FDA 510(k) Search - By Innolitics (SaMD and AI Experts)

The Roche Scripts for COBAS AMPLICOR CT/NG Test are intended to provide software scripts to direct the automated Tecan Genesis RSP 150 Workstation to process swab samples or control material for analysis using either of the following 510(k)-cleared assay test systems:

. COBAS AMPLICOR™ CT/NG test for Chlamydia trachomatis
COBAS AMPLICOR™ CT/NG test for Neisseria gonorrhoeae

Device Description

The Roche Scripts for COBAS AMPLICOR CT/NG Test accessory consists of a compact disc (CD) containing software scripts which direct the automated Tecan Genesis RSP 150 Workstation to process swab samples or control material for analysis.

AI/ML Overview

Here's an analysis of the provided text regarding the COBAS AMPLICOR CT/NG test for Neisseria gonorrhoeae with Roche Scripts Accessory, focusing on acceptance criteria and supporting studies.

Based on the provided K053289 document, the submission is primarily for the Roche Scripts for COBAS AMPLICOR CT/NG Test accessory, which consists of software scripts to automate sample processing for an already cleared assay. The acceptance criteria and study detailed in the document are therefore focused on demonstrating that the use of these scripts with the automated workstation does not negatively impact the performance of the existing COBAS AMPLICOR CT/NG test.

The document states: "We claim equivalence to the currently marketed COBAS AMPLICOR CT/NG test for Neisseria gonorrhoeae cleared under K974342." This implies that the performance of the new system (Automated workstation + Roche Scripts + COBAS AMPLICOR assay) must be comparable to the performance of the predicate device (Manual sample processing + COBAS AMPLICOR assay).

However, the provided text does not explicitly state specific numerical acceptance criteria or detail a study that directly reports device performance against such criteria for the Roche Scripts accessory itself. The document acts as a 510(k) summary, which generally provides an overview rather than full study reports.

Thus, I will describe the implied acceptance criteria based on the information provided, and what kind of study would typically be performed for such a submission, even if the specific data is not present in this summary.

Acceptance Criteria and Study Information for the Roche Scripts for COBAS AMPLICOR CT/NG Test

Given that this submission is for an accessory (software scripts) to automate sample processing for an existing, cleared diagnostic test, the primary "acceptance criterion" is demonstrations of substantial equivalence to the predicate device, specifically that the automation does not compromise the accuracy, precision, or overall performance of the existing assay.

Implied Acceptance Criteria and Reported Device Performance (based on principles of substantial equivalence for an automation accessory):

Acceptance Criteria Category	Specific Criterion (Implied)	Reported Device Performance (Inferred from 510(k) context)
Accuracy	The qualitative results (positive/negative for N. gonorrhoeae) obtained using the automated sample processing with Roche Scripts should be concordant with, and show no significant degradation compared to, the results obtained using manual sample processing with the predicate COBAS AMPLICOR CT/NG test for N. gonorrhoeae.	The 510(k) summary does not provide specific accuracy metrics (e.g., sensitivity, specificity, positive predictive value, negative predictive value) for the Roche Scripts accessory itself. However, the FDA's clearance implies that concordance studies were performed demonstrating comparable performance to the predicate device (K974342). For a device that automates a cleared assay, studies typically show high positive and negative agreement rates between automated and manual methods, often aiming for >95% agreement and demonstrating that the automated system does not introduce false positives or negatives beyond acceptable levels compared to the manual method.
Precision/Reproducibility	The automated sample processing with Roche Scripts should yield reproducible results (e.g., consistent positive/negative calls) when repeatedly testing the same samples or controls under varying conditions (e.g., different operators, different runs).	No explicit precision data related to the Roche Scripts is presented in this summary. However, comparable reproducibility to the predicate device would have been demonstrated to ensure substantial equivalence. This would involve showing consistent results across replicates and different testing scenarios when utilizing the automated workflow.
Functionality/Reliability	The Roche Scripts should successfully direct the Tecan Genesis RSP 150 Workstation to properly process swab samples and control material, without errors that lead to invalid results or interruptions. The software should perform its intended function as an accessory for sample preparation.	The 510(k) clearance itself indicates that the functionality and reliability of the software in directing the automated workstation were deemed acceptable, allowing for the intended sample preparation steps to occur correctly and interface with the COBAS AMPLICOR system without issue.

Study Details (Based on typical requirements for automation accessories, and inferences from the provided text):

Since the provided document is a 510(k) Summary, it contains limited detail about the actual study methodology and results. However, we can infer the type of study that would have been conducted for such a device: a method comparison or bridging study demonstrating that the automated sample processing system performs comparably to the manual (predicate) method.

