(422 days)
The Pivot Breath Sensor is a breath carbon monoxide monitor intended for single-user use by cigarette smokers as an educational and motivational tool to inform the user about how breath carbon monoxide levels are affected by smoking behavior. The device is not intended to be used with other combustible, inhaled products.
The Pivot Breath Sensor is a personal, portable, lithium ion battery powered breath carbon monoxide ("CO") monitoring device that measures the level of CO in an individual's exhaled breath. It is intended for single-user over-the-counter ("OTC") use by cigarette smokers (users) to measure CO levels in their exhaled breath. This parameter correlates closely with carboxyhemoglobin levels and with cigarette smoking behavior. Hence, the more a person smokes, the higher are their exhaled breath CO levels. The user submits a breath sample by exhaling (blowing) into the mouthpiece of the Pivot Breath Sensor which is directed over electrochemical sensors to quantify the CO level in the breath. The sensor has two buttons - a front, center button and a side button - to help with user inputs and navigation. It also has a rechargeable battery that can be charged using a micro-USB cable by plugging into USB compatible charging sources such as a computer, USB adapter for power outlet, or car USB port. The calculated CO concentration/ level of the exhaled breath is displayed to the user in whole number parts-per-million ("ppm") on the LCD screen of the sensor. The Pivot breath sensor measures and displays CO concentrations from 0 to 100 ppm. Each of the breath sample results is shown to the user with a corresponding color and a number. The color is intended to aid in giving context to the quantitative CO value, aligning with the predicate device's color coding and scientific literature. The sensor can display multiple samples as the CO log and helps to graphically show the user their relative levels of exhaled breath CO throughout the day and between days. Hence, periodic measurements of CO levels may provide users with feedback regarding their smoking exposure, thus helping them to become educated and motivated to quit smoking, as supported by reference literature.
The provided text details the 510(k) summary for the Carrot Inc. Pivot Breath Sensor. Here's a breakdown of the acceptance criteria and the study that proves the device meets them:
1. Table of Acceptance Criteria and Reported Device Performance
The regulatory document outlines several studies and their success criteria, which act as acceptance criteria.
| Acceptance Criteria Study | Objective | Success Criteria | Reported Device Performance |
|---|---|---|---|
| Bench Tests | |||
| Shelf Life | Not explicitly stated, implied to ensure device remains functional for the specified duration. | Passed with 18 month shelf life | Passed with 18 month shelf life. |
| Biocompatibility | Ensure materials are safe for human contact. | Passed ISO-10993 tests for cytotoxicity, sensitization and irritation. | Passed ISO-10993 tests for cytotoxicity, sensitization and irritation. |
| Software Validation | Ensure firmware functions correctly. | Passed unit, integration and system testing of firmware. | Passed unit, integration and system testing of firmware. |
| Wireless Coexistence | Ensure device operates without interference. | Passed requirements. | Passed requirements. |
| EMC testing | Ensure electromagnetic compatibility. | Passed ISO 60601 testing requirements. | Passed ISO 60601 testing requirements. |
| Sensor Performance | Ensure accuracy, precision, linearity, and cross-sensitivity. | Passed testing related to accuracy, precision, linearity and cross sensitivity. Testing included multiple lots at various temperature and humidity conditions. | Passed testing related to accuracy, precision, linearity and cross sensitivity. Testing included multiple lots at various temperature and humidity conditions. |
| Interfering Gases | Evaluate impact of other gases on sensor readings. | Completed testing of interfering gases and included in labeling where applicable. | Completed testing of interfering gases and included in labeling where applicable. |
| Hardware Verification | Ensure hardware and battery life meet specifications. | Passed hardware and battery life related testing. | Passed hardware and battery life related testing. |
