prepare the library and a potentially need for additional enrichment steps, the technology is being adapted to detect a potentially large number of pathogens from a single sample thus making up for the time needed to prepare the samples for processing. Cha et al., 2023 [37] have shown that second generation sequencing platform (Illumina) and third generation sequencing platform (Oxford Nanopore, MinION) have reproducible results in gut microbiome profiling in newborns. The difference between generations of sequencers can be seen in the duration in which the results can be obtained and the length of reads produced by the sequencer, second generation being capable of managing a plethora of samples in parallel with small reads and a long time to process, compared to third generation where a smaller number of samples can be processed with greater length, at a faster pace with ready to go devices that are more compact and easier to transport [38].

Third generation sequencers (TGS) has been shown to be more in line with POC REASSURED criteria in spite of the time needed for the result to be obtained of about 8–12 h which correlates to the considerable amount of data produced in order to profile patient’s gut microbiota with high fidelity in identifying pathogens with no additional enrichment required [39,40]. The current research papers published specify the need to further adapt TGS as a way to identify as many pathogenic strains in infected patients. Grumaz et al., 2020 [41]. managed to optimize a workflow capable of profiling the microbiota of septic patients in a manner of 6 h by obtaining the cell-free DNA from blood sample. Another research team managed to create a database of over 500.000 general markers for microscopic eukaryotes and by combining that with a shotgun sequencing procedure, they have obtained a workflow capable of profiling non-bacterial GI organisms which can be transferred through fecal matter, thus further proving that TGS is a very promising technology which should be adapted further for POC [42].

2.3. Biosensors

The human microbiome could be a potential key to the understanding of human health and improving the diagnosis process and treatment of various infections. A large spectrum of microorganisms that includes bacteria, eukaryotes, archaea, and viruses could colonize the human body. Activity and interaction between them and the host could affect health [43]. A large variety of biosensors have been used to detect multiple targets of interest from the various samples such as serum, plasma, or feces [44,45]. Biosensors are also involved in the detection process of contaminants or diverse allergens in food samples [46].

Biosensors are utilized as an analysis instrument made up of two components: the biosensor itself and the transducer. Pathogens can be identified first of all through their antigens or antibodies, sensitive enzymes, and their gene sequence [47–49]. The tested materials suffer biological reactions when they are in contact with pathogen-derived molecules. Next, signal transducers can transform these interactions into quantifiable electric signals that are detected and read by amplifiers [50]. A large variety of biosensors can be classified by their functional principles, this classification can include: optical biosensors, enzymatic, electrochemical, physical, and mechanical ones [51] (Figure 2).

Figure 2. Representation of the various detection methods used in the process of gut pathogen identification and the most common infectious bacteria detected.Created using Biorender.com.

Two types of biosensors are used usually for the rapid detection of pathogenic agents that are foodborne: optical biosensors and electrochemical ones [52,53]. The capacity of optical biosensors to detect rapidly, also their sensitivity and specificity, make them extremely useful in the identification process of various pathogens [54]. So far, the most frequently used optical detection technologies are chemiluminescence, fluorescence, colorimetry, and SPR (Surface Plasmon Resonance) [55].

There is a constant demand for novel rapid detection methods with different principles and mechanisms to detect gut pathogens in different types of samples. For example, a new detection technique was developed using spot dye-based sensors to detect Vibrio load in shrimp culture farms. The spot dye-based sensor test kit involves only two simple steps and is designed to be user-friendly, not requiring specialized personnel. A selective media is added to the water sample to encourage Vibrio growth while inhibiting the growth of other bacteria and subsequently, a dye-based sensor is added to detect active Vibrio bacteria, with a positive result indicated by a change in color to red or pink. The method is sensitive, capable of detecting low concentrations of Vibrio species (at 10² CFU/mL) within six to seven h. The kit is designed for use in aquaculture farms, where quick detection of Vibrio species is crucial for disease management [56].

Even though all the examples given could lead to the detection of gut pathogens by only using biosensors, they rely primarily on complex pieces of equipment, sometimes having an incredibly high cost and therefore they are not usually portable. Besides that, the truly point-of-care devices should be sufficiently fast to offer results before the patient can leave the clinic. Although significant developments have been made in the integration process of microbiome analysis in the diagnosis setups, it still leaves much more to be improved.

To create a complete point-of-care biosensor, the full process, starting from the processing of the sample to the detection of the target in question, needs to be portable and fast, the preferred time allocated for the test should be no longer than 30 min. This means that all of the sample preparation steps (i.e. centrifuge, liquefy, and homogenization), also hydrolysis of the sample should be converted into portable platforms, together with the detection method [57]. More than that, the interpretation of the results by the physician should take place through an interface with the user or any other visible format, without necessitating costly and complex machinery. These obstacles should be overcome to be able to create analysis methods of gut pathogen/microbiome based on biosensors.

Biosensors could be used to generate reference data for the microbiome, because of their large capacity. With all of that in mind, the large quantity of data collected from an individual regarding his microbiome in one region of the body, let alone the large quantities of data that should be obtained from vast masses of population in different states of health and sickness, is discouraging.

To make matters worse, despite the identification of specific pathogenic strains, there is a large variability between individuals’ microbiomes, even when parameters like the diet of patients are controlled. That means, from a diagnostic point of view, the approach of the reference interval will not be sufficient to predict the state of health or sickness of an individual based on sets of data collected from similar patients [58]. A summary of different biosensors types, their LOD and structures is presented in Table 3.

Table 3. Detection of various gut bacteria with multiple types of biosensors.

