The T2Biothreat Panel is a qualitative, multiplexed, nucleic acid-based in vitro diagnostic test intended for use with the T2Dx Instrument. The T2Biothreat Panel detects nucleic acids from the following organisms directly from K2EDTA whole blood samples:

1. Bacillus anthracis (plasmids pXO1 and pXO2)
2. Francisella tularensis
3. Burkholderia spp. (B. mallei/B. pseudomallei)
4. Yersinia pestis
5. Rickettsia prowazekii

The T2Biothreat Panel will not distinguish between detection of Burkholderia mallei and Burkholderia pseudomallei but will present valid detections as a positive detection of Burkholderia species.

The T2Biothreat Panel is intended to test individuals with signs and symptoms of infection from biothreat agents and/or individuals who are at risk for exposure or may have been exposed to these agents. The T2Biothreat Panel is indicated as an aid in the diagnosis of anthrax, tularemia, melioidosis, glanders, typhus fever and plague in response to suspected or confirmed bioterrorism events or outbreaks. Diagnosis of infection must be made in conjunction with clinical, epidemiologic and other laboratory data. Results are for the presumptive identification of Bacillus anthracis, Francisella tularensis, Burkholderia spp. (B. mallei), Yersinia pestis and Rickettsia prowazekii. Results are meant to be used in conjunction with other clinical, epidemiologic, and laboratory data, in accordance with the guidel by the relevant public health authorities. The definitive identification of Bacillus anthracis, Francisella tularensis, Burkholderia mallei, Burkholderia pseudomallei, Yersinia pestis or Rickettsia prowazekii requires additional testing and confirmation procedures in consultation with the appropriate public health authorities for whom reports may be required. Positive results do not rule out co-infections with pathogens not included on the T2Biothreat Panel. Negative results do not preclude infection with the biothreat microbial agents targeted by the device and should not be used as the sole basis for diagnosis, treatment, or other patient management decisions.

The T2Biothreat Panel is indicated for use in laboratories that have the appropriate biosafety equipment, personal protective equipment (PPE), containment facilities, and personnel trained in the safe handling of clinical specimens potentially containing biothreat organisms.

The T2Biothreat Panel is indicated for use in laboratories that follow public health guidelines that address appropriate biosafety conditions, interpretation of test results, and coordination of findings with public health authorities.

This assay is not FDA-cleared or approved for testing blood or plasma donors.

The T2Biothreat Panel is a qualitative, multiplexed, nucleic acid-based in vitro diagnostic test intended for use with the T2Dx Instrument. The T2Biothreat Panel detects nucleic acids from the following organisms directly from K2EDTA whole blood samples: Bacillus anthracis (plasmids pXO1 and pXO2), Francisella tularensis, Burkholderia spp. (B. mallei/B. pseudomallei), Yersinia pestis, and Rickettsia prowazekii. The T2Biothreat Panel is run on the T2Dx, a fully automated, benchtop instrument. During processing on the T2Dx, intact pathogen cells are concentrated directly in whole blood, then lysed to release the target DNA. After amplification, target amplicon is hybridized with superparamagnetic particles and then detected by T2MR. The Internal Control on the T2Biothreat Panel monitors performance for each sample. The T2Biothreat Panel is a qualitative molecular diagnostic assay that employs whole blood compatible PCR amplification followed by T2 Magnetic Resonance (T2MR) detection. The T2Biothreat Panel is performed on the T2Dx Instrument, which executes all steps after specimen loading, with the capability of loading up to seven blood specimens at the same time. Individually, a KeEDTA whole blood specimen containing a minimum of 3 mL is loaded directly onto the T2Biothreat Sample Inlet, which is then placed on the T2Biothreat Cartridge along with the T2Biothreat Reagent Tray. The Cartidge and Reagent Tray contain the lysis reagent, internal control, primers, enzyme, buffer and probe-coupled superparamagnetic particles for each detected tarqet. After loading into the T2Dx. the blood specimen is mixed with the red blood cell lysing reagent and the bacterial cells and human cellular debris are concentrated by centrifygation. The internal control is added to the concentrated pellet and a bead-beating process lyses the bacterial cells. The supernatant containing the DNA from the lysed bacterial cells and the internal control is amplified using the target and internal control-specific primers. The generated amplified product is aliquoted into individual tubes containing target-specific probe conjugated particles for each detected target and the internal control. The amplified to target-specific probes attached to superparamagnetic particles causing clustering of the particles. The hybridization occurring in individual tubes is analyzed in the T2MR reader and a signal for each target is generated, which indicates the presence of the target organism(s). This automated process is the same process followed by the FDA cleared T2Candida and T2Bacteria Panels performed on the T2Dx Instrument system. When running a single specimens simultaneously, the first specimen will be reported in approximately 4 hours from the specimen is loaded onto the instrument. The results are interpreted by the device software as valid or invalid (based on the result of the internal control or target detections), and if valid, results are reported as "Positive" or "Target not Detected" for each specific target.