| Packaging Testing | Ensure device integrity during shipping and handling. | Passed functionality testing after being subjected to ISTA 3A conditioning. | Passed functionality testing after being subjected to ISTA 3A conditioning. |
| Device Use Life | Evaluate long-term performance under repeated use. | Passed long-term repeated use testing. | Passed long-term repeated use testing. |
| Clinical Studies | |||
| 18-RP-1061A (Human Factors) | Assess whether an untrained lay user group can operate the device and interpret results correctly using only provided instructions. Validate appropriate mitigations of use-related hazards. | Ensure that untrained lay users can properly operate the device, and can interpret the results correctly using only the labeling to be provided. Validate appropriate mitigations of use-related hazards identified in risk management documentation. | Found the device to be safe and effective for the intended users, uses, and use environments. All participants, overall, were observed to safely perform critical tasks. |
| 18-RP-1062A (Comparative Performance) | Assess correlation between measured CO levels of the Pivot Breath Sensor (self-trained user) and a prescription-use CO breath sensor (trained healthcare professional guidance). | Based on the null hypothesis that the Pearson correlation coefficient of prescription device and Pivot Breath Sensor is 0.90 and the alternative hypothesis that it is >0.90, passing criterion is refuting the null hypothesis with a power of ≥90% assuming an 0.05 alpha level. | Using regression analysis, the 70 paired measurements of CO from Pivot Breath Sensor and the prescription device produced a line with a slope of 0.9202, a y-intercept of 0.0041 and a correlation coefficient of 0.9710. |
| 20-RP-1083A (Expanded Indications) | Assess changes in attitudes and understanding towards quitting smoking as well as smoking behavior change with use of the Pivot breath sensor. | Primary: Assess change in the proportion of participants' Stage of Change response at day 28 versus baseline. | |
| Secondary: Proportion of participants who report ≥ 1 quit attempt by day 28, and proportion of participants who reduce their CPD by ≥ 50% by day 28, compared to baseline. | Primary: Motivation to quit smoking improved in a statistically significant manner, with 38.9% of subjects at day 28 indicating they were thinking of quitting in the next 30 days versus 14.4% at baseline. At 28 days, motivation to quit smoking increased in 29.6%, was unchanged in 66.7%, and decreased in 3.7% of subjects. | ||
| Secondary: By day 28, 28.2% of the intent to treat (ITT) population reported making ≥ 1 quit attempt, and 23.1% reduced their CPD by ≥ 50% compared to baseline. |
2. Sample Size Used for the Test Set and the Data Provenance
- 18-RP-1061A (Human Factors):
- Sample Size: 15 subjects.
- Data Provenance: Prospective, single-center study. The document does not specify the country of origin but implies it was conducted under the direct supervision of Carrot Inc. or a contracted research institution.
- 18-RP-1062A (Comparative Performance):
- Sample Size: 70 subjects.
- Data Provenance: Prospective, single-center study. No country of origin is specified.
- 20-RP-1083A (Expanded Indications):
- Sample Size: 234 subjects, split into two cohorts (40-60% smoking 10-19 CPD, 40-60% smoking 20+ CPD).
- Data Provenance: Prospective, single-center study. No country of origin is specified.
3. Number of Experts Used to Establish the Ground Truth for the Test Set and the Qualifications of Those Experts
The document does not explicitly state the number or qualifications of experts used to establish ground truth for the test sets in the same way one might describe for an imaging AI device. Instead, the "ground truth" for each study is inherent to its design:
- 18-RP-1061A (Human Factors): The ground truth was the observable ability of untrained lay users to operate the device and interpret results against explicit instructions, likely assessed by study personnel. No specific "experts" are mentioned for establishing ground truth in this context.
- 18-RP-1062A (Comparative Performance): The ground truth was established by a prescription-use CO breath sensor used with guidance by a trained health care professional. This serves as the reference standard against which the Pivot Breath Sensor's performance was compared. The number and qualifications of these healthcare professionals are not specified, but they are implied to be "trained."
