(59 days)
The BD Phoenix™ Automated Microbiology System is intended for in vitro quantitative determination of antimicrobial susceptibility by minimal inhibitory concentration (MIC) of most gram-negative aerobic and facultative anacrobic bacteria isolates from pure culture for Enterobacteriaceae and Non-Enterobacteriaceae and most gram-positive bacteria isolates from pure culture belonging to the genera Staphylococcus and Enterococcus.
This premarket notification is for the Modifications to Phoenix System Software and for the BDXpert System resident on the EpiCenter System.
The BD Phoenix Automated Microbiology System (Phoenix System) is an automated system for the rapid identification (ID) and antimicrobial susceptibility testing (AST) of clinically relevant bacterial isolates. The system includes the following components:
- . BD Phoenix instrument and software.
- BD Phoenix panels containing biochemicals for organism ID testing and antimicrobial agents . or AST determinations.
- BD Phoenix ID Broth used for performing ID tests and preparing AST Broth inoculum. ●
- . BD Phoenix AST Broth used for performing AST tests only.
- . BD Phoenix AST Indicator solution added to the AST Broth to aid in bacterial growth determination.
Software incorporated into the BD Phoenix™ Automated Microbiology System provides the user with the ability to process Phoenix ID and/or Phoenix AST panels in a walk-away fashion. The software allows for the establishment of a panel test, automated/unattended testing of the panel and the ability to review the panel results. In order to accomplish the overall general purpose of the instrument, the software provides users with the ability to interact with the instrument, when necessary to facilitate the testing that they require. The Phoenix software includes the BDXpert System which is a collection of rules intended to evaluate results for specific organism-drug combinations.
The BDXpert System that is resident on the EpiCenter System is basically the same Expert system that exists on the Phoenix System. However, the BDXpert System utilizes additional information, such as specimen site, to provide the user with a more refined SIR interpretation.
The Phoenix software was developed and tested according to its Software Dcvelopment Plan and Software Verification and Validation Plan. The software test results demonstrated that the software operated in accordance to the software specifications
The Phoenix panel is a sealed and self-inoculating molded polystyrene tray with 136 micro-wells containing dried reagents. Organisms for susceptibility testing must be a pure culture and preliminarily identified as a Gram-negative or Gram-positive isolate. For each isolation equivalent to a 0.5 McFarland standard is prepared in Phocnix ID broth.
The Phoenix AST mothod is a broth based microdilution test. The Phoenix system utilizes a redox indicator for the detection of organism growth in the presence of an antimicrobial agent. Measurements of changes to the indicator as well as bacterial turbidity are used in the determination of bacterial growth. Each AST panel configuration contains several antimicrobial agents with a wide range of two-fold doubling dilution concentrations.
The instrument houses the pancls where they are continuously incubated at a nominal temperature of 35°C. The instrument takes readings of the pancls every 20 minutcs. The readings are interpreted to give an identification of the isolate, minimum inhibitory concentration (MIC) values and category interpretations, S, I, or R (sensitive, intermediate, or resistant).
The provided text primarily focuses on the 510(k) submission for modifications to the BD Phoenix™ Automated Microbiology System software and the BDXpert System. It does not contain detailed information about the acceptance criteria or a specific study proving the device meets those criteria.
However, based on the context of a 510(k) submission for an antimicrobial susceptibility test system, we can infer some general acceptance criteria and the type of "study" implied for regulatory clearance.
Here's an attempt to answer your questions based on the available information and general regulatory practices for such devices:
1. A table of acceptance criteria and the reported device performance
The document does not explicitly state acceptance criteria or provide a table of performance metrics. For devices of this type, acceptance criteria typically involve demonstrating substantial equivalence to a predicate device, specifically in terms of:
- Accuracy of Antimicrobial Susceptibility Testing (AST) results: This usually includes agreement rates for MIC values and categorical interpretations (Sensitive, Intermediate, Resistant) compared to a gold standard method (e.g., CLSI reference broth microdilution).