Sample size used for the test set and the data provenance:
- Test Set Sample Size: Not explicitly stated in the summary. For a method comparison, typically hundreds to thousands of clinical samples (both positive and negative for N. gonorrhoeae) would be tested in parallel using both the automated and manual methods, alongside analytical samples (e.g., spiked samples, controls).
- Data Provenance: Not explicitly stated. Likely retrospective clinical samples, potentially supplemented by archived prospective samples or newly collected prospective samples. The country of origin is not mentioned, but given Roche Diagnostics' global operations, it could be multi-site, potentially including US sites.
Number of experts used to establish the ground truth for the test set and the qualifications of those experts:
- Number of Experts/Qualifications: This information is not provided for the Roche Scripts accessory. For diagnostic tests, ground truth for sample status (N. gonorrhoeae positive/negative) would typically involve a combination of:
  - Culture: Considered a gold standard for N. gonorrhoeae. Testing a subset of samples by culture would establish definitive ground truth.
  - Discrepancy Testing: For samples with discordant results between the candidate device (automated) and the predicate device (manual), or between either and culture, secondary, highly sensitive and specific molecular methods (e.g., a another PCR assay or sequencing) would be used, often by reference laboratories, to adjudicate the true status. This would be overseen by clinical microbiologists or infectious disease specialists.
Adjudication method for the test set:
- Not explicitly stated. For diagnostics comparing a new method to a predicate, and where culture or another highly reliable method isn't applied to all samples, the adjudication typically involves:
  - Discrepancy Resolution: Samples that yield different results between the new method and the predicate method, or between either method and an alternative reference method (if used), are re-tested, often using a "tie-breaker" method (e.g., a confirmatory PCR, sequencing, or a different validated assay) whose results are then used as the 'adjudicated truth' for that specific sample. There isn't typically a "2+1" or "3+1" expert consensus method as would be seen in image-based diagnostic AI.
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:
- Not applicable. This device is an in vitro diagnostic (IVD) for detecting bacterial DNA, not an imaging device or an AI application designed to assist human readers in interpretation. There are no "human readers" interpreting images or clinical data that would be assisted by this software. The software automates a laboratory process.
If a standalone (i.e. algorithm only without human-in-the-loop performance) was done:
- Yes, effectively. The primary "performance" studied for the Roche Scripts accessory would be its standalone ability to correctly direct the Tecan Genesis RSP 150 Workstation to prepare samples, and for the prepared samples to yield results comparable to manually processed samples when run on the COBAS AMPLICOR assay. The scripts themselves are not making diagnostic interpretations; they are automating a physical sample preparation process that leads to a result from the COBAS AMPLICOR instrument. The "algorithm" (scripts) performance is evaluated by the outcome of the sample preparation it controls.
The type of ground truth used (expert consensus, pathology, outcomes data, etc.):
- Likely a composite reference standard involving clinical culture, possibly combined with an alternative validated molecular diagnostic test (e.g., a different PCR assay), especially for samples with discordant results between the automated and manual methods, or for samples where culture is not feasible or sensitive enough (e.g., for certain N. gonorrhoeae sample types or low titers). The status of the patient (symptomatic/asymptomatic) is also considered in the context of the intended use, but the N. gonorrhoeae positivity is based on laboratory results.
The sample size for the training set:
- The provided document does not mention a training set sample size. For an automation accessory like this, there isn't a "training set" in the machine learning sense. The scripts are deterministic instructions. Development and verification of the scripts would involve internal testing, debugging, and validation using a variety of engineered and clinical samples, but this isn't typically termed a "training set" for statistical performance evaluation like a machine learning model.
How the ground truth for the training set was established:
- Not applicable as there is no "training set" in the context of machine learning for these deterministic software scripts. The functionalities directed by the scripts would have been verified during software development against engineering specifications and the known correct sequence of operations for sample processing.

Summary

Regulation Number and Section

§ 866.3390

Neisseria spp. direct serological test reagents.(a)
Identification. Neisseria spp. direct serological test reagents are devices that consist of antigens and antisera used in serological tests to identifyNeisseria spp. from cultured isolates. Additionally, some of these reagents consist ofNeisseria spp. antisera conjugated with a fluorescent dye (immunofluorescent reagents) which may be used to detect the presence ofNeisseria spp. directly from clinical specimens. The identification aids in the diagnosis of disease caused by bacteria belonging to the genusNeisseria, such as epidemic cerebrospinal meningitis, meningococcal disease, and gonorrhea, and also provides epidemiological information on diseases caused by these microorganisms. The device does not include products for the detection of gonorrhea in humans by indirect methods, such as detection of antibodies or of oxidase produced by gonococcal organisms.(b)
Classification. Class II (performance standards).