Search Results

The BioCode Gastrointestinal Pathogen Panel (GPP) is a qualitative multiplexed in vitro diagnostic test intended for use with the BioCode MDx 3000 Instrument. The BioCode GPP is capable of the simultaneous detection of nucleic acids from multiple bacteria, viruses, and parasites extracted directly from unpreserved in Cary-Blair transport medium obtained from individuals with simptoms of gastrointestinal infection. The following bacteria, parasites, and viruses are identified using the BioCode Gastrointestinal Pathogen Panel:

• . Campylobacter (C. jejuni/C. coli)
• Clostridium difficile (C. difficile) toxin A/B (Fresh samples only)
• l Salmonella spp
• Vibrio (V. parahaemolyticus/V. vulnificus/ V. cholerae), including specific identification of Vibrio parahaemolyticus .
• . Yersinia enterocolitica
• . Enteroaggregative Escherichia coli (EAEC)
• Enterotoxigenic Escherichia coli (ETEC) lt/st
• E. coli 0157 serogroup
• Shiga-like toxin-producing Escherichia coli (STEC) stx1/stx2
• Shigella/ Enteroinvasive Escherichia coli (EIEC)
• Cryptosporidium spp (C. parvum/C. hominis)
• Entamoeba histolytica
• Giardia lamblia (also known as G. intestinalis and G. duodenalis)
• . Adenovirus F 40/41
• Norovirus GI/GII ■
• . Rotavirus A

The BioCode GPP is indicated as an aid in the diagnosis of gastrointestinal illness and results are meant to be used in conjunction with other clinical, laboratory, and epidemiological data. For In Vitro Diagnostic Use Only. For Prescription Use Only.

Positive results do not rule out co-infection with organisms not included in the BioCode GPP. The agent detected may not be the definite cause of the disease. Negative results in the setting of clinical illness compatible with gastroenteriis may be due to infection by pathogens that are not detected by this test or non-infectious causes such as ulcerative colitis, irritable bowel syndrome, or Crohn's disease. Concomitant culture is necessary for organism recovery and further typing of bacterial agents. This device is not intended to monitor or guide treatment for C. difficile infection.

Due to the small number of positive specimens collected for certain organisms during the prospective clinical study, performance characteristics for Adenovinus 40/41, Campylobacter, E. coli 0157, Shigella(EIEC, Yersinia enterocolitica, and Giardia lamblia were established additionally with retrospective clinical specimens. Performance characteristica, Giardia lamblia, Yersinia enterocolitica and Vibrio (V. parahaemolyticus, V. cholerae) were established primarily using contrived clinical specimens.

The BioCode® Gastrointestinal Pathogen Panel (GPP) is a multiplexed nucleic acid-based test designed to be used with the BioCode MDx-3000 system. The BioCode MDx-3000 is an automated system that integrates PCR amplification, target capture, signal generation and optical detection for multiple gastrointestinal pathogens from a single stool specimen, either unpreserved or in Cary Blair. Stool specimens are processed, and nucleic acids extracted with easyMAG, MagNA Pure 96, KingFisher Flex and KingFisher Apex Dx. Once the PCR plate is set up and sealed, all other operations are automated on MDx-3000. The BioCode MDx-3000 Gastrointestinal Infection Panel simultaneously tests for 17 pathogens (see table below) from unpreserved stool specimens or stool collected in Cary-Blair transport medium. Results from the BioCode Gastrointestinal Pathogen Panel (GPP) test are available within less than 4 hours.

The provided text describes the BioCode Gastrointestinal Pathogen Panel (GPP), a diagnostic test for gastrointestinal pathogens, and a study to demonstrate its substantial equivalence to a predicate device.

Here's an analysis of the acceptance criteria and study data:

1. Acceptance Criteria and Reported Device Performance

The acceptance criteria for each pathogen are implied by the reported Positive Percent Agreement (PPA) and Negative Percent Agreement (NPA) compared to the easyMAG extraction method (used with the predicate device) for each extraction system (KingFisher Flex and KingFisher Apex Dx). While explicit pre-defined acceptance criteria values are not stated, the tables present the achieved performance with 95% Confidence Intervals. Generally, a high PPA and NPA (typically >90-95%) with narrow confidence intervals are expected for substantial equivalence for diagnostic tests.