Type	Biosensor Structure	Target analyte	LOD	Ref.
Resonance	Tungsten oxide-based sensor functionalized with vancomycin	S. aureus	80–100 CFU/mL	[59]
Resonance	AuNPs and MoS2-based LSPR sensor	Shigella	1.56 CFU/mL	[60]
Fluorescence	Quantum dot-based sensor	E. coli, L. monocytogenes, S.Typhimurium	102, 103, and 103 CFU/mL	[61]
Colorimetric	M13 phage-based sensor	E. Coli, P. aeruginosa, V. cholera, X. campestris	100 cells	[62]
Colorimetric	CRISPR-Cas12a and AuNPs based colorimetric sensor	Salmonella	1 CFU/mL	[63]
Colorimetric	Peptide aptamer-based MZI sensor	E. coli	11 CFU/mL	[64]
Piezoelectric	Antibody-AuNPs	Bifidobacterium bifidum	2.1 × 10^2 CFU/mL	[65]
Electrochemical	Molecularly imprinted polymer	TMAO(trimethylamine N-oxide) - metabolite	1–15 ppm	[66]
Electrochemical (Impedance)	OCMCS-Fe3O4 NPs	Campylobacter jejuni	103–107 CFU/mL	[67]
Electrochemical	Bismuth-immobilized carbon nanotube	Helicobacter pylori DNA	0.72–7.92 μg/mL	[68]

2.4. Mass spectrometry

An impartial and precise method that helps in the detection of pathogenic bacteria is mass spectrometry (MS). From the moment, this method first appeared, it evolved at a rapid pace into a new way to detect different types of bacteria [69,70]. This specific technique can be considered analytical, instrumental, and non-biochemical, which examines the characteristics of pathogens or a specific part of the proteins in their structure as a research subject. MS refers to the analyzing technique of the ions produced after ionization to identify and detect the target bacteria [71].

Mass spectrometry is most often mentioned in many multiplex detections from complex samples and presents high sensitivity and selectivity. In the biological field, specific MS techniques, like inductively coupled plasma mass spectrometry (ICP – MS), necessitate that the targets in question should be marked first with another molecule that could be detected. This obstacle arises from the fact that proteins have as main components hydrogen, carbon, oxygen, and nitrogen that are present in the solvents used and also in the air which unfortunately leads to background noise Even though there are multiple disadvantages to using ICP-MS, it has been proved time and time again that this technique also has many benefits, for example: it can detect albumin in urine samples [72] and polymorphism with a single nucleotide in serum [73]. However, other MS methods can identify known and unknown bacteria, without tagging molecules. These methods use ionization technology with a soft impact to obtain information regarding mass specters, which are necessary for protein analysis [74].

A new method to test the antimicrobial resistance of various bacterial strains was developed by Feucherolles et al. For the relevant foodborne pathogens, for example, Campylobacter coli and Campylobacter jejuni, protein mass spectra from matrix-assisted laser desorption/ionization time-of-flight mass spectrometry or MALDI-TOF MS were developed together with a method of prediction [71]. For example, matrix-assisted laser desorption/ionization-time-of-flight MS has been proven useful in the detection of biological war agents [75] and of various strains of E. coli [76].

Another study conducted by Li et al [77] describes a unique MALDI-TOF MS method specialized in the simultaneous detection of numerous bacteria by using mass tags induced surface engineering in combination with mass spectrometry in the same study. By using this method many types of bacteria can be identified simultaneously, without needing to collect entire libraries of microbial mass spectrum before the analysis procedure.

Dias et al [78] examined three essential oils and their potential active ingredients with beneficial effects against pathogens present in spoiled food. The techniques described by Dias et al were gas chromatography-mass spectrometry and gas chromatography with flame ionization. Mass spectrometry is considered to be the most efficient detection method of bacteria transmitted through food, offering accuracy and a fast detection time but also an easy operation, but many problems regarding the actual process of identification can occur [79]. It is necessary to improve the flow rate, spray voltage, but also temperature of the capillary, and other difficulties that have arisen in the detection process, to increase the sensitivity and reliability of the technology in the detection of gut pathogens and also of foodborne pathogenic bacteria. MS presents many limitations, such as the incapacity to make the distinction between closely related species due to their similarity. Furthermore, the wrong identification can appear when some of the members of a species complex are listed in the database, but others are not [80].

Similar species cannot be correctly identified because of a lack of sufficient spectra in the database. It is possible to get an incorrect species identification or no identification for that matter. A study conducted by Rychert et al [80], proved that a specific species, Trichophyton species, was always misidentified. Another example, in a study by Body et al, M. mucogenicum isolates were accurately identified, but a closely related organism, M. practicum wasn’t identified properly, because it was not in the database, and was most often misidentified as M. mucogenicum [81]. In many clinical cases, such errors can pose a significant threat to the well-being of patients. Various updates of the database and other created libraries can overcome this problem. Another alternative would be using backup methods, for example, sequencing can be effective as long as the issue is known.

Several emerging technologies have been discovered and used, such as multiplex real-time PCR combined with matrix-assisted laser desorption ionization-time of flight mass spectrometry (MALDI – TOF MS) method for clinical application [82]. This method can also be described as a scalable protocol with a suite of applications for quantitative and qualitative nucleic acid analysis, which provides flexible assay design, fast time results, and the ability to run from tens to thousands of samples daily. These features make it the ideal genetic analysis system for validation and fine-mapping studies in basic and translational research settings, it has also been applied in the field of microbial and virus detection [83,84]

With the help of the advantages presented by the system’s multiplex capability and high sample throughput, Zhang et al. [85] developed and validated a bacterial pathogen-MS panel (BP-MS) to screen simultaneously, 11 key bacterial pathogens associated with pneumonia and meningitis. Based on the sensitivity test, the LOD of the bacterial pathogen-mass spectrometry method is approximately 10 gene copies/reaction. This evidence suggests that the BP-MS method performs well in clarifying complicated polymicrobial etiologies.

