(146 days)
The Trak® Male Fertility Testing System is intended for semi-quantitative assessment of sperm concentration at 15 million sperm per milliliter (M/mL) or below, between 15 and 55 M/mL, and above 55 M/mL. Sperm concentration is only one factor that could impact a man's fertility status and time to pregnancy. For complete assessment of male reproductive health, the user should consult a physician. For in vitro, over the counter home use.
The Trak® Male Fertility Testing System (Trak) includes a small instrument (the Engine), disposable units in which liquefied semen sample is introduced and the result is interpreted (the Props), and consumables, including collection cups and sample droppers. Trak uses the principle of density gradient separation to isolate sperm cells from human semen to provide an estimation of sperm concentration. The Trak Engine spins a test Prop to compact sperm cells within an introduced semen sample into a visible column (or "pellet"). The Prop gives a defined shape to the column, the height of which corresponds to the concentration of sperm cells in the sample.
Acceptance Criteria and Device Performance for Trak® Male Fertility Testing System
The Trak® Male Fertility Testing System is intended for semi-quantitative assessment of sperm concentration at three distinct levels: ≤ 15 M/mL, between 15 and 55 M/mL, and > 55 M/mL. The device must demonstrate sufficient accuracy and user interpretability for its intended over-the-counter home use.
1. Table of Acceptance Criteria and Reported Device Performance
The core acceptance criteria are based on the conditional probability of a correct Trak result when compared to a Computer Aided Semen Analysis (CASA) reference method.
| Category | Acceptance Criteria (Conditional Probability) | Reported Device Performance (95% CI) |
|---|---|---|
| ≤ 15 M/mL | Not explicitly stated as a numerical threshold, but implied to be high for substantial equivalence. | 93.3% (84.1 - 97.4%) |
| 15 – 55 M/mL | Not explicitly stated as a numerical threshold, but implied to be high for substantial equivalence. | 82.4% (73.3 - 88.9%) |
| > 55 M/mL | Not explicitly stated as a numerical threshold, but implied to be high for substantial equivalence. | 95.5% (88.9 - 98.2%) |
Additional Acceptance Criteria (Demonstrated in Non-Clinical Studies):
- Near-Cutoff Validation: Trak generates results adequately close to the 15 M/mL and 55 M/mL thresholds.
- Precision: The Trak system demonstrates adequate measurement precision for consistent semi-quantitative results.
- Consumer Interpretation: Lay users are able to correctly interpret Trak results, particularly near the 55 M/mL threshold (Overall Percent Agreement (OPA) of 97.4%).
- Interference Testing: Substances like saliva, urine, most microorganisms (E. coli, C. albicans, C. trachomatis, N. perflava), and hormones (testosterone, D-norgestrel, B-estradiol) do not interfere with Trak results at relevant concentrations. Note: 3 M/mL leukocytes and 1% whole blood failed acceptance criteria, requiring labeling limitations.
- QC Material Precision: QC material formulations meet acceptance criteria with 100% correct calls.
- Cleaning Robustness: The Trak Engine maintains performance after repeated cleaning and disinfection cycles.
- Prop Stability: Trak Props demonstrate real-time stability beyond the claimed expiry date.
- QC Material Stability: QC material demonstrates stability beyond the claimed expiry date.
2. Sample Size Used for the Test Set and Data Provenance
- Sample Size for Clinical Study (Test Set): 239 male subjects provided semen specimens for analysis.
- Data Provenance: The clinical study was a cross-sectional, multi-site investigation conducted at three clinical sites in the United States. Subjects were either presumptively healthy, a partner in a couple having difficulty conceiving, diagnosed with male factor infertility, post-vasectomy patients, or post-vasectomy reversal patients. This suggests a prospective study design.
3. Number of Experts Used to Establish the Ground Truth for the Test Set and Qualifications
- The document implies that one "HCP technician" per site analyzed an aliquot of the original semen specimen on a Computer Aided Semen Analysis (CASA) system to establish the ground truth.