Here's a summary of the reported performance for the BioCode GPP using KingFisher Flex and KingFisher Apex Dx, focusing on the "All Archived" data as it covers a larger sample size for each target. Targets with PPA < 90% are highlighted as indicated in the document; however, the document notes that these are "Positive agreement <90%" for Archived samples only, and specifically points out individual false positives/negatives that were retested and often resolved. Targets for which "N/A" is reported for PPA/NPA indicate no positive/negative archived samples were available for comparison, and thus performance was primarily established with contrived samples (as indicated in the 'Indications for Use' section).

Table of Device Performance (Archived Samples - All Combined)

Note:

• Asterisks (**) in the table indicate that the specific performance measure (PPA) was below 90% for archived samples. The document explicitly notes for Vibrio spp. (not parahaemolyticus) that 1 out of 2 positive concordance was observed for both KingFisher Flex and KingFisher Apex Dx, resulting in 50% PPA.
• Retesting of discordant samples often showed resolution in favor of the predicate or new device, but the summary tables reflect the initial comparison. The footnotes (a-q) provide detailed explanations for discordant cases, including instances where samples were invalid or initial false results were later identified as true negatives/positives upon retesting.

2. Sample Size and Data Provenance

• Test Set (Clinical Method Comparison Study):
  • Archived Samples: 468 remnant, de-identified samples (254 frozen unpreserved stool and 214 inoculated Cary-Blair stool). These samples were prospectively collected for the clinical study that resulted in the predicate device (K180041 BioCode® GPP FDA clearance). The country of origin is not explicitly stated, but clinical studies for FDA clearance typically involve multi-center studies within the US or compliant international sites. "Mayo-GI" and "TRI" sample names suggest US clinical sites.
  • Fresh Samples (for C. difficile testing): 54 freshly collected leftover samples (unpreserved).
  • Contrived Samples: 120 samples (15 at 3x LoD and 15 at 6xLoD for each of three targets: Entamoeba histolytica, Yersinia enterocolitica, and Vibrio spp.).

3. Number of Experts to Establish Ground Truth for Test Set and Qualifications

The ground truth for the clinical method comparison study was established by comparing the results of the new device (BioCode GPP with KingFisher Flex and KingFisher Apex Dx extraction) against the results obtained with the easyMAG extraction, which was used with the predicate device (BioCode GPP K190585). The predicate device itself would have undergone its own clinical validation against a pre-established ground truth. Therefore, the "ground truth" for this specific study is the performance of the BioCode GPP with easyMAG extraction.

The document does not explicitly state the number or qualifications of experts used for establishing the initial ground truth of the predicate device, or for adjudicating discordant results in this study. However, the retesting notes for discordant samples mention "consensus results" were not used in the agreement calculation, implying that the retesting was performed to understand the discrepancy rather than to establish a new ground truth.

4. Adjudication Method for the Test Set

The adjudication method is described in the footnotes for Tables 11 and 16. For discordant results between the easyMAG (reference) and the KingFisher Flex/Apex Dx (new device), samples were retested twice with both original extraction methods. The outcome of these retests served to clarify the discrepancy, though the initial raw comparison was used to calculate the agreement, with retest findings noted in the footnotes. For example, a "false negative" for the new device might become a "true negative" upon retesting if the reference system also failed to detect it consistently. However, "consensus results of the discordant samples were not used in the calculation of the agreements" in the main tables but rather for detailed understanding.

5. Multi-Reader Multi-Case (MRMC) Comparative Effectiveness Study

No Multi-Reader Multi-Case (MRMC) comparative effectiveness study was done. This study focuses on a laboratory diagnostic device, not an imaging device requiring human reader interpretation. Therefore, the concept of "human readers improve with AI vs without AI assistance" is not applicable here.

6. Standalone Performance Study

Yes, a standalone performance study was done for the device in the context of the new extraction platforms. The "Nonclinical testing summary" details a Reproducibility Study and a Limit of Detection (LoD) study. These studies evaluate the algorithm's (BioCode GPP assay's) consistency and sensitivity when integrated with the new KingFisher Flex and KingFisher Apex Dx extraction systems, independent of direct human interpretative input beyond standard lab procedures.

• Reproducibility Study: Assessed intra-assay, inter-assay, day-to-day, and instrument-to-instrument (operator-to-operator) reproducibility using quantitative controls diluted in stool.
• Limit of Detection (LoD) Study: Determined the lowest concentration at which ≥95% detection was achieved in 20 replicates for each target, for both unpreserved and Cary-Blair stool, using the new extraction methods.