MS techniques can allow fast detection but the equipment is lacking in terms of simplicity still being considered a high-cost method and also requires complex abilities to utilize [74]. Usage of MALDI-TOF MS in the clinical workspace has given us many opportunities to identify several microorganisms that had subsequent involvement in the process of diagnosis and of course, the reduction in the time necessary to prescribe an appropriate treatment.

3. Integrated ‘’Sample-to-answer’’ devices for gut pathogens

Currently, there is an important and practical research direction to improve sample processing efficiency in POCT devices. Integrated POC molecular diagnostic devices follow a ‘sample-to-answer’ approach, where the entire diagnostic workflow (from sample collection to result interpretation) is integrated into a single device. As shown in Figure 3a, these devices encompass various stages of the diagnostic process, including: sample collection, pathogen isolation/enrichment, NA purification/enrichment, NA amplification and signal transduction [86–88].

Figure 3. Integrated sample-to-answer POCT devices – a. Technical criteria for designing an integrated POCT device; b. Trends and challenges in POCT development and clinical implementation. Created using Biorender.com.

In lab-on-a-chip devices, the pathogen capture can be performed using different methods such as: microspheres [89], filters or membranes [90], magnetophoresis [91], dielectrophoresis [92], or acoustophoresis [93,94].

Yi et al described an integrated LOC system with a sample preparation step using magnetic beads for LAMP real-time quantitative detection of Salmonella in enriched pork meat samples [95]. Recently, Chen et al (2023) developed an integrated, origami-operated lab-on-a-chip device for the detection of Salmonella using RPA. After 6 h enrichment, the device could detect Salmonella within 30 min with an LOD of 6 CFU/g in lettuce, 9 CFU/g in chicken breast meat and 58 CFU/mL in milk [96]. Chen et al (2023) developed an easy-to-use, hybrid paper/polymer-based microfluidic device integrating paper-based DNA extraction, RPA, and lateral flow detection for C. jejuni detection in spiked chicken meat at concentrations ranging from 10¹ to 10² CFU/g after an enrichment of 5 to 10 h [97].

The FDA-cleared BIOFIRE integrated system stores all the necessary reagents for sample preparation, reverse transcription PCR, PCR, and detection in a freeze-dried format. During the diagnostic workflow, this system extracts and purifies all NAs from unprocessed clinical samples [98]. This commercially available POC system offers a range of clinical applications, including detection of gut pathogens and antimicrobial resistance genes.

Developing an integrated sample-to-answer for molecular diagnostic of infectious agents needs to overcome an important barrier related to sample preparation. Biological matrices, such as blood and fecal samples, are complex and heterogeneous, containing a wide range of different molecules, making it difficult to identify and isolate specific target pathogens. Particularly, stool is composed of a wide range of constituents, including genetic material from the host, commensal microorganisms, and potential pathogens, and this diversity can lead to complex mixtures that impact the performance of NAs amplification techniques.

Many rapid diagnostic methods rely on sample preparation steps to concentrate and isolate the target pathogens. However, samples with very low pathogen concentrations might have DNA levels that fall below the capabilities of common sample preparation techniques. For instance, magnetic bead-based methods could be employed for concentrating pathogen DNA [25]. These beads can then be separated from the sample using a magnetic field, effectively concentrating the pathogens. This concentration step enhances the chances of detecting low-abundance pathogens and can eliminate the need for additional enrichment steps in the diagnostic process. Future advances in this research field should focus on developing assays for direct use of raw clinical specimens (i.e. fecal sample) without the need for sample pre-treatment.

Detection of pathogens from food samples comes with several challenges including low concentration of target microorganisms and NAs, as well as the presence of various inhibitory substances that lead to false-negative results or reduced amplification efficiency. Different food sample matrices can exert specific effects on the amplification process. For example, the presence of calcium ions in milk can interact with DNA or RNA and inhibit polymerase activity [99]. Moreover, these matrices often require labor-intensive sample preparation steps like centrifugation, pH adjustment, serial dilution, or filtration to ensure accurate analysis.

4. Microbiome, IoT and AI in detection of gut pathogens

Current trends in POCT are the integration of wireless communication and machine learning (ML) [100] as well as study of the gut microbiome [101]. Usually, a general workflow of machine learning modeling includes several basic steps, among which we can mention: data acquisition (raw data obtained from sequencing, POC devices, or images), data preprocessing, data extraction and transformation, classification and grouping, modeling and prediction, data integration (if data from multiple sources are used), post-analysis and comparison with other data and methods. For microbiome engineering, classic classification techniques have still proven effective. Whether we are talking about clustering, classification, regression methods or data modeling and obtaining predictions, each one of the various techniques like decision trees, random forest, support vector machine (SVM), Bayesian networks, neural networks, genetic algorithms, fuzzy techniques, naïve bayes, hidden Markov model (HMM), k-nearest neighbors (kNN), logistic regression, or various hybrid techniques, comes to the aid of microbiome research.

Summaries of machine learning applications for gut microbiome studies, including data sets, features, and aims are presented in [102–104] while resources for applying ML, others from those already known (R, Matlab, Python), such as MetAML or MicroPheno software, can be found in [105].