Here's a breakdown of the acceptance criteria and study information for the T2Biothreat Panel, based on the provided text:

Acceptance Criteria and Device Performance

Study Information

The provided document describes standalone performance studies for the T2Biothreat Panel. It does not mention any multi-reader multi-case (MRMC) comparative effectiveness study.

2. Sample Sizes and Data Provenance

• Test Set (Clinical Performance):
  • Negative Arm: Undisclosed number of K2EDTA whole blood samples from healthy donors (no signs/symptoms of infection) and febrile donors (fever ≥ 100.4 °F).
  • Positive Arm: Sequence-verified clinical bacterial strains spiked into whole blood collected from febrile donors. The sample sizes for the positive arm are listed as:
    • B. anthracis (pXO1 & pXO2): 6 (

Ask a specific question about this device

The BioFire® Global Fever Special Pathogens Panel is a qualitative, mucleic acid-based test intended for use with BioFire® FilmArray® 2.0 and BioFire® FilmArray® Torch Systems. The BioFire Global Fever Special Pathogens Panel is for the simultaneous qualitative detection and identification of multiple bacterial, viral, and protozoan nucleic acids directly from EDTA whole collected from individuals with signs and or symptoms of acute febrile illness and known or suspected exposure to the target pathogens described below.

Pathogens identified:
Chikungunya virus Dengue virus (serotypes 1, 2, 3 and 4) Leishmania spp. that cause visceral leishmaniasis (e.g., L. donovani and L. infantum) Leptospira spp. Plasmodium spp. (including species differentiation of Plasmodium falciparum and Plasmodium vivax/ovale) West Nile virus

Pathogens presumptively identified:
Bacillus anthracis Crimean-Congo hemorrhagic fever virus Ebolavirus spp. Francisella tularensis Lassa virus Marburgvirus Yellow fever virus Yersinia pestis

Pathogens for which the panel provides presumptive identification results resting and confirmation procedures in consultation with the appropriate public health authorities for whom reports may be necessary.

Positive results do not rule out co-infections with pathogens not included on the BioFire Global Fever Special Pathogens Panel. Not all pathogens that cause acute febrile illness are detected by this test, and nectude infection with the pathogens targeted by the device and should not be used as the sole basis for diagnosis, treatment, or other patient management decisions.

Evaluation for more common causes of acute illness (e.g., infections of the upper and lower respiratory tract or gastroenteritis, as well as non-infectious causes) should be considered prith this panel. In the United States, patient travel history, exposure risk, and consultation of the CDC Yellow Book should be considered prior to use of the BioFire Global Fever Special Pathogens Panel as some pathogens are more common in certain geographical locations. Results are meant to be used in conjunction with other clinical, epidemiologic, and laboratory data, in accordance with the guided by the relevant public health authorities.

The BioFire Global Fever Special Pathogens Panel is indicatories having appropriate biosafety equipment, personal protective equipment (PPE), contamment facilities and persomel trained in the safe handling of diagnostic clinical specimens potentially containing any of the pathogens detected by this panel.

The BioFire Global Fever Special Pathogens Panel is indicated for use in laboratories that follow public health guidelines that address appropriate biosafety conditions, interpretation of test results, and coordination of findings with public health authorities.

For In Vitro Diagnostic Use.

The BioFire® Global Fever Special Pathogens Panel is a multiplexed nucleic acid-based test designed to be used with BioFire® FilmArray® Systems (BioFire® FilmArray® 2.0 or BioFire® FilmArray® Torch). The BioFire Global Fever Special Pathogens Panel pouch contains freezedried reagents to perform nucleic acid purification and nested, multiplexed polymerase chain reaction (PCR) with DNA melt analysis. The BioFire Global Fever Special Pathogens Panel conducts tests for the identification of bacterial, viral, and protozoan pathogens from whole blood specimens collected in EDTA tubes (Table 1). Results from the BioFire Global Fever Special Pathogens Panel are available in about 1 hour.

