Search Results

The IA ALIS device is intended to measure alcohol in human breath. Measurements obtained by this device are used in the diagnostic of alcohol intoxication.

The IA ALIS is a self-contained device that uses a semiconductor-based gas sensor and disposable straws for breath sampling and three 7-segment LED displays present the detected alcohol level. On the upper part of the screen there are fifteen lights for visual indication of the level of alcohol detected. The LED screen displays the user's blood alcohol equivalent concentration in two decimal places. On the bottom part of the screen there are four boxes for user instructions (step 1, step 2, step 3 and Wait). The device starts working when money is inserted into the machine. After the WAIT light goes on and the system performs a cleaning process, the BLOW light goes on and there is a beep when the device is ready to accept an air sample. As the user exhales with the straw into the hole, the device monitors the air pressure and blowing time. If the flow of the breath sample is sufficient, the indicator BLOW light goes from flashing to solid and the WAIT light goes on. Thereafter the system processes the air sample and gives a numerical result.

Here's a summary of the acceptance criteria and the study that proves the device meets them, based on the provided text:

1. Table of Acceptance Criteria and Reported Device Performance

Note: The document explicitly states the IA ALIS met all of the applicable NHTSA requirements for precision and accuracy. The specific numerical results for temperature testing (e.g., "20/40") likely imply that across the 40 devices tested, 20 trials were performed for each BAC concentration at each temperature, and these trials met the criteria within the acceptable range (implied by the "met all applicable NHTSA requirements" statement).

2. Sample Size for the Test Set and Data Provenance

• Sample Size:
  • Precision/Reproducibility: 20 trials for 0.008% BAC, 20 trials for 0.032% BAC.
  • Blank Reading: 20 trials.
  • Cigarette Smoke Interference: 5 alcohol-free volunteers, 5 tests per volunteer per device cut-off (total tests not explicitly stated but implied to be sufficient for 25 tests for one cut-off).
  • Temperature: 40 devices, 20 trials at 0.008% BAC (total 800 trials), 20 trials at 0.032% BAC (total 800 trials). This was performed at 3 different temperatures (10°C, 25°C, 40°C).
  • Vibration: Not specified in terms of devices, but 20 samples using a Breath Alcohol Sample Simulator were performed for each BAC concentration (0.008% and 0.032%).
  • Consumer Study (Comparison with Predicate): 65 volunteers ranging from 25 to 65 years old (12 females, 53 males).
• Data Provenance: The document states "The sponsor performed their own testing and collected performance data." This suggests the data was generated specifically for this submission by the applicant. No specific country of origin is mentioned beyond "4376081 Canada Inc", implying the testing likely occurred in Canada or under their direct supervision. The nature of the testing suggests it was prospective as it involved designed test protocols and, for the consumer study, live participant data collection.

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

No human experts were used to establish the ground truth for the device's analytical performance (precision, accuracy, blank reading, temperature, vibration, cigarette smoke). The ground truth for these tests was established using:

• Breath Alcohol Sample Simulator (BASS), "MARK IIA SIMULATOR": This device provides an alcohol-in-air test sample with a known, simulated alcohol level (e.g., 0.008% BAC, 0.032% BAC, 0.00% BAC).
• Commercially available certified alcohol reference solution at 0.10% BAC: Used for device calibration, providing an external traceable reference for accuracy.

For the consumer study, the predicate device (AlcoMateCA2000), operated by a trained individual, served as a reference for comparison, but this wasn't ground truth in the traditional sense of an independent expert diagnosis.

4. Adjudication Method for the Test Set

Not applicable. The analytical performance tests used a simulated known concentration as the ground truth. The consumer study compared the device to a predicate, not to an adjudicated expert 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

No MRMC comparative effectiveness study was performed. This device is a standalone breath alcohol measurement system, not an AI-assisted diagnostic tool for human readers. A "consumer study" was performed, comparing the device's results to a predicate device and evaluating user comprehension and ease of use. This is not an MRMC study.

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

Yes, a standalone performance characterization was conducted. The precision, accuracy, blank reading, linearity, specificity, temperature, and vibration tests were all performed with the IA ALIS device acting as the sole measurement system against known, simulated alcohol concentrations. The results presented in Section M.1 ("Analytical Performance") directly reflect the algorithm's performance in isolation.

7. The Type of Ground Truth Used

The ground truth used for the analytical performance studies (precision, reproducibility, blank reading, temperature, vibration) was a simulated known alcohol concentration provided by a Breath Alcohol Sample Simulator (BASS) or a certified alcohol reference solution for calibration traceability. For the consumer study, the AlcoMateCA2000 predicate device provided the reference for comparison of BAC measurements.

8. The Sample Size for the Training Set

The document does not explicitly mention a "training set" in the context of machine learning or AI. The device uses a semiconductor gas sensor that functions based on physiochemical principles (Rs = A[C] * a). The document describes the sensor's operation but doesn't indicate a machine learning algorithm that would require a distinct training set. The device is calibrated at the factory using a certified alcohol reference solution.

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

Not applicable, as a machine learning-based training set is not indicated for this device. The device's operational parameters (like the relationship between sensor resistance and alcohol concentration) are based on the inherent properties of the SnO2 sensor and its calibration against certified reference solutions.

Ask a specific question about this device