- 20-RP-1083A (Expanded Indications): The ground truth was self-reported data from participants regarding their "Stage of Change" for quitting smoking, quit attempts, and reduction in cigarettes per day (CPD) compared to baseline. No external "experts" were used to establish this ground truth; it was based on participant responses.
4. Adjudication Method for the Test Set
The document does not describe a formal adjudication method (like 2+1, 3+1 consensus by multiple readers) typically found in AI imaging studies. The studies described are either:
- Observational (Human Factors): Study personnel observed participants' interactions.
- Comparative Measurement (Comparative Performance): Direct comparison of measurements from two devices.
- Self-Reported Outcomes (Expanded Indications): Based on participant responses and observed changes.
Therefore, "none" in the traditional sense of expert adjudication for diagnostic discrepancies.
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, an MRMC comparative effectiveness study of human readers with vs. without AI assistance was not done. The Pivot Breath Sensor is a direct-to-consumer device that provides readings to a single user, not an AI system assisting human experts in making a diagnosis.
6. If a Standalone (i.e., algorithm only without human-in-the loop performance) was done
Yes, in the context of this device, the "standalone" performance is effectively the performance of the Pivot Breath Sensor itself.
- The "Sensor Performance" bench tests (accuracy, precision, linearity, cross-sensitivity) are a direct assessment of the device's standalone algorithmic and hardware performance against controlled environments and known gas concentrations.
- The 18-RP-1062A (Comparative Performance) study also assesses the standalone performance of the Pivot Breath Sensor by comparing its readings (taken by self-trained users) to a "prescription-use CO breath sensor submitted with guidance by a trained health care professional." The correlation coefficient of 0.9710 indicates a strong standalone performance in agreement with a reference standard.
7. The Type of Ground Truth Used
- Bench Tests: Controlled laboratory measurements and standards for physical and electrical properties, and established ISO standards for biocompatibility.
- 18-RP-1061A (Human Factors): Observational data of user interaction against predefined criteria for correct operation and interpretation.
- 18-RP-1062A (Comparative Performance): Measurements from a "prescription-use CO breath sensor submitted with guidance by a trained health care professional." This acts as the clinical gold standard for CO breath measurement.
- 20-RP-1083A (Expanded Indications): Self-reported outcomes data from participants regarding their smoking behavior, motivation to quit, and quit attempts.
8. The Sample Size for the Training Set
The document does not explicitly mention a "training set" in the context of an AI/ML algorithm development as one would typically see for complex learning models. Given that the device relies on electrochemical sensors, the "training" analogous to machine learning would typically involve:
- Sensor calibration: Manufacturers perform extensive calibration on sensor batches using known gas concentrations. This is implied by the "Sensor Performance" bench tests covering multiple lots.
- Algorithm development: The internal algorithm that converts sensor signals to ppm readings is developed and refined based on engineering principles and empirical data, not necessarily a separate "training set" in the common AI sense.
Therefore, a specific "training set sample size" as might be used for supervised machine learning is not provided or applicable in the traditional sense for this type of device.
9. How the Ground Truth for the Training Set Was Established
As explained above, a distinct "training set" with ground truth in the AI/ML sense is not described. For sensor calibration and algorithm development, the "ground truth" would be established through:
- Known gas concentrations: When calibrating electrochemical sensors, the device is exposed to precise, certified concentrations of carbon monoxide (and potentially interfering gases). These known concentrations serve as the ground truth for fine-tuning the sensor's response curve and the device's internal conversion algorithm.
- Engineering and chemical principles: The fundamental operation of an electrochemical sensor is based on established scientific principles, which guide the development of the algorithms that translate raw sensor signals into ppm values.
§ 868.1430 Carbon monoxide gas analyzer.
(a)
Identification. A carbon monoxide gas analyzer is a device intended to measure the concentration of carbon monoxide in a gas mixture to aid in determining the patient's ventilatory status. The device may use techniques such as infrared absorption or gas chromatography.(b)
Classification. Class II (performance standards).