Here's a breakdown of the acceptance criteria and study information for the BioFire Global Fever Special Pathogens Panel:

1. Table of Acceptance Criteria and Reported Device Performance

The application does not explicitly state "acceptance criteria" for PPA and NPA. However, it presents the clinical performance results in a way that suggests these are the key metrics for evaluating agreement with comparator methods. The "Expected percent agreement was >95%" in the reproducibility study might be inferred as a general target for performance. For this table, I'll use the Reported Clinical Performance Summary (Tables 4, 5, 6) as the "Reported Device Performance" against an implied high standard of agreement.

Note on "Acceptance Criteria": The document provides performance results but doesn't explicitly state quantitative acceptance criteria (e.g., "PPA must be >95%"). However, the high percentages and confidence intervals presented imply that high sensitivity and specificity are expected for proper function. The "Reproducibility" section mentions ">95%", which can serve as a proxy for the general expectation of performance accuracy.

2. Sample size used for the test set and the data provenance

• Prospective Clinical Study Test Set:
  • Sample Size: 2139 prospectively collected whole blood specimens.
  • Data Provenance: The specimens were collected between March 2018 and March 2021 from 11 undisclosed sites. The country of origin is not explicitly stated, but the mention of "CDC Yellow Book" in the "Indications for Use" suggests a relevance to North American (US) context, although the pathogens detected indicate global relevance. The data is prospective.
• Archived Specimen Study Test Set:
  • Sample Size: 416 archived specimens.
  • Data Provenance: The specimens were collected from undisclosed sites (Site 01: 199, Site 02: 82, Site 03: 135). The country of origin is not explicitly stated. The data is retrospective.
• Contrived Specimen Study Test Set:
  • Sample Size: 50 replicates for each analyte for which archived specimens were unavailable or insufficient. This resulted in varying total number of specimens tested per analyte (e.g., CCHF virus and Marburgvirus sp. had 100 replicates, others had 50).
  • Data Provenance: Contrived specimens were prepared using residual human whole blood. This is a laboratory-based study.

3. Number of experts used to establish the ground truth for the test set and the qualifications of those experts

The document does not specify the number or qualifications of experts used to establish the ground truth for the test set. Ground truth was established using "plate-based PCR comparator methods" and "additional PCR" for discrepant results. For archived specimens, they had "known analyte content" or "high likelihood of containing a given analyte." For contrived specimens, the "known composition of the contrived specimen" was the ground truth.

4. Adjudication method for the test set

The document describes "discrepancy testing" for samples where the BioFire Global Fever Special Pathogens Panel results differed from the initial comparator method results. For example:

• For Chikungunya virus, "Evidence of Chikungunya virus was found in 2/2 FP specimens by additional PCR."
• For Dengue virus, "15/17 FN specimens were positive upon BioFire Global Fever Special Pathogens Panel retest and by additional PCR, two were positive Global Fever Special Pathogens Panel retest, and eight were detected only by additional PCR."
• For Leptospira spp., "Evidence of Leptospira spp. was found in 1/1 FN specimens by BioFire Global Fever Special Pathogens Panel retest and by additional PCR, and in 3/4 FP specimens by additional PCR."
• For Plasmodium spp., similar retesting by the BioFire panel and "additional PCR" or "species-level comparator assay" was used.

This indicates an adjudication method that involves retesting with the device and/or additional PCR/comparator methods for discrepant results, rather than explicitly stating an "X+Y" consensus model among human experts.

5. If a multi-reader multi-case (MRMC) comparative effectiveness study was done, if so, what was the effect size of how much human readers improve with AI vs without AI assistance

No, a multi-reader multi-case (MRMC) comparative effectiveness study was not done. This device is a diagnostic assay (nucleic acid-based test), not an AI-assisted interpretation tool for human readers. Therefore, the concept of "human readers improve with AI vs without AI assistance" does not apply to this submission.

6. If a standalone (i.e. algorithm only without human-in-the-loop performance) was done

Yes, the performance studies (clinical, archived, contrived) evaluate the BioFire Global Fever Special Pathogens Panel as a standalone device (algorithm only). The device provides automated test interpretation and report generation, and the "user cannot access raw data" (Table 2). This means the performance metrics (PPA, NPA) reflect the algorithm's detection capabilities without human intervention in the interpretation process.

7. The type of ground truth used

The ground truth for the test sets was primarily established by:

• Comparator methods (e.g., plate-based PCR/additional PCR): For both prospective clinical and archived specimens.
• Known composition: For contrived specimens, where the analytes were intentionally spiked into the samples.
• Discrepancy testing: For cases where the device result and initial comparator result differed, further testing (usually additional PCR) was performed to resolve the discrepancy and establish the final ground truth.