7. Type of Ground Truth Used

For the clinical method comparison study (test set), the ground truth was the result generated by the predicate device's established extraction method (easyMAG). This is a comparative effectiveness study where the predicate device's performance established the "truth" for evaluating the new device's substantial equivalence.

For the LoD and reproducibility studies, the ground truth was based on quantified stocks of bacteria, viruses, or parasites (except for Norovirus GI/GII which used positive clinical specimens with serial dilutions) introduced into negative stool matrices.

8. Sample Size for the Training Set

The document does not explicitly describe a "training set" for the BioCode GPP itself, as it's a multiplex nucleic acid-based test rather than a machine learning or AI algorithm in the common sense requiring explicit training data. The assay's design (probes, primers, etc.) would have been developed and optimized against a range of target and non-target organisms.
The 468 remnant clinical samples used in the method comparison study were prospectively collected for the predicate device's clinical study (K180041 BioCode® GPP FDA clearance), suggesting they were part of the validation for the predicate, not a training set for the current device's underlying assay design.

9. How the Ground Truth for the Training Set Was Established

As noted above, an explicit "training set" with ground truth in the AI/ML context isn't specified for this type of diagnostic assay. The development of such an assay involves extensive analytical validation to establish specificity, sensitivity, and inclusivity using characterized microbial strains, clinical isolates, and negative controls. The ground truth for these developmental studies would be based on molecular characterization (e.g., sequencing), culture, or other reference methods for the target organisms. The performance of the predicate device (K190585) would have been established through a clinical study where its results were compared against clinical diagnosis, culture, or other validated reference methods.

Ask a specific question about this device

The BioCode Respiratory Pathogen Panel (RPP) is a qualitative multiplexed nucleic acid-based in vitro diagnostic test intended for use with BioCode MDx 3000 Instrument. The BioCode of the simultaneous detection and identification of nucleic acids from multiple viruses and bacteria extracted from nasopharyngeal swab (NPS) samples obtained from individuals with signs and/or symptoms of respiratory tract infection. The following pathogens and subtypes are identified using the BioCode RPP:

• Adenovirus
• · Coronavirus (229E, OC43, HKU1, and NL63)
• Human Metapneumovirus A/B
• · Influenza A, including subtypes H1, H1 2009 Pandemic, and H3
• Influenza B
• Parainfluenza Virus 1
• Parainfluenza Virus 2
• · Parainfluenza Virus 3
• · Parainfluenza Virus 4
• · Respiratory Syncytial Virus A/B
• Rhinovirus/Enterovirus
• · Bordetella pertussis
• Chlamydia pneumoniae
• Mycoplasma pneumoniae

The detection and identification of specific viral and bacterial nucleic acids from individuals exhibiting signs and/or symptoms of a respiratory infection aids in the diagnosis of respiratory infection if used in conjunction with other clinical and epidemiological information. The results of this test should not be used as the sole basis for diagnosis, treatment, or other patient management decisions.

Negative results in the setting of a respiratory illness may be due to infection with pathogens that are not detected by this test, or lower respiratory tract infection that may not be detected by a nasopharyngeal swab specimen. Positive results do not rule out co-infection with other organisms: the agent(s) detected by the BioCode RPP may not be the definite cause of disease. Additional laboratory testing (e.g. bacterial and viral culture, immunofluorescence, and radiography) may be necessary when evaluating a patient with possible respiratory tract infection.

Due to the genetic similarity between Human Rhinovirus, the BioCode RPP cannot differentiate them. A positive BioCode RPP Rhinovirus/Enterovirus result should be followed up using an alternate method (e.g., cell culture or sequence analysis) if differentiation is required. The BioCode RPP detects Human Rhinovirus with reduced sensitivity. If a more accurate Rhinovirus result is required, it is recommended that specimens found to be negative for Human Rhinovirus/Enterovirus after examination using BioCode RPP be confirmed by an alternate method (e.g. FDA cleared molecular tests).

Performance characteristics for Influenza A were established when Influenza A H1 2009 Pandemic and A H3 were the predominant Influenza A viruses in circulation. Performance of detecting Influenza A may vary if other Influenza A strains are circulating or a novel Influenza A virus emerges.