Using machine learning, researchers intend to capture microbiome signatures to make the connection to impending diseases. In recent studies [106], the method called MB-SupCon (Microbiome-based Supervised Contrastive Learning Framework) provides highly accurate predictions about inflammatory bowel disease. The model which maximizes the similarities between microbiome and metabolome samples, was validated using data generated in two other studies. According to scientists, it outperformed the existing machine learning methods, providing more accurate rates, but with some limitations: it does not offer biological interpretations between the microbiome and metabolomics and does not incorporate correlations among longitudinal samples.

In a study by Topçuoğlu et al [107], seven models were trained using fecal 16S rRNA sequence data to predict the presence of colonic screen-relevant neoplasia. The random forest model obtained the best AUROC in terms of predictive performance, but in terms of boundaries, more samples are required for each category for better accuracy, and in opposition to [106] the analysis did not explore deep learning methods.

Although very popular, ML predictions face some common problems and computational challenges [103]. Like gene profiles, many of the samples used are short, thus leading to the problem of high dimensionality. Not all of the sequences are necessary for classification, heterogeneous data increase the size of the problem, thus leading to overfitting, multicollinearity, and unnecessary noise. Other problems are those related to data considering sequences as statistical observations or to the quality of data, which can create background noise, hybridization due to uneven fluorescence on the chip, lot/batch effects, or the type of experimental protocol [108].

As the advances in bioanalytical techniques have evolved, more and more researchers are combining the traditional analysis methods for pathogen detection with smartphone-POC devices. Mobile phones have quickly evolved, from big and heavy cellular with 3 or 4 features, to small smartphones with wireless charging, augmented reality, foldable components and increased 5 G-enabled connectivity. Modern smartphones have high-quality cameras and better computational power to perform complex analyzes than the greatest supercomputers used in the 20^th century, making them suitable and convenient for use in POC devices.

Main advantage of such detecting systems is related to faster and more friendly manners than the classical biological equipment. If used at home, it accelerates treatment response, being of great help to people who cannot move or who are at a great distance from the clinic, or in resource limited areas, being able to be used worldwide. Of course, one of the biggest disadvantages regarding a smartphone POC-based device is the security part, both hacking and privacy data.

In general, smartphones can be used in two manners: the one in which it is part of the POC device as a whole incorporated system [109], and the one in which are used for a certain stage in the POC testing, as a stand-alone device (i.e. reader, light source, video recording, image capturing through attached ball lenses, converting the smartphone to an efficient microscope [110,111].

The use of smartphones as highly connected, portable, and versatile read‐out diagnostic platforms has been enabled by advancements in different biological fields, allowing them to be part of POC methods based on different sensing (i.e. optical, fluorescent, colorimetric, chemiluminescent, electrochemical [112,113]. Smartphone‐ POC applications are used in all kinds of areas such as neurology, oncology, hematology [109,110] virology [114,115] surgical diagnostics [113] or virus and bacteria detection [109,110,116], allowing a new paradigm to be developed. Different types of POC devices integrated with smartphones were used for investigation and detection of bacteria [111,117]. Table 4 presents some of the smartphone-based POC tests used in human bacteria detection.

Table 4. Smartphone-based POC tests used for bacterial detection.

POC device	Assay type	Detected bacteria	LOD	Assay duration	Ref.
Smartphone-based fluorescence microscope	Peptide nucleic acid (PNA) based FISH assay	Cronobacter sakazakii, Staphylococcus aureus	≥10^4 CFU/mL	Not specified	[118]
Smartphone-based rt LAMP	Biotin-labeled aptamer	Mycobacterium tuberculosis	10^4 CFU/mL	5 h	[119]
Smartphone-controlled, automated, and portable system	Colorimetric LAMP approach	Staphylococcus aureus	3.2 × 10−3 hemagglutinating units (HAU)/reaction; 30 colony-forming units (CFU)/reaction	40 min	[27]
Smartphone-based detection fluorescence microscopy	Aptamer-functionalized fluorescent magnetic nanoparticles	Staphylococcus aureus	10 CFU/ml	10 min	[120]
Pipette-actuated capillary array comb based on a smartphone-based hand-held device	LAMP	Klebsiella pneumoniae	200 CFU/capillary	85 min	[121]
Smartphone-based rt LAMP	LAMP	urine bacterial analysis	5 × 10^6 CFU/reaction	1 hour	[122]

There are a few reports based on smartphone-based technologies for exclusive human gut pathogens detection, most of them focusing on the detection of foodborne bacterial pathogens, like E. coli, Salmonella Enteritidis [121,123–125].

Xu et al [126] presents the POCKET DNA test, a versatile point-of-care DNA testing platform with a lot of areas of applications (from clinics, agriculture and environment to food applications). The kit weighs 60 g, is smaller than 25 cm, and it consists of an integrated chip (which integrates the sample preparation with a signal amplification), and a foldable box (which uses a smartphone as a heater, signal detector, and result readout). The detection of different types of DNA from sources like blood, buccal, urine, milk, river water, or plant leaves was sensitive (<103 copies/mL), specific (single-base differentiation), speedy (<2 h), and stable, making it a powerful device in healthcare and medical applications.

Even if several smartphone-based POC diagnostic tests have been outlined to be ‘REASSURED,’ most of them cannot be used on a large scale or in clinical practice. A team of researchers presented a new test for the detection of bacterial pathogen Helicobacter pylori (RCE@test) and demonstrated its efficacy by comparing it to a commercially available rapid urease test, the ‘CLO test’ [127]. They used an anthocyanin-rich red cabbage extract (RCE) as a natural indicator for the preparation of colorimetric diagnostic tests, which were interpreted by an image processing software using a smartphone application that works with a color analysis method based on the Euclidean distance formula.