8. The sample size for the training set

The document does not specify the sample size for a training set. The BioFire Global Fever Special Pathogens Panel is a diagnostic assay, and while its development would involve internal validation and optimization, the provided performance data relates to its analytical and clinical performance after development, rather than the data used for machine learning model training.

9. How the ground truth for the training set was established

Since a "training set" in the context of the requested information (e.g., for an AI model) is not explicitly mentioned or relevant to this type of device submission, the document does not describe how ground truth for a training set was established. The performance studies presented are for the finished device's evaluation against established laboratory methods.

Ask a specific question about this device

The Applied Biosystems Bacillus anthracis Detection Kit is a real-time polymerase chain reaction (PCR) test kit intended for the qualitative in vitro diagnostic (IVD) detection of target DNA sequences for Bacillus anthracis, or BA). The Applied Biosystems Bacillus anthracis Detection Kit is intended to test human whole blood (EDTA) specimens and blood culture specimens with growth detected by a continuous monitoring blood culture system. Blood culture specimens must be determined to contain gram-positive bacilli by Gram stain prior to testing of whole blood specimens must be performed concomitantly with standard of care blood culture.

The Applied Biosystems Bacillus anthracis Detection Kit is indicated for use in CLIA-certified high-complexity laboratories in response to a confirmed Bacillus anthracis event only in accordance with the guidelines provided by public health authorities prior to or during a public health emergency. Testing with the Applied Biosystems Bacillus anthracis Detection Kit must only be performed when public health authorities have determined the need for this test must only be used with specimens from individuals with clinical signs and symptoms of B. anthracis infection and who have either been exposed to B. anthracis or may have been exposed to B. anthracis.

The Applied Biosystems Bacillus anthracis Detection Kit is intended for use as an aid in the diagnosis of anthrax infection and results are for the presumptive identification of Bacillus anthracis. The diagnosis of B. anthracis infection must be made based on history, signs, symptoms, exposure likelihood, and other laboratory evidence, in addition to the identification of B. anthracis from cultures or directly from clinical specimens. The definitive identification of B. anthracis requires additional testing and confirmation procedures in consultation with the appropriate public health authorities for whom reports may be required.

The Applied Biosystems Bacillus anthracis Detection Kit has not been clinically evaluated with specimens collected from individuals with B. anthracis infection or those presumed to B. anthracis. 'B. anthracis Not detected' results do not preclude infection with Bacillus anthracis and should not be used as for diagnosis, treatment, or other patient management decisions.

Laboratories implementing this test must have the appropriate biosafety equipment, personal protective equipment (PPE), containment facilities and personnel trained in the safe handling of diagnostic clinical specimens potentially containing B. anthracis. Anthrax is a nationally notifiable disease caused by a biothreat microbial agent and must be reported to public health authorities.

The distribution of in vitro diagnostic devices for Bacillus spp. detection is limited to laboratories that follow public health guidelines that address appropriate biosafety conditions, interpretation of test results, and coordination of findings with public health authorities.

The Applied Biosystems Bacillus anthracis Detection Kit is intended for use with the AB1 7500 Fast Dx Real-Time PCR Instrument with analysis using the Applied Biosystems Bacillus anthracis Interpretive Software (BaIS).

The Applied Biosystems™ Bacillus anthracis Detection Kit is a multiplexed real-time polymerase chain reaction (PCR) test kit intended for the qualitative in vitro diagnostic (IVD) detection of target DNA sequences for B. anthracis. Reagents are lyophilized in a 96-well plate format as a fully formulated Mastermix and are stable at room temperature for up to one year. The kit is specifically designed for performing real-time PCR using the Applied Biosystems (ABI) 7500 Fast Dx instrument and software, with nucleic acids extracted from clinical specimens using a Qiagen manual extraction method or Roche MagNA Pure automated extraction methods. An automated interpretative software component (BalS) is included in the kit but supplied separately and operates on a computer(s) that is separate from the ABI 7500 Fast Dx computer.

Here's a breakdown of the acceptance criteria and the study proving the device meets them, based on the provided text:

1. Table of Acceptance Criteria and Reported Device Performance

The document doesn't explicitly present a formal "Acceptance Criteria" table with pass/fail thresholds for all aspects. Instead, it describes performance studies and their outcomes. Based on the performance data provided, we can infer the de facto acceptance associated with demonstrating strong performance indicators.