The BioCode® Respiratory Pathogen Panel is a respiratory pathogen multiplex nucleic acid test designed for use with the BioCode® MDx-3000 system. The BioCode® MDx-3000 is an automated system that integrates PCR amplification, target capture, signal generation and optical detection for multiple viral and bacterial pathogens from a single nasopharyngeal swab specimen collected in transport media. Specimens are processed and nucleic acids extracted with the NucliSens easyMAG or Roche MagNA Pure 96 automated systems. Once the PCR plate is set up and sealed, all other operations are automated on MDx-3000. The BioCode® RPP simultaneously tests for 17 pathogens and/or subtypes (see table below) from nasopharyngeal swab specimens collected in UTM or VTM. Results from the BioCode RPP test are available within about 5 hours, including off-board nucleic acids extraction.

Here's a breakdown of the acceptance criteria and the study that proves the device meets them, based on the provided text.

Based on the provided text, the device is an in vitro diagnostic test (BioCode Respiratory Pathogen Panel - RPP). The primary evidence for meeting acceptance criteria comes from clinical performance studies and analytical performance studies. The document does not explicitly state pre-defined acceptance criteria values (e.g., "Positive Agreement must be > X%"), but rather presents the results achieved and implicitly confirms they meet regulatory expectations for substantial equivalence.

Acceptance Criteria and Reported Device Performance

Since explicit numerical acceptance criteria were not stated, I'll present the reported performance, which implicitly met the necessary thresholds for FDA clearance. The document focuses on Positive Percent Agreement (PA) and Negative Percent Agreement (NA) compared to an FDA-cleared molecular multiplexed respiratory pathogen panel (the "comparator test").

Table of Acceptance Criteria (Implicit) and Reported Device Performance (Clinical Study - Total Specimens)

Study Details:

1. Sample Size and Data Provenance:

   • Clinical Test Set (Prospective Study):
     • Sample Size: 2649 residual nasopharyngeal swab (NPS) specimens in VTM or UTM.
     • Data Provenance: Prospectively collected from patients suspected of respiratory tract infections at five geographically diverse clinical sites in the U.S. (August 2017 to May 2019). Specimens were either tested freshly (stored 2-8°C for no more than 7 days) or stored frozen and thawed later.
   • Clinical Test Set (Archived Specimens - Retrospective Study):
     • Sample Size: 165 clinical specimens (archived NPS in VTM or UTM).
     • Data Provenance: Retrospective, preselected archives from source laboratories, chosen because they had previously tested positive for low-prevalence pathogens or were negative. These specimens were then randomized and blinded.
   • Contrived Specimens (Analytical Performance):
     • Chlamydia pneumoniae & Influenza A H1: 50 unique negative natural NPS in VTM or UTM specimens were spiked to create 100 positive samples (2X LOD or greater) and interspersed with negative samples. A total of 110 samples were tested.
     • Reproducibility Study: 6 contrived positive samples and 1 negative sample, each extracted in triplicate, each assayed in singlet. This translates to 90 samples per concentration level per extraction type (easyMAG, MagNA Pure 96) across multiple sites/runs, total 90 (3x LoD) + 90 (1.5x LoD) + 450 (no analyte) = 630 for each virus/bacteria category listed (e.g., Adenovirus, Coronavirus, Human Metapneumovirus).
     • Limit of Detection (LoD): 20 replicates for each target at or near the presumptive LoD.
     • Analytical Reactivity/Inclusivity: Triplicate extractions for each strain/serotype.
     • Analytical Specificity/Cross Reactivity: Triplicate extractions for each off-panel and on-panel organism.
     • Interfering Substances/Microbes: Triplicate extractions for each sample/substance combination.
     • Competitive Inhibition: Triplicate extractions for each pooled sample.
     • Specimen Stability: Triplicate extractions for each time point and storage condition.
     • Matrix Equivalency: Quadruplicate extractions for each pool/matrix combination.

2. Number of Experts and Qualifications for Ground Truth:

   • The document states that the BioCode RPP test results were compared against those from an "FDA-cleared molecular multiplexed respiratory pathogen panel" (Standard of Care/Comparator Test). This FDA-cleared comparator test serves as the primary "ground truth" for the clinical studies.
   • For discrepant results in the clinical studies, further investigation was conducted by "performing independent molecular tests, including analytically validated PCR followed by bi-directional sequencing assays and alternate NAATs." This implies the use of specialized laboratory personnel with expertise in molecular diagnostics, but the exact number or specific qualifications (e.g., "radiologist with 10 years of experience") are not specified as this is a molecular diagnostic device, not an imaging device typically read by radiologists. The ground truth for presence/absence of pathogens is established by these comparator and confirmatory molecular tests.