5. Internet of things

The concept of Internet of Things (IoT) has opened the way to a new era, that of interweaving the Internet with essential electronic devices that make our lives easier (smart home security, adjust temperatures apps, smart watches, car assistance). Combining medicine, microbiology, or molecular biology, with machine learning techniques and artificial intelligence generates the idea of Internet of Medical Things (IoMT), focused especially on medical applications and healthcare.

Starting from various machine learning applications for microbiome and metagenomic prediction [128–130], the role of AI in gut metagenomics is under continuous exploration. Synergistic approach of employing AI and assisted biology for improved gut microbiome health control have reached another level. Nowadays, great emphasis is placed on a much closer analysis than working on samples, by creating ingestible sensors and smart video pills, which capture live intestinal microbiota, with the aim of finding the cause of suffering much faster and efficiently [131].

Classic imaging techniques of the intestinal tract (i.e. colonoscopy or colonography), are much more invasive and offer a state of discomfort to patients. Instead, modern technology describes new techniques for direct visualization of the microbiota, through small sensors that can be swallowed, whose substances do not affect human health and can capture data from the digestive system or even provide information about pH levels, pressure measurements or internal bleeding [132]. Connected to a medical device, the ingested sensors send a small signal to the receiver or medical application, from where different measurements or images of the gastrointestinal tract and colon can be used for forward investigation (i.e. start taking the meds when needed or just understand better the exact history of the patient). According to Smart Pills Market Brochure [133], the global smart pills market will reach over 1 billion $ by 2028. The research also categorizes smart pills according to the function they have: diagnostic (image capturing), monitoring, or therapeutic (targeted drug delivery to particular regions of the gastrointestinal tract such as esophagus, small intestine, fat, stomach) or the place where can be used (specialized units or at home, depending on the severity and needs of the patient).

Kalantar-zadeh et al report that the factors in the design of ingestible capsules are ranging from physical dimensions for easy ingestion, biocompatibility of the materials used, lifetime of the battery to high fidelity of data measurements of chemical constituents inside the gut [134].

In [135] the authors summarized some of the digital pills with ingestible sensors used in Gastroenterology, among which we can mention a released immunosuppressant at or near the site of disease in the gastrointestinal tract [136] devices which can identify a location within the gastrointestinal tract to release a special medication [137] or encapsulated devices for direct visualization and images collecting of the small bowel and colon [138].

Another smart device for studying the gastrointestinal microbiome is presented in [139], a 3D-printed capsule (9 × 15 mm) designed to collect microbiome samples throughout the gastrointestinal tract. The hydrogel within the sealed capsule was validated as a host for bacterial culture using a liquid sample containing E. coli.

Some limitations are related to bowel obstruction due to capsule retention [134] or high cost due to the effort of designing process, manufacturing, sensor testing, machine learning algorithms etc.

To conclude, the integration of Internet of Things, machine learning and smartphone-based POC applications have grown rapidly, gaining more and more trust in personalized, predictive and preventive medicine. The future sounds promising, and through the interaction between smart pills and engineered bacteria, new ways of detecting and eradicating gut pathogen diseases will be developed.

With the rise of IoT devices used as monitoring sensors for potential pathogens in the food chain, the quantity of data produced is of large scale and can hardly be managed by the regular users of the devices in question. Considering this, the need for implementation of methods to utilize the data rose alongside the implementation of the devices. The issue presented falls under the umbrella of the term Big Data and, although the term itself does not have a proper definition, the general idea accepted is that Big Data refers to the amount of data (in terms of terabytes or higher quantities) a set of devices produce and store in a specifically allocated space (either local server or cloud) [140,141].

The adoption of IoT capable devices in the food chain can provide real time information about the current situation of animal feed, health status of the animal and growth environment or even potential exposure to pathogens [142] has shown a promising device capable of detecting microbial particles in a rapid manner, such a process could be adopted in food processing plants and extend it for bacterial pathogen detection in order to constantly screen for threats.

The data collected from food processing lines provide an insight into potentially dangerous outbreaks, which could lead to an entire lot being compromised, thus the need for implementation of predictive models using the obtained data in order to avoid critical situations [143]. The process of creating a digital system capable of managing the amount of data should use a reference model as proposed by [144] which is taking into account the fact that all implementations of such systems should process and print data in a similar manner in order to keep consistency between applications.

6. Challenges and perspectives

The development of an ideal POCT for pathogen detection has indeed been a significant research area in recent years, but several challenges have hindered its widespread adoption and market revolution. These challenges are related to sample complexity, assay design, multiplexed detection, scaling up, infrastructure requirements, contamination and specificity hurdles (Figure 3b).

Several challenges have been identified in the field of POCT and these include:

6.1. Minimal manual steps and long shelf life

POC systems should exhibit features such as sample-to-answer detection, flexible design, and, if possible, automation coupled with artificial intelligence. Achieving sample-to-result pathogen detection with minimal manual steps and long shelf life is crucial [145].

6.2. Assay design

Selecting the detection method for a POC system involves recognizing the assay and design challenges, sensitivity requirements, reagent storage, prevention of contamination, and commercial resources. Isothermal amplification methods like LAMP, RCA, and RPA are advantageous for resource-limited settings due to their flexibility and compatibility with direct detection from crude samples. Nevertheless, these methods have several constraints including specificity, inability to accurately detect clinically relevant mutations, complex primer design and the need for reagent refrigeration.