- Accuracy of organism identification (ID): Agreement rates with a reference identification method.
- Reliability and functionality of software: Demonstrated through software verification and validation.
The document states:
- "The Phoenix software was developed and tested according to its Software Development Plan and Software Verification and Validation Plan. The software test results demonstrated that the software operated in accordance to the software specifications."
This implies that the software met its specified performance, but specific numerical criteria and results are not provided.
2. Sample size used for the test set and the data provenance (e.g. country of origin of the data, retrospective or prospective)
The document does not specify the sample size for any test set or the data provenance. It mentions "The Phoenix AST method is a broth based microdilution test," but does not give details about how many isolates or tests were included in the validation for this specific submission.
3. Number of experts used to establish the ground truth for the test set and the qualifications of those experts (e.g. radiologist with 10 years of experience)
This information is not provided. For AST systems, the "ground truth" (or reference method) is typically established by trained microbiology laboratory personnel following standardized protocols (e.g., CLSI guidelines for broth microdilution). It doesn't usually involve "experts" in the same way as, for example, image interpretation.
4. Adjudication method (e.g. 2+1, 3+1, none) for the test set
This information is not provided. Adjudication methods are more common in studies involving subjective interpretations, such as radiology. For AST, the reference method provides a more objective "ground truth."
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
An MRMC study is not mentioned, and such a study is not typically applicable to an automated microbiology system like the BD Phoenix. This device automates the testing process and provides results, reducing direct "human reader" interpretation in the classic sense of a diagnostic image. The BDXpert System provides "a more refined SIR interpretation" by utilizing additional information like specimen site, which is a form of decision support, but not a "human reader improvement with AI" study in the MRMC context.
6. If a standalone (i.e. algorithm only without human-in-the-loop performance) was done
The BD Phoenix system itself is essentially a standalone (algorithm only) device in its core function of determining MICs and SIR interpretations. While a human initiates the test and reviews results, the determination of susceptibility is automated. The BDXpert System is described as "a collection of rules intended to evaluate results for specific organism-drug combinations," which is an algorithmic component operating without continuous direct human intervention during the interpretation process. The submission details "Software incorporated... provides the user with the ability to process Phoenix ID and/or Phoenix AST panels in a walk-away fashion." This confirms standalone algorithmic performance for test processing and initial interpretation.
7. The type of ground truth used (expert consensus, pathology, outcomes data, etc.)
For antimicrobial susceptibility testing, the ground truth is typically established using a Clinical and Laboratory Standards Institute (CLSI) reference broth microdilution method or other recognized reference methods for antimicrobial susceptibility testing. This is the industry standard for validating AST devices. The document states: "The Phoenix AST method is a broth based microdilution test," implying it's designed to align with such standards, and its performance would be compared against a reference method.
8. The sample size for the training set
The document does not provide any information about a "training set." This type of device, particularly the software modifications described, is likely validated against established performance criteria and reference methods rather than through a machine learning training paradigm that requires a distinct training set. The BDXpert System, described as "a collection of rules," suggests a rule-based expert system rather than a machine learning model that would be "trained" on data.
9. How the ground truth for the training set was established
Since no training set is mentioned in the provided text, this question cannot be answered.
§ 866.1645 Fully automated short-term incubation cycle antimicrobial susceptibility system.
(a)
Identification. A fully automated short-term incubation cycle antimicrobial susceptibility system is a device that incorporates concentrations of antimicrobial agents into a system for the purpose of determining in vitro susceptibility of bacterial pathogens isolated from clinical specimens. Test results obtained from short-term (less than 16 hours) incubation are used to determine the antimicrobial agent of choice to treat bacterial diseases.(b)
Classification. Class II (special controls). The special control for this device is FDA's guidance document entitled “Class II Special Controls Guidance Document: Antimicrobial Susceptibility Test (AST) Systems; Guidance for Industry and FDA.”