3. Adjudication Method for the Test Set:

   • For the clinical studies (prospective and archived), the method for addressing discrepancies between the BioCode RPP and the comparator test was to perform "independent molecular tests, including analytically validated PCR followed by bi-directional sequencing assays and alternate NAATs." This serves as an adjudication method, where an orthogonal, high-accuracy method is used to determine the true positivity or negativity of discrepant samples. The specific "2+1" or "3+1" concensus type is not explicitly mentioned, but the retesting with confirmatory methods acts as a form of adjudication.

4. Multi-Reader Multi-Case (MRMC) Comparative Effectiveness Study:

   • No, an MRMC study was not done. This type of study (comparative effectiveness with human readers) is typical for AI-powered imaging devices where human interpretation is a key component. The BioCode RPP is an in vitro diagnostic (IVD) molecular test where the output is directly from the instrument analysis, not reliant on human interpretation of complex images. The study focuses on comparing the device's accuracy against a legally marketed predicate device/standard of care, not on improving human reader performance.

5. Standalone Performance (Algorithm Only):

   • Yes, the primary performance evaluation is standalone. The device ("BioCode RPP") is a qualitative multiplexed nucleic acid-based in vitro diagnostic test for use with the "BioCode MDx-3000 Instrument." The instrument integrates PCR, target capture, signal generation, and optical detection. The results are generated by the system (the "algorithm" in this context refers to the instrument's processing logic and interpretation algorithm), and the performance metrics (PA, NA) are reported for this system. There is no human interpretation component in the direct testing process that could be "assisted" by the algorithm.

6. Type of Ground Truth Used:

   • The ground truth for the clinical test sets (prospective and archived) was established by an FDA-cleared molecular multiplexed respiratory pathogen panel (Standard of Care).
   • For discrepant results, expert molecular laboratory retesting using analytically validated PCR followed by bi-directional sequencing assays and alternate NAATs served as the confirmatory ground truth. This is a form of expert consensus/confirmatory testing using highly accurate molecular methods.
   • For the analytical studies (e.g., LoD, inclusivity, specificity, inhibition), the ground truth was based on known concentrations of purified organisms/nucleic acids or genetically characterized strains (e.g., ATCC strains, Zeptometrix controls). This is a form of analytical ground truth where the content is precisely controlled and known.

7. Sample Size for the Training Set:

   • The document does not specify a separate "training set" sample size. For IVD devices, especially those based on molecular assays, a distinct "training set" as understood in machine learning (where an algorithm learns from labeled data) is not typically described in the same way. The device's underlying "algorithm" is the biochemical and optical detection system itself, and its performance characteristics are established through extensive analytical and clinical validation, not by training on a large dataset of patient samples in the AI sense. The design and validation are based on principles of molecular biology and traditional assay development.

8. How the Ground Truth for the Training Set Was Established:

   • As there isn't a "training set" in the machine learning sense, this question isn't directly applicable for this type of IVD device. The development of the assay (e.g., primer and probe design) would rely on known genetic sequences of the target pathogens. The validation data (clinical and analytical studies) demonstrate that the final, developed assay meets its intended performance, rather than being used to train a model in an iterative machine learning manner.

Ask a specific question about this device

The BioCode Gastrointestinal Pathogen Panel (GPP) is a qualitative multiplexed nucleic acid-based in vitro diagnostic test intended for use with the BioCode MDx 3000 Instrument. The BioCode GPP is capable of the simultaneous detection and identification of nucleic acids from multiple bacteria, viruses, and parasites extracted directly from unpreserved stool samples or stool preserved in Cary-Blair transport medium obtained from individuals with signs and/or symptoms of gastrointestinal infection. The following bacteria, parasites, and viruses are identified using the BioCode Gastrointestinal Pathogen Panel:

• · Campylobacter (C. jejuni/C. coli)
• · Clostridium difficile (C. difficile) toxin A/B (Fresh samples only)
• · Salmonella spp
• · Vibrio (V. parahaemolyticus/V. vulnificus/ V. cholerae), including specific identification of Vibrio parahaemolyticus
• · Yersinia enterocolitica
• · Enteroaggregative Escherichia coli (EAEC)
• · Enterotoxigenic Escherichia coli (ETEC) lt/st
• · E. coli 0157 serogroup
• Shiga-like toxin-producing Escherichia coli (STEC) stx1/stx2
• Shigella/ Enteroinvasive Escherichia coli (EIEC)
• · Cryptosporidium spp (C. parvum/C. hominis)
• Entamoeba histolytica
• · Giardia lamblia (also known as G. intestinalis and G. duodenalis)
• Adenovirus F 40/41
• Norovirus GI/GII
• Rotavirus A

The BioCode GPP is indicated as an aid in the diagnosis of gastrointestinal illness and results are meant to be used in conjunction with other clinical, laboratory, and epidemiological data.