6.3. Multiplex detection

POC devices should be adaptable for various applications and multiplexing to identify different pathogens. For example, the Biofire GI panel that uses Multiplex PCR allows the detection (from stool samples collected in Cary Blair medium) of multiple GI pathogenic bacteria such as Salmonella, Campylobacter (C. jejuni/C. coli/C. upsaliensis), Yersinia enterocolitica, Vibrio (V. parahaemolyticus/V. vulnificus/V. cholerae), Vibrio cholerae, Clostridioides, Plesiomonas shigelloides, Enteroaggregative E. coli, Enterotoxigenic E. coli, Shiga-like toxin-producing E. coli, Shigella/Enteroinvasive E. coli (EIEC), Enteropathogenic E. coli (EPEC), E. coli O157. Despite its clear clinical benefits, this POC platform is not available for low resource locations. Oh et al [146] reported the development of a centrifugal microfluidic device for the identification of multiple foodborne pathogens (Escherichia coli O157: H7, Salmonella typhimurium and Vibrio parahaemolyticus) with a LOD of 380 copies genomic DNA of using LAMP and a colorimetric-based detection method. However, the sample preparation needs to be performed off-chip.

In case of lab-on-chip devices, parallel reactors on a single chip enable the testing of several targets at once. For example, Sun et al [147] were able to detect five equine respiratory pathogens on a single chip using digital LAMP by drying and reconstituting the specific primers into solution during the addition of LAMP reaction mix. Recently, solid phase-LAMP has opened a promising direction toward the development of a multiplex POC system for rapid detection of multiple pathogens such as Salmonella spp, Campylobacter coli, and Campylobacter jejuni [148].

A recent development by Zhang et al (2021) involves a microchip-based approach using LAMP for the simultaneous detection of E. coli, Salmonella, S. aureus, and Vibrio parahaemolyticus. The microchip has reaction chambers embedded with paper that contains LAMP reagents and primers specific to the pathogens under consideration, and the microchip’s setup allows for the distribution of a sample containing the target pathogens into different reactive chambers for each pathogen. The sample is injected into the central chamber of the microchip and further distributed into the specific reactive chambers that correspond to the target pathogens [112].

6.4. Qualitative versus quantitative results

Most POCT developed for the detection of gut pathogens are qualitative. However, the ability to quantitatively detect microbial NAs is crucial for assessing the severity of an infection, tracking disease progression, and monitoring the effectiveness of treatments. By integrating real-time quantification into a smartphone-based platform can provide valuable information beyond simple presence/absence pathogen detection. For example, Yin et al [149] combined enhanced colorimetric LAMP assay and smartphone-based color analysis to quantitatively detect HPV in saliva samples/clinical vaginal swab samples, and HIV in plasma samples.

Real-time colorimetric quantitative molecular detection of infectious diseases on smartphone-based diagnostic platforms enables rapid and accurate identification and quantification of specific pathogens causing infectious diseases. This technology leverages colorimetric assays to detect the presence of target NAs from the pathogens of interest and the diagnostic results are interpreted and displayed on a smartphone or other portable device. These smartphone-based platforms could potentially enable remote monitoring and data sharing for epidemiological studies. While the concept is promising, there may be challenges related to assay sensitivity, specificity, and the need for rigorous validation to ensure accurate and reliable results.

6.5. Detection from fecal samples

Most studies focus on detection of gut microbes from either food samples or patient’s blood and only few focus on direct detection from fecal samples. From a clinical point of view, it is essential to rapidly decipher the source of diarrheal disease in order to improve antimicrobial stewardship [150]. A recent study performed by Charkraborty et al described a rapid and simple diagnostic assay for detection of ETEC and Shigella directly from fecal samples using lyophilized reaction strips and LAMP [151].

6.6. Accuracy of automated systems

Advances in sensor technology have contributed to the portability of pathogen detection devices, making them compatible with smartphones for combined analysis and readout. Despite the advancements, there are still some challenges that need to be addressed. For instance, automated microfluidic smartphone-based biosensors combine microfluidic technology with smartphones to create portable and user-friendly diagnostic tools [152]. However, one challenge in designing smartphone-based microfluidic biosensor systems is finding a balance between the number of detection components attached to the smartphone and achieving target accuracy. The inclusion of accessories in the system can affect convenience, flexibility, and interfere with the diagnostic process, potentially leading to accuracy loss [153]. Miniatured sensors used in smartphone-based systems might not attain the same level of accuracy as conventional laboratory instruments [154].

6.7. Diagnostic of viable but non-culturable bacteria

Foodborne pathogens (i.e. Listeria monocytogenes) have developed different survival strategies. For example, refrigeration is a common practice to control microbial growth and extend the shelf-life of foods. However, exposure to such conditions can cause pathogens to enter a state known as viable but non-culturable (VBNC), which presents challenges for their detection. In the VBNC state, the bacteria are alive but cannot be cultured using traditional methods [155] and specialized growth enrichment conditions are required to resuscitate the VBNC cells. POCTs such as LFIA may yield false-negative results when VBNC cells are present since antigen expression may have been severely impaired by refrigeration. In this case, NA – based lateral flow immunoassays (NALFIA) could be employed as alternative POCTs [156].

6.8. Biomolecule adhesion to the testing device

Biomolecules (such as NAs, proteins, or other cellular components) can adhere to the surfaces of microfluidic channels in a nonspecific manner triggering loss of target molecules, alteration of concentrations, and interference with reactions, and subsequent inaccurate results. To mitigate these challenges, various surface treatments (polyethylene glycol, linear polyacrylamide) can be employed to modify the properties of the microchannel walls [157].