• Whole Blood: 150 CFU/mL
• Blood Culture, Aerobic: 10,000 CFU/mL
• Blood Culture, Anaerobic: 10,000 CFU/mL
  Roche MagNA Pure:
• Whole Blood: 50 CFU/mL
• Blood Culture, Aerobic: 2270 CFU/mL
• Blood Culture, Anaerobic: 3040 CFU/mL
  (All values are presumptively "met" the >95% detection criteria given they are reported as LODs). |
  | Analytical Inclusivity| 100% detection rate for B. anthracis strains carrying both plasmid targets. Correctly identify 'Bacillus anthracis suspected' for strains with only one plasmid target. | 100% detection rate for B. anthracis strains that carry both plasmid targets. For three strains known to carry only one plasmid target, the assay generated 'Bacillus anthracis suspected' results as expected. |
  | Analytical Exclusivity| High no detection rate (close to 100%) for non-target organisms. No false positives. | Initial no detection rate for exclusivity testing was 95.68%. Repeat testing was conducted, resulting in a final no detection rate of 100%. Some known Bacillus cereus strains (BAG1X1-1, 03BB102, and G-9241) that contain a pXO1-like plasmid were correctly identified as 'Bacillus anthracis suspected'. |
  | Interfering Substances| No significant interference from common substances unless noted with appropriate labeling limitations. | Identified specific technique-specific substances that interfere (e.g., MagNA Pure Wash Buffer, 10% Bleach, Ethanol >5%). Appropriate limitations were added to product labeling. |
  | Microbial Interference| 100% detection rate of B. anthracis in the presence of other clinically-relevant organisms. | 100% detection rate of B. anthracis in the presence of potentially interfering organisms in whole blood and blood culture. |
  | Reproducibility | Consistent results across operators, instruments, and reagent lots. Limited false negatives/positives in control samples. Positive percent agreement metrics indicate reproducibility. | One low positive sample returned negative. One high positive sample returned 'Bacillus anthracis suspected' due to pXO1 assay failure, but returned 'Bacillus anthracis detected' on repeat. No false positive events out of 270 negative PCR tests. 24 sample replicates (across all three lots) returned initial indeterminate results, all determined negative by supervisor review. No invalid results. |
  | Carry-Over/Cross-contamination| Minimal to no false positive events due to carry-over/cross-contamination. | Automated Method (MagNA Pure): 2 false positive events out of 135 negative samples (Specificity: 98.5%).
  Manual Method (Qiagen DSP): No false positive events out of 139 negative samples (Specificity: 100%).
  Contamination events identified on surfaces during testing, highlighting areas of risk. |
  | Clinical Specificity | 100% Negative Percent Agreement (NPA) for B. anthracis negative samples. | For 401 blood culture specimens and 439 whole blood specimens: 100% Negative Percent Agreement (NPA) when compared to the expected negative result. |
  | Clinical Sensitivity | High Positive Percent Agreement (PPA) for B. anthracis positive samples. | 96% Positive Percent Agreement (PPA) (CI: 90.4-96.5%) for 87 low-positive whole blood specimens. Three specimens initially generated 'Bacillus anthracis suspected' results. |

2. Sample Sizes Used for the Test Set and Data Provenance

• Limit of Detection (LOD): 1612 technical replicates were tested. Data provenance is not specified by country, but it's analytical data likely generated in-house at MRIGlobal (developer). Prospective (controlled spike-in experiments).
• Analytical Inclusivity: 24 different Bacillus anthracis strains tested in triplicate (24 strains * 2 extraction methods * 3 replicates = 144 technical replicates)
• Analytical Exclusivity: 154 non-target organisms tested in wet lab.
• Interfering Substances: 50 potentially interfering substances. Each substance tested in triplicate-paired samples (with and without B. anthracis). 50 * 2 * 3 = 300 samples (assuming paired means 3 concentrations per, and triplicate-paired means 3 replicates per substance per condition).
• Reproducibility: 7 panel members tested twice a day by three teams on five non-consecutive days. Details not fully specified for total sample count, but likely included multiple replicates for each panel member/condition. (e.g., 7 panel members * 2 times/day * 3 teams * 5 days = 210 runs, each run likely involving multiple replicates). "No false positive events occurred out of the 270 PCR tests of negative samples."
• Carry-Over/Cross-contamination:
  • Automated method (MagNA Pure): 9 runs, each with 31 samples, for a total of 279 samples. 135 negative samples.
  • Manual method (Qiagen DSP): 12 batches, total of 279 samples. 139 negative samples.
• Clinical Specificity: 401 blood culture specimens and 439 whole blood specimens. Samples were "left-over fresh and frozen blood culture samples," "randomly accessed, residual blood," and "febrile whole blood samples." Collected prospectively and serially from three point-of-care collection sites within the US.
• Clinical Sensitivity: 87 low-positive whole blood specimens (simulated/contrived). Each aliquot was spiked with 1 of 18 Bacillus anthracis strains. "Febrile whole blood specimens ... determined to be negative for Bacillus anthracis" were used as the matrix for spiking. Samples were contrived (prospective experimental design using collected negative clinical matrix). Clinical performance testing was conducted at three laboratories within the US.