Positive results do not rule out co-infection with organisms not included in the BioCode GPP. The agent detected may not be the definite cause of the disease. Negative results in the setting of clinical illness compatible with gastroenteritis may be due to infection by pathogens that are not detected by this test or non-infectious causes such as ulcerative colitis, irritable bowel syndrome, or Crohn's disease.

Concomitant culture is necessary for organism recovery and further typing of bacterial agents.

This device is not intended to monitor or guide treatment for C. difficile infection.

Due to the small number of positive specimens collected for certain organisms during the prospective clinical study, performance characteristics for Adenovirus 40/41, Campylobacter, E. coli 0157, Shigella/EIEC, Yersinia enterocolitica, and Giardia lamblia were established additionally with retrospective clinical specimens. Performance characteristics for Entamoeba histolytica, Giardia lamblia, Yersinia enterocolitica and Vibrio (V. parahaemolyticus, V. vulnificus, and V. cholerae) were established primarily using contrived clinical specimens.

The BioCode Gastrointestinal Pathogen Panel (GPP) is a multiplexed nucleic acid-based test designed to be used with the BioCode MDx 3000 system. The BioCode MDx 3000 is an automated system that integrates PCR amplification, target capture, signal generation and optical detection for multiple gastrointestinal pathogens from a single stool specimen, either unpreserved or in Cary Blair. Stool specimens are processed and nucleic acids extracted with the easyMAG and MagNa Pure. Once the PCR plate is set up and sealed, all other operations are automated on MDx 3000. The BioCode MDx 3000 Gastrointestinal Infection Panel simultaneously tests for 17 pathogens (see table below) from unpreserved stool specimens or stool collected in Cary-Blair transport medium. Results from the BioCode Gastrointestinal Pathogen Panel (GPP) test are available within less than 4 hours.

Here's a breakdown of the acceptance criteria and study information for the BioCode Gastrointestinal Pathogen Panel (GPP) as described in the provided document:

Acceptance Criteria and Device Performance for BioCode GPP (with MagNA Pure 96 extraction)

The document focuses on demonstrating the substantial equivalence of the BioCode GPP when used with the MagNA Pure 96 extraction system, compared to its previously cleared version with the easyMAG extraction system (K180041, the predicate device). Therefore, the acceptance criteria and performance are largely derived from the comparison to the existing predicate and the analytical studies.

General Acceptance Criteria (Implied from Clinical Performance Tables):
The acceptance criteria are generally implied to be high positive agreement (PPA) and negative agreement (NPA) with reference methods, typically in the range of 80-100% for PPA and 90-100% for NPA, within a 95% Confidence Interval. For analytical studies, reproducibility is expected to be high (>95%), and the limit of detection (LoD) should be comparable between extraction methods.

Table of Acceptance Criteria and Reported Device Performance (Summary):

2. Sample Size for the Test Set and Data Provenance:

• Clinical Test Set:

  • Archived Samples: 466 leftover, de-identified samples (275 frozen unpreserved and 191 inoculated Cary-Blair).
    • Provenance: Prospectively collected for the clinical study that resulted in the K180041 BioCode GPP clearance. This suggests the samples were originally collected for diagnostic purposes in a real-world setting, then archived and de-identified for the current study. The country of origin is not explicitly stated but is implied to be within the scope of previous FDA clearance.
  • Fresh Samples: 53 freshly collected leftover samples (specifically for C. difficile testing initially, but used for other targets as well).
    • Provenance: Freshly collected, though the specific clinical sites or countries are not mentioned.
  • Contrived Samples: 120 samples
    • Provenance: Primarily used to establish performance characteristics for Entamoeba histolytica, Giardia lamblia, Yersinia enterocolitica and Vibrio (V. parahaemolyticus, V. vulnificus, and V. cholerae) due to small numbers of naturally positive clinical specimens. These are laboratory-prepared samples.