- Qualifications: "HCP technician employed at the site". Specific qualifications (e.g., years of experience, specific certifications) are not detailed beyond "HCP technician."
4. Adjudication Method for the Test Set
- For the clinical study, the process involved:
- Subject/Tester analyzes semen with Trak and records result.
- Health-Care Professional (HCP) observes the Prop and records their own interpretation of the subject/tester's result.
- HCP performs their own Trak test using a saved aliquot and records the result.
- HCP technician analyzes an additional aliquot on a CASA system (ground truth).
- The primary comparison for the performance parameters (conditional probability) is between the Subject/Tester Trak results and the CASA reference method. The document does not explicitly state an adjudication method (like 2+1, 3+1) for resolving discrepancies between multiple expert interpretations of the CASA results or between the subject's Trak interpretation and the HCP's interpretation of the subject's Trak result. The CASA system is presented as the definitive ground truth. The HCP's interpretation of the subject's Trak result and their own Trak test seem to be for internal validation or comparison rather than a formal adjudication of the ground truth.
5. Multi-Reader Multi-Case (MRMC) Comparative Effectiveness Study
- A formal MRMC comparative effectiveness study, comparing human readers with and without AI assistance, was not explicitly described or performed.
- However, the Consumer Interpretation Study can be viewed as a form of reader study, assessing lay users' ability to interpret results. In this study, 61 lay subjects (readers) interpreted images of 7 Trak Props (cases). The "correct interpretation" served as the reference. The Overall Percent Agreement (OPA) was 97.4%. This study only involved interpretation of the device's output by lay users, not an "AI assistance" scenario for human readers evaluating a separate output.
6. Standalone Performance (Algorithm Only Without Human-in-the-Loop Performance)
- This device is not an AI algorithm. It is a physical diagnostic device that produces a visual output (a column of sperm cells) that is interpreted by the user. Therefore, a "standalone algorithm only" performance study is not applicable in the traditional sense of AI.
- The analytical sensitivity and precision studies, where Trak results were compared to CASA measurements, represent the closest equivalent to a "standalone" or objective performance evaluation of the device itself, independent of user interpretation (beyond the initial reading by lab personnel). These studies measured the device's output (column height, categorized into concentration ranges) rather than a human's interpretation of that output.
7. Type of Ground Truth Used
- For the clinical study (test set): Computer Aided Semen Analysis (CASA) results in M/mL were used as the reference method and ground truth.
- For non-clinical studies (e.g., analytical sensitivity, precision, interference): CASA results were also used as the ground truth. For some studies, "reference values" or "expected category" based on formulations were used, which were confirmed by CASA.
8. Sample Size for the Training Set
- The document describes premarket notification (510(k)) studies for a medical device that does not involve artificial intelligence or machine learning. Therefore, there is no specific "training set" in the conventional sense of machine learning. The device's operational principles (density gradient separation, centrifugal action) are based on established scientific principles rather than being "trained" on a dataset.
9. How the Ground Truth for the Training Set Was Established
- As there is no "training set" for an AI/ML algorithm, this question is not applicable. The device's design and parameters would have been developed and refined through engineering and analytical studies, likely using samples with known characteristics, but not in the format of an AI training set with ground truth labels.
§ 864.5220 Automated differential cell counter.
(a)
Identification. An automated differential cell counter is a device used to identify one or more of the formed elements of the blood. The device may also have the capability to flag, count, or classify immature or abnormal hematopoietic cells of the blood, bone marrow, or other body fluids. These devices may combine an electronic particle counting method, optical method, or a flow cytometric method utilizing monoclonal CD (cluster designation) markers. The device includes accessory CD markers.(b)
Classification. Class II (special controls). The special control for this device is the FDA document entitled “Class II Special Controls Guidance Document: Premarket Notifications for Automated Differential Cell Counters for Immature or Abnormal Blood Cells; Final Guidance for Industry and FDA.”