6.9. Contamination risk

To prevent contamination risks, various approaches are used, such as droplet-based microfluidic systems and designing sealed amplification chambers. To address the issue of reagent storage, a possible solution would be to use polymer/paper hybrid environments for storing reagents in a dried form for extended periods [158].

6.10. Oversensitivity, specificity, and false positives results

CRISPR system’s ability to tolerate some degree of mismatched nucleotides could lead to unintended binding and potential false positives. CRISPR-Cas systems are highly sensitive to their target sequences, which can sometimes lead to false-positive results due to the recognition of closely related sequences [159]. Adoption of CRISPR-based diagnostics into the clinic can be a complex process due to regulatory approval, validation studies, and the existing prevalence of established methods like RT-PCR. Despite its potential, CRISPR technology also raises ethical concerns, particularly in the context of gene editing and potential unintended consequences, therefore responsible research and a careful approach to its applications are essential.

6.11. Clinical validation and implementation

Validating POCT devices on real clinical samples is a critical step in ensuring the accuracy and reliability of these devices. It is imperative that the results obtained from these devices are consistent with those from established reference methods (i.e. PCR, cultivation) [160]. Validation studies should be designed to mimic real clinical scenarios, with an accurate sample size and a blinded testing approach, in order to prevent interpretation biases.

POCT implementation on a large scale has to overcome certain barriers related to costs, availability, awareness, and accurate interpretation of results. In addition, patient-specific biomarkers profile may be linked to many other medical disorders (i.e. inflammatory bowel disease), hence a qualified professional should examine the test findings and other symptoms of the patient. Raising public awareness about the availability, benefits, and proper usage of POCT is essential to encourage their use on a large scale. Without professional guidance, there could be concerns regarding the accurate interpretation of the POCT results.

6.12. Low resource considerations

POC assays must maintain consistent performance over extended storage periods, including against different environmental conditions like increased temperature and humidity. In some settings, the lack of a reliable infrastructure, including electricity and internet connectivity, can raise significant challenges for the implementation of POCT. For example, self-powered sensors that generate energy sustainably, instead of using a power grid or batteries, could be helpful in leveraging POCT applicability (i.e. enzymatic fuel cells) [161].

6.13. POCT Scaling-up, lifecycle and cost-to-benefit ratio

POCT development should be based on a balance between simplicity, cost, and environmental impact (i.e. sustainable manufacturing and disposal). A major POCT limitation occurs when transitioning from lab-designed devices to scaling-up for clinical implementation. Scaling up POC devices requires careful planning, collaboration (i.e. stakeholder engagement), and adaptation to local contexts. Challenges related to power supply and connectivity, especially in remote areas should also be taken into consideration. A thorough analysis and pilot studies should be performed taking into account the overall cost of deploying and maintaining the devices at scale, including device costs, consumables, maintenance, training, and infrastructure requirements. A strategy for device storage, sustainability, distribution, and disposal of waste materials should be implemented alongside a supply chain strategy to ensure a consistent and reliable supply of POCT devices and consumables. Successful development of commercial POC devices needs to overcome engineering hurdles related to chip materials, software costs and amplification and detection techniques. An elegant review by Nguyen et al [145] presents possible solution on how to overcome these challenges. In addition, disposal and recycling of POCT materials (i.e. LFA casettes) should be advisable. Nevertheless, in low resource regions (for which POCT is needed more), landfilling is often used regardless of its environmental effects [162].

7. Conclusions

The COVID-19 pandemic has exposed deficiencies in rapid detection and early screening of infectious diseases. The need for portable POC diagnostics and correlated devices is urgent to contribute to accurate and quick infection detection.

The continuous advancements in molecular detection techniques have an impact on analysis throughput, accuracy, and sensitivity, as well as the potential for automation and customization.

Molecular detection techniques can significantly decrease turnaround time for obtaining results but despite their advantages, rapid detection methods have limitations, including pathogen amount (detecting low levels of pathogens may require pre-enrichment steps to increase the concentration of target organisms) and NA integrity (DNA/RNA in samples can degrade or be affected by PCR inhibitors, affecting the reliability of results). Pre-enrichment steps can increase the concentration of target organisms in samples with low bacterial counts but this step may introduce contamination risks, which can be mitigated through proper personnel training and sample management systems.

While each diagnostic method has its strengths and weaknesses, there is a need for methods that combine accuracy, sensitivity, simplicity, and rapid results for on-site detection and large-scale screening, especially in the context of pandemic prevention and control. Combining both conventional detection methods and molecular techniques can offer a comprehensive approach to studying the ecology and epidemiology of disease-causing organisms.

Finally, integration of POCT results into diagnostic algorithms and treatment decisions requires careful validation and consideration of clinical guidelines. To achieve this, POCT devices need to exhibit consistent performance across different settings and to provide accurate and reliable results comparable to those obtained from centralized laboratory testing. The integration of wireless data transmission capabilities into smartphone-based platforms can enable real-time pathogen tracking and disease mapping. This data sharing has the potential to support epidemiological studies, aid in disease surveillance, and inform public health interventions to respond more effectively to outbreaks, allocate resources, and implement targeted interventions. While this technology holds great promise, regulatory approval and integration into healthcare systems would have significant hurdles to overcome. Also, given the sensitive nature of medical data, robust data security measures would need to be in place to protect patient information during wireless transmission and storage.

8. Expert opinion

Given the alarming rates of antimicrobial resistance, it is essential to rapidly decipher the source of diarrheal diseases to improve antimicrobial stewardship. The landscape of healthcare is dynamic, and the evolution of POCT for bacteria is being influenced by a combination of technological, regulatory, and societal factors.