• Analytical Test Set (Reproducibility Study): Consisted of 7 contrived samples, with combinations of 12 representative targets at 1.5x LoD (Low) and 3x LoD (Medium). Each sample was extracted in triplicate and assayed in singlet. This involved testing across 3 instruments by 3 operators, 2 runs per day for 5 days (total of 30 runs).

• Analytical Test Set (LoD Study):

  • Initial screening: 4 replicates of each concentration (near LoD) in negative stool and Cary-Blair, extracted on both easyMAG and MagNA Pure 96, and tested in singlet.
  • Confirmation: 20 replicates of each sample type/extraction method, tested in singlet, at or near presumptive LoD.
  • Norovirus GI and GII LoD used positive clinical specimens with serial dilutions.

3. Number of Experts Used to Establish Ground Truth for the Test Set and Qualifications:

The document mentions "Reference methods" which include culture, FDA cleared NAT, PCR/sequencing, and enrichment culture/cleared antigen test. While reference methods are used, the document does not explicitly state the number of experts used to establish the ground truth or their qualifications. It refers to "historical Reference results (K180041)" and "Reference assay". This implies that the ground truth was established by laboratory testing using established diagnostic methods, rather than clinical experts' consensus.

4. Adjudication Method for the Test Set:

The document describes discordant analysis for clinical samples. Specifically, "Sixty-four (64) archived samples with discordant results were retested twice with both easyMag and/or MagNA Pure 96 systems." The retesting appears to be the primary adjudication method. There is no mention of a particular adjudication method like 2+1 or 3+1 involving human experts. The "Comment" section for discordant results often indicates further molecular testing (PCR/Seq) to resolve discrepancies.

5. If a Multi-Reader Multi-Case (MRMC) Comparative Effectiveness Study Was Done, and Effect Size:

No, an MRMC comparative effectiveness study was not done. This device is an in vitro diagnostic test that provides qualitative detection of pathogens. Its performance is evaluated based on its agreement (sensitivity and specificity) with established reference methods, not on how it assists human readers in interpreting images or making diagnoses. Therefore, there is no discussion of human reader improvement with or without AI assistance.

6. If a Standalone (Algorithm Only Without Human-in-the Loop Performance) Was Done:

Yes, the studies are standalone performance evaluations of the BioCode GPP diagnostic test. The device itself is an automated system that performs nucleic acid extraction, amplification, detection, and decoding to provide qualitative results. The "Clinical Performance" and "Analytical Performance" sections directly assess the accuracy of the device's output against reference methods. There is no human-in-the-loop component in the direct operation or interpretation of the assay results in the way it would apply to, for example, an AI-powered image analysis tool. The results are interpreted by the MDx 3000 system.

7. The Type of Ground Truth Used:

The ground truth for the clinical studies was established using reference laboratory methods, including:

• Culture: For Campylobacter (C. jejuni, C. coli), Escherichia coli (E. coli) 0157, Salmonella, Shigella, Vibrio spp., Yersinia enterocolitica.
• FDA cleared NAT (Nucleic Acid Test): For Clostridium difficile toxin A/B.
• PCR/sequencing: For Adenovirus 40/41, Cryptosporidium (C. parvum, C. hominis), Entamoeba histolytica, Enteropathogenic E. coli (EPEC), Enterotoxigenic E. coli (ETEC) LT/ST, Enteroaggregative E. coli (EAEC), Giardia lamblia /intestinalis, Norovirus GI/GII, Rotavirus A.
• Enrichment culture/cleared antigen test: For Shiga-like Toxin producing E. coli (STEC) stx1/stx2.

Some ground truth for less common organisms (Entamoeba histolytica, Giardia lamblia, Yersinia enterocolitica and Vibrio) was established with contrived clinical specimens, where the pathogen presence and concentration are precisely known.

8. The Sample Size for the Training Set:

The document does not explicitly mention a "training set" in the context of machine learning for an AI device. This is a molecular diagnostic assay, not an AI/ML algorithm that requires training data in the typical sense. The studies presented are primarily for validation and verification of the device's performance. The "K180041 BioCode GPP clearance" mentioned as the source of archived samples suggests previous validation, but that wouldn't necessarily be considered a "training set" for the current device.

9. How the Ground Truth for the Training Set Was Established:

Since there is no explicit "training set" for an AI/ML algorithm mentioned, this question is not directly applicable. The performance is validated against clinical and analytical studies using reference methods as described in point 7.

Ask a specific question about this device