(63 days)
The Stryker CrossFlow Integrated Arthroscopy Pump is a dual arthroscopic pump system intended to provide fluid distension and irrigation of the knee, shoulder, hip, elbow, ankle and wrist joint cavities and fluid suction during diagnostic and operative arthroscopic procedures.
The Stryker CrossFlow Integrated Arthroscopy Pump (CrossFlow) is a microprocessorcontrolled dual (inflow and outflow) pump system designed to provide liquid distension and irrigation of joint cavities and aspiration of liquids out of the joint cavities during diagnostic and operative arthroscopy. Both the irrigation and aspiration pump of the device function according to the peristaltic principle. The Stryker CrossFlow Integrated Arthroscopy Pump consists of the following main components: console housing, power supply, two peristaltic pumps, three pinch valves, and a touch-screen display panel. The device is to be used with specially designed irrigation and aspiration tube sets and can be operated by remote hand and foot controls. A constantly-performed pressure sensing algorithm controls the value of the actual pressure in the joint cavity as compared to the set pressure determined by the user.
The proposed software modification consists of a graphical user interface (GUI) aesthetic update and the addition of a temperature estimation algorithm and on-screen indicator.
The provided document is an FDA 510(k) clearance letter for the Stryker CrossFlow Integrated Arthroscopy Pump, with a focus on a software modification (GUI aesthetic update and addition of a temperature estimation algorithm with an on-screen indicator).
This document does not contain the detailed, quantitative acceptance criteria and study results typically found for AI/ML-driven diagnostic or prognositic devices that require extensive performance data demonstrating accuracy in the context of human interpretation, such as sensitivity, specificity, AUC, or reader studies.
The device discussed here, the Stryker CrossFlow Integrated Arthroscopy Pump, is a medical device with a software component, not primarily an AI/ML-driven diagnostic/prognostic algorithm. The software modification is focused on estimating in-joint temperature and implementing mitigating actions to reduce high temperatures, and a GUI aesthetic update. The performance testing is engineering-centric, verifying that the temperature estimation algorithm meets accuracy specifications and reduces the occurrence of high fluid temperature.
Therefore, a significant portion of the requested information regarding AI/ML study specifics (e.g., sample size for test/training sets, data provenance, number of experts for ground truth, adjudication methods, MRMC studies, standalone performance, training ground truth establishment) is not applicable or not present in this type of 510(k) submission for this specific device.
However, I can extract and infer information based on the document related to the software modification's performance data:
Acceptance Criteria and Device Performance for Stryker CrossFlow Integrated Arthroscopy Pump (Software Modification)
Given the nature of the device as an arthroscopic pump with a new temperature estimation algorithm, the "acceptance criteria" discussed are largely functional and performance-based for the new software feature, rather than diagnostic accuracy metrics common for AI/ML image analysis.
1. Table of Acceptance Criteria and Reported Device Performance
| Acceptance Criteria (Inferred from "Software Testing" and "Mechanical Testing") | Reported Device Performance (Inferred from "Software Testing" and "Bench Performance Testing" sections) |
|---|---|
| Temperature estimation algorithm performs according to specification. | Software verification testing confirmed the algorithm and resulting mitigating actions performed according to specification. |
| Temperature estimation algorithm meets accuracy specifications. | Bench performance testing confirmed the temperature estimation algorithm meets accuracy specifications. |
| Algorithm reduces the occurrence of high fluid temperature. | Bench performance testing confirmed the algorithm reduces the occurrence of high fluid temperature. |
| Updated GUI allows users to successfully navigate and use the device as intended. | Design validation testing in a simulated-use environment showed surgeon and nurse users were successfully able to navigate the updated GUI and use the CrossFlow as intended. |
2. Sample size used for the test set and data provenance:
- Sample Size: Not explicitly stated as a number of "cases" or "patients" in the context of an algorithm evaluation study. Performance testing involved "software verification testing" and "bench performance testing," and "design validation testing" in a "simulated-use environment." The sample size refers to the number of tests performed or configurations evaluated, not a patient cohort. No specific numbers are provided beyond that these tests were conducted.
- Data Provenance: Not applicable. This is not a study derived from patient data in the sense of image analysis. The "data" tested are simulated or real-time measurements within the pump system during bench and simulated-use testing. It is implicitly prospective testing done in a lab/simulated environment. No country of origin for data is relevant as it's not a clinical data set.
3. Number of experts used to establish the ground truth for the test set and their qualifications:
- Number of Experts: Not explicitly stated for generating "ground truth" as it would be for diagnostic imaging. For "Design validation testing," it states it was conducted by "surgeon and nurse users." This implies clinical experts.
- Qualifications: "Surgeon and nurse users." No specific years of experience or board certifications are mentioned. Their role was to assess usability of the GUI and intended use, not to establish data-driven ground truth for temperature estimation. For the temperature estimation accuracy, the ground truth is likely established by precise measurement instruments in a lab setting, not human experts.
4. Adjudication method for the test set:
- Method: Not applicable. This is not a study requiring adjudication of human expert interpretations. The validation of the temperature algorithm would rely on objective physical measurements.
5. If a multi-reader multi-case (MRMC) comparative effectiveness study was done:
- Done?: No. This type of study (comparative effectiveness of human readers with/without AI assistance) is not applicable to an arthroscopic pump's temperature estimation algorithm and GUI update.
- Effect Size: Not applicable.
6. If a standalone (i.e. algorithm only without human-in-the loop performance) was done:
- Done?: Yes, to a degree. The "Software verification testing" and "Bench performance testing" of the "temperature estimation algorithm" are effectively standalone performance tests, ensuring the algorithm itself calculates and acts on temperature data correctly, independently of user interaction beyond setting parameters. The results state it "performed according to specification" and "meets accuracy specifications."
7. The type of ground truth used:
- Type:
- For the temperature estimation algorithm: Objective measurements from calibrated physical sensors during bench testing.
- For the GUI user experience: User feedback/success criteria from "surgeon and nurse users" in a "simulated-use environment."
8. The sample size for the training set:
- Sample Size: Not applicable/not disclosed. This is a software modification (algorithm update + GUI change) to an existing device, not an AI/ML system that undergoes traditional training on large datasets in the way a diagnostic image analysis algorithm would. The algorithm's parameters and logic would be developed and refined using engineering principles and calibration data, not a "training set" in the common AI/ML sense.
9. How the ground truth for the training set was established:
- Method: Not applicable. Given the nature of the device's software update, there isn't a "training set" with established "ground truth" in the manner of AI/ML diagnostic tools. The development of the temperature estimation algorithm would involve engineering design, thermodynamic modeling, and empirical calibration using temperature sensors, not a ground truth derived from clinical experts on a patient dataset.
§ 888.1100 Arthroscope.
(a)
Identification. An arthroscope is an electrically powered endoscope intended to make visible the interior of a joint. The arthroscope and accessories also is intended to perform surgery within a joint.(b)
Classification. (1) Class II (performance standards).(2) Class I for the following manual arthroscopic instruments: cannulas, currettes, drill guides, forceps, gouges, graspers, knives, obturators, osteotomes, probes, punches, rasps, retractors, rongeurs, suture passers, suture knotpushers, suture punches, switching rods, and trocars. The devices subject to this paragraph (b)(2) are exempt from the premarket notification procedures in subpart E of part 807 of this chapter, subject to the limitations in § 888.9.