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

The document does not explicitly state the number or qualifications of experts used to establish ground truth.

• For analytical studies (LOD, Inclusivity, Exclusivity, Interference, Reproducibility, Carry-Over), ground truth is established by the known concentration/presence of cultured organisms or purified DNA, or by the known absence of the target.
• For clinical specificity, samples were "assumed to be negative for Bacillus anthracis" based on their collection context (e.g., routine CBC, fever of unknown origin that would not indicate B. anthracis exposure, or confirmed to be positive for other bacterial species but not B. anthracis). No independent expert assessment or gold standard positive confirmation was used for these negative samples.
• For clinical sensitivity, the ground truth was established by artificial spiking of known concentrations of Bacillus anthracis into negative clinical samples.

4. Adjudication Method for the Test Set

The document mentions adjudication in two instances:

• Reproducibility: "24 sample replicates (across all three reagent lots) that returned initial indeterminate results. All were determined to be negative by supervisor review." This suggests a form of expert review for indeterminate results.
• Analytical Exclusivity: "Some known Bacillus cereus strains... Detection of these strains was indicated by a test result of ‘Bacillus anthracis suspected’ with amplification of the plasmid manually confirmed by the supervisor." This also indicates supervisor review for specific expected "suspected" results.

No broad 2+1 or 3+1 adjudication method for clinical test sets is described. For the clinical performance, the results are directly compared to the assumed/contrived ground truth.

5. If a Multi-Reader Multi-Case (MRMC) Comparative Effectiveness Study was Done

No MRMC comparative effectiveness study was done. This device is an in vitro diagnostic (IVD) PCR test kit, not an imaging AI designed to assist human readers. Its performance is assessed as a standalone diagnostic tool.

6. If a Standalone (i.e., algorithm only without human-in-the-loop performance) was Done

Yes, the studies presented (Analytical Sensitivity, Inclusivity, Exclusivity, Interference, Microbial Interference, Reproducibility, Carry-Over/Cross-contamination, Clinical Specificity, Clinical Sensitivity) represent the standalone performance of the "Applied Biosystems™ Bacillus anthracis Detection Kit" which includes the AB1 7500 Fast Dx Real-Time PCR Instrument and the Applied Biosystems Bacillus anthracis Interpretive Software (BaIS). It is an automated test with automated interpretation and report generation, so its performance is inherently a "standalone" or "algorithm only" type of assessment.

7. The Type of Ground Truth Used

The ground truth varied by study type:

• Analytical Studies (LOD, Inclusivity, Exclusivity, Interference, Microbial Interference, Reproducibility, Carry-Over): Controlled laboratory settings with known concentrations of purified DNA or cultured organisms. For negative controls, known absence of target.
• Clinical Specificity: "Assumed to be negative for Bacillus anthracis" for clinical samples collected from individuals not suspected of anthrax, or confirmed to have other bacteria. This is a clinical "expected negative" based on the patient population and standard clinical practice, rather than an independent gold standard for B. anthracis negativity.
• Clinical Sensitivity: Contrived samples where known concentrations of Bacillus anthracis were spiked into clinical samples previously determined to be negative for B. anthracis. This is a simulated positive ground truth.

8. The Sample Size for the Training Set

The document describes premarket validation studies for a diagnostic kit. It does not provide information about a "training set" size for model development, as this device is a PCR assay with interpretive software, not a machine learning model that undergoes a separate training phase with a distinct dataset of this nature. The "training" in this context refers to the development and optimization process, which is not typically quantified in terms of a "training dataset size" in the same way as for AI/ML algorithms.

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

As above, the concept of a "training set" and its ground truth establishment, as it applies to AI/ML, is not directly applicable to this PCR diagnostic kit. The "ground truth" during the development phase would have involved known positive and negative controls, spiked samples, and characterized bacterial strains, similar to the analytical studies performed for validation.

Ask a specific question about this device