(140 days)
The AHI System is intended for use by healthcare professionals managing patients 18 years or older who are receiving continuous physiological monitoring with electrocardiography (ECG) in hospitals.
AHI provides a frequently updated binary output over time based on pattern analysis of a lead-II ECG waveform intended to describe a patient's hemodynamic status and indicate if a patient is showing signs of hemodynamic stability or instability. Signs of hemodynamic instability (HI) are defined as hypotension (systolic blood pressure
The AHI System is a multiparameter system designed to meet clinicians' need to identify patient hemodynamic status and predict patient hemodynamic instability episodes using two analytics:
-
Analytic for Hemodynamic Instability (AHI) (as granted in DEN200022): Utilizing data from a single existing lead of a non-invasive electrocardiograph (ECG), AHI analyzes heart rate variability (HRV) and ECG morphology features to rapidly detect signs of hemodynamic stability or instability and categorize each window of data as either "AHI Stable" or "AHI Unstable." Time trending of AHI outputs is also provided.
-
Analytic for Hemodynamic Instability Predictive Indicator (AHI-PI): Utilizing AHI outputs from up to the most recent 30 minutes of ECG data, AHI-PI indicates the likelihood of a future episode of hemodynamic instability, defined as ten continuous minutes or more where signs of hemodynamic instability are present.
Here's a breakdown of the acceptance criteria and the study proving the device meets them, based on the provided text:
1. Table of Acceptance Criteria and Reported Device Performance
| Acceptance Criteria | Reported Device Performance (AHI-PI) |
|---|---|
| Distinguish Risk Levels for Hemodynamic Instability | Probability of HI in next 1 hour: |
- Low Risk (Green): 0.7% (CI: 0.4%, 1.3%)
- Moderate Risk (Yellow): 6.5% (CI: 3.7%, 10.3%)
- High Risk (Red): 35.9% (CI: 28.1%, 44.0%) |
| Likelihood of HI compared to Low Risk | - Moderate Risk: 9x more likely to have an episode of HI in the next 1 hour than Low Risk. - High Risk: 51x more likely to have an episode of HI in the next 1 hour than Low Risk. |
| Prediction of patient deterioration (among those who had HI) | 89% of cases correctly predicted. |
| Lead Time for Prediction | Median lead time of 48 minutes. |
| Clinical Benefit | Demonstrate significant discrimination between risk of future hemodynamic instability events, providing adjunctive information to clinicians to facilitate fewer missed diagnoses of emerging HI/patient deterioration. |
2. Sample Size Used for the Test Set and Data Provenance
- Sample Size: 65,969 "windows" (data points representing a period of time for which an AHI-PI indicator was generated). The exact number of patients is not explicitly stated, but the note mentions: "The unit of analysis here is at the windows level and not the patient level." The study used bootstrapping to account for multiple measurements per subject.
- Data Provenance: Prospectively collected from consecutive patients at Michigan Medicine. This indicates the data is from the United States and is prospective.
- Targeted Patient Population: Hospitalized patients 18 years or older who are receiving continuous physiological monitoring with electrocardiography (ECG) and are not contraindicated. Due to study design considerations, the primary study population was limited to patients who were invasively monitored with an arterial line.
3. Number of Experts Used to Establish the Ground Truth for the Test Set and Qualifications of Those Experts
- The document does not mention the use of experts to establish the ground truth for the test set.
4. Adjudication Method for the Test Set
- The document does not mention an adjudication method for the test set.
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 is mentioned for the AHI-PI system. The study focuses on the standalone performance of the AHI-PI algorithm. The device is intended for "adjunctive use" meaning it supports human healthcare professionals, but specific studies on human improvement with AI assistance are not described here.
6. If a Standalone (i.e., algorithm only without human-in-the-loop performance) was done
- Yes, a standalone performance study was done for the AHI-PI. The reported data in the table directly reflects the algorithm's ability to predict hemodynamic instability without human intervention in the prediction process itself. The study compared AHI-PI outputs to a vital signs reference standard.
7. The Type of Ground Truth Used (expert consensus, pathology, outcomes data, etc.)
- Outcomes Data/Physiological Reference Standard: The ground truth for hemodynamic instability (HI) was defined using a "hemodynamic vital signs reference standard" based on continuous vital signs:
- Hypotension (systolic blood pressure
§ 870.2220 Adjunctive hemodynamic indicator with decision point.
(a)
Identification. An adjunctive hemodynamic indicator with decision point is a device that identifies and monitors hemodynamic condition(s) of interest and provides notifications at a clinically meaningful decision point. This device is intended to be used adjunctively along with other monitoring and patient information.(b)
Classification. Class II (special controls). The special controls for this device are:(1) Software description, verification, and validation based on comprehensive hazard analysis and risk assessment must be provided, including:
(i) Full characterization of technical parameters of the software, including algorithm(s);
(ii) Description of the expected impact of all applicable sensor acquisition hardware characteristics on performance and any associated hardware specifications;
(iii) Specification of acceptable incoming sensor data quality control measures;
(iv) Mitigation of impact of user error or failure of any subsystem components (signal detection and analysis, data display, and storage) on output accuracy; and
(v) The sensitivity, specificity, positive predictive value, and negative predictive value in both percentage and number form for clinically meaningful pre-specified time windows consistent with the device output.
(2) Scientific justification for the validity of the hemodynamic indicator algorithm(s) must be provided. Verification of algorithm calculations and validation testing of the algorithm must use an independent data set.
(3) Usability assessment must be provided to demonstrate that risk of misinterpretation of the status indicator is appropriately mitigated.
(4) Clinical data must support the intended use and include the following:
(i) The assessment must include a summary of the clinical data used, including source, patient demographics, and any techniques used for annotating and separating the data;
(ii) Output measure(s) must be compared to an acceptable reference method to demonstrate that the output represents the measure(s) that the device provides in an accurate and reproducible manner;
(iii) The data set must be representative of the intended use population for the device. Any selection criteria or limitations of the samples must be fully described and justified;
(iv) Where continuous measurement variables are displayed, agreement of the output with the reference measure(s) must be assessed across the full measurement range; and
(v) Data must be provided within the clinical validation study or using equivalent datasets to demonstrate the consistency of the output and be representative of the range of data sources and data quality likely to be encountered in the intended use population and relevant use conditions in the intended use environment.
(5) Labeling must include the following:
(i) The type of sensor data used, including specification of compatible sensors for data acquisition, and a clear description of what the device measures and outputs to the user;
(ii) Warnings identifying factors that may impact output results;
(iii) Guidance for interpretation of the outputs, including warning(s) specifying adjunctive use of the measurements;
(iv) Key assumptions made in the calculation and determination of measurements; and
(v) A summary of the clinical validation data, including details of the patient population studied (
e.g., age, gender, race/ethnicity), clinical study protocols, and device performance with confidence intervals for all intended use populations.