The shift toward patient-centered care and decentralized healthcare models emphasizes the importance of point-of-care diagnostics, promoting rapid decision-making and reducing the need for centralized laboratory testing. Rapid detection of gut pathogens can be done using a wide array of techniques including NA amplification, immunological assays, electrochemical impedance spectroscopy, and mass spectrometry.

However, developing effective assays for the detection of gut microbes can be complex. Assay design must consider factors like target selection, possibility for real-time and multiplex detection, type of sample used, reagent stability and storage, primer/probe design, and optimizing reaction conditions for accuracy and sensitivity.

Developing affordable POCT devices that are economically viable for widespread use, especially in resource-limited settings, remains a significant hurdle. Balancing cost-effectiveness with high performance is crucial, so, there is a need to explore the use of innovative materials and manufacturing processes to reduce production costs without compromising the performance of these devices.

Importantly, implementing robust quality control mechanisms for POCT devices is crucial to maintain their accuracy over time, hence, ensuring that end-users can perform quality assurance checks easily is important for sustained reliability.

In addition, the integration of wireless data transmission into miniaturized devices may promote real-time pathogen tracking and disease mapping, supporting authorities to respond more effectively to outbreaks, allocate resources, and implement targeted interventions. Validation of these devices using real clinical samples is essential in demonstrating the specificity, accuracy, and reliability of these devices.

Mitigation of these challenges requires interdisciplinary collaboration among scientists, medical practitioners, engineers, and industry partners. Future research and development efforts are essential to improve the sensitivity, specificity, and versatility of POCT systems for the detection and quantification of gut microbes, ultimately advancing the field of infectious disease diagnostics and management.

In the coming years, a significant progress is foreseen in areas such as identification and validation of novel biomarkers associated with infectious diseases and sepsis, development of predictive models for disease trends and personalized treatment strategies based on POCT data, as well as utilization of POCT in telehealth for monitoring chronic conditions and managing infectious diseases in remote or underserved populations. Integration of AI algorithms to analyze test results will continue to improve accuracy and speed in identifying bacterial infections and AI-driven predictive analytics will enable the early detection of bacterial outbreaks and trends, aiding in timely public health interventions.

In the next years, a continued refinement and expansion of POCT capabilities is expected. The current POCT research goal is to make diagnostic testing more accessible, rapid, and accurate, addressing the diverse needs of healthcare settings and patient populations. As long as infectious diseases persist and healthcare continues to advance, the development and improvement of point-of-care testing for bacteria are likely to persist.

Article highlights

• Gut microbes pose a significant health concern and their fast detection is crucial.

• POCT systems offer rapid, sensitive, low-cost and sample-to-answer methods for the detection of gut microbes.

• Detection of gut pathogens can be done using a wide array of techniques including NA amplification, immunological assays, electrochemical impedance spectroscopy, and mass spectrometry.

• The current trends in POCT involve incorporating wireless communication and machine learning along with exploring the intricacies of the gut microbiome.

• Given the complexity inherent in creating assays for the detection of gut microbes, the design process must take into account various factors such as target selection, the potential for real-time and multiplex detection, stability and storage of reagents, primer/probe design, and optimization of reaction conditions.

• In the forthcoming years, substantial advancements are anticipated in domains such as the discovery and authentication of new biomarkers linked to infectious diseases and sepsis, as well as the creation of predictive models for disease patterns and personalized treatment approaches utilizing POCT data.

Abbreviations

CFU	=	colony forming units
CRISPR	=	clustered regularly interspaced short palindromic repeats
EIEC	=	Enteroinvasive E. coli
ELISA	=	enzyme-linked immunosorbent assay
EPEC	=	Enteropathogenic E. coli
HDA	=	helicase-dependent amplification
HIV	=	human immunodeficiency virus
HPV	=	human papillomavirus
IoT	=	Internet of Things
LAMP	=	loop-mediated isothermal amplification
ML	=	machine learning
NA	=	nucleic acid
NASBA	=	nucleic acid sequence-based amplification
PCR	=	polymerase chain reaction
POC	=	point-of-care
POCT	=	point-of-care-testing
RCA	=	rolling circle amplification
RPA	=	Recombinase Polymerase Amplification

Author contributions

G Gradisteanu Pircalabioru and C Iliescu were involved in the conception and design of the manuscript. M Raileanu, G Gradisteanu Pircalabioru, M Viorel Dionisie, and IO Lixandru-Petre were involved in drafting of the paper. C Iliescu revised the paper critically for intellectual content; All authors approved of the final version to be published and agree to be accountable for all aspects of the work.

Declaration of interest

The authors have no other relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript apart from those disclosed.

Reviewer disclosures

Peer reviewers on this manuscript have no relevant financial or other relationships to disclose.

Acknowledgments

Views and opinions expressed are those of the authors only and do not necessarily reflect those of the European Union or European Research Executive Agency (REA). Neither the European Union nor the granting authority can be held responsible for them.

Additional information

Funding

This manuscript was funded by the European Union’s Horizon Europe framework program 2021-2027, under the Coordination and Support Actions, HORIZON-WIDERA-2022-TALENTS-01 [grant agreement - 101087007 – eBio-hub], funded by the European Union as well as via the Unitatea Executiva pentru Finantarea Invatamantului Superior, a Cercetarii, Dezvoltarii si Inovarii (PN-III-P4-ID-PCE-2020-1886 Contract PCE 180/17/02/2021)