The retrospective clinical dataset was used to conduct a multi-center clinical validation study to evaluate the safety and effectiveness of the device against a clinically accepted reference standard.
Retrospective clinical validation; Clinical performance; Routine clinical care; Multi-center study
Clinical Evidence
Study Design
Population
Comparator
Key Endpoints
Multi-center clinical validation study; Retrospective clinical validation study; Follow-up/Duration: Patient encounters between 2021 and 2024; Study Period: 2021-2024
Adult patients (age ≥22 years) presenting with symptoms suspicious for acute coronary syndrome; Sample Size: 1,805 patients (primary analysis population); Number of Sites: 4 US sites
Standard of care (manual ECG read by a health care provider)
Sensitivity and specificity for detection of type 1 myocardial infarction (STEMI/STEMI-equivalent)
AI Performance
Output
Algorithm
Acceptance
Observed
Dev DS
Dev Readers
Test DS
Test Readers
Acute Coronary Syndrome (ACS)
—
Sensitivity superior to standard of care; Specificity non-inferior to standard of care
Development dataset consists of training and tuning sets from US and OUS sources.
—
Multi-center clinical validation study: 1,805 patients from 4 US sites.
—
Indications for Use
Queen of Hearts™ (ACS AI ECG Model) is a software as a medical device (SaMD) intended for analyzing resting 12-lead ECGs in patients (age ≥22 years) presenting with symptoms suspicious for an acute coronary syndrome in whom referral to a cardiologist for early invasive diagnostic testing and clinically indicated coronary intervention may be considered. When using the device, healthcare professionals should take the following factors into account: - The device should be used in conjunction with the standard of care in ED and EMS settings for improved evaluation of patients with potential acute coronary syndromes. - Device use is targeted for potential acute coronary syndrome patients based on presenting clinical characteristics in whom the initial ECG is interpreted as low or indeterminate risk. - A negative device output does not exclude an acute coronary syndrome, STEMI, STEMI-equivalent, acute coronary occlusion, or other cardiac conditions, and the standard of care should not be modified in response to a negative device output. - No therapeutic decisions should be made based solely on the device output, and device findings should be reviewed by a qualified cardiology specialist.
Device Story
Standalone software module (SaMD) for healthcare professionals; integrates with EHR/HIS/medical platforms. Inputs: digital 12-lead resting ECG waveforms (≥200Hz, 2.5-15s duration). Processing: AI-based algorithm identifies STEMI or STEMI-equivalent characteristics. Output: positive/negative notification for suspected acute coronary occlusion. Usage: ED/EMS settings; supports triage/prioritization for cardiology consultation. Clinical impact: aids early identification of ACS patients; does not replace clinical evaluation; negative results do not rule out ACS; standard of care remains unchanged for negative outputs. Benefits: improved sensitivity over manual interpretation; facilitates timely invasive diagnostic testing/intervention.
Clinical Evidence
Multi-center clinical validation study (n=1,805) across 4 US sites (2021-2024). Reference standard: type 1 myocardial infarction (angiographic culprit lesion + adjudicated diagnosis). Primary endpoints: sensitivity and specificity compared to standard of care (manual ECG read). Results: Device sensitivity (58%) superior to standard of care (46.4%); device specificity (96.7%) non-inferior to standard of care (96.2%).
Technological Characteristics
Standalone software; AI-based algorithm. Inputs: digital 12-lead ECG waveforms (sampling 200-2000 Hz). Integration: EHR/HIS/medical device platforms. Cybersecurity: compliant with Section 524B FD&C Act. Software documentation: Enhanced level. Input validation: signal quality assessment (missing leads, lead reversal, artifact detection).
Indications for Use
Indicated for patients age ≥22 years presenting with symptoms suspicious for acute coronary syndrome (ACS) where initial ECG is low or indeterminate risk; intended for use in ED and EMS settings to assist in identifying patients for potential early invasive diagnostic testing and coronary intervention.
Submission Summary (Full Text)
{1}
# DE NOVO CLASSIFICATION REQUEST FOR
# Queen of Hearts™ (ACS AI ECG Model)
# REGULATORY INFORMATION
FDA identifies this generic type of device as:
Acute cardiovascular event detection and notification software. Acute cardiovascular event detection and notification software uses software algorithms to analyze inputs such as electrocardiograph signals, medical imaging data, or other medical data to detect acute cardiovascular events or conditions for further referral or diagnostic follow-up. The software provides an indicator that a patient is experiencing an acute cardiovascular event based on the algorithmic evaluation of one or more non-invasive physiological inputs as part of acute medical care. It is not intended to rule out acute cardiovascular events or conditions. It is not intended to identify or detect arrhythmias, stroke, or other neurological events.
NEW REGULATION NUMBER: 21 CFR 870.2230
CLASSIFICATION: Class II
PRODUCT CODE: SHS
# BACKGROUND
DEVICE NAME: Queen of Hearts™ (ACS AI ECG Model)
SUBMISSION NUMBER: DEN250044
DATE DE NOVO RECEIVED: September 22, 2025
# SPONSOR INFORMATION:
Powerful Medical, Inc.
7th Floor, 33 W. 17th St.
New York, NY 10011
# INDICATIONS FOR USE
The Queen of Hearts™ (ACS AI ECG Model) is indicated as follows:
Queen of Hearts™ (ACS AI ECG Model) is a software as a medical device (SaMD) intended for analyzing resting 12-lead ECGs in patients (age ≥22 years) presenting with symptoms suspicious for an acute coronary syndrome in whom referral to a cardiologist for early invasive diagnostic testing and clinically indicated coronary intervention may be
{2}
considered. When using the device, healthcare professionals should take the following factors into account:
- The device should be used in conjunction with the standard of care in ED and EMS settings for improved evaluation of patients with potential acute coronary syndromes.
- Device use is targeted for potential acute coronary syndrome patients based on presenting clinical characteristics in whom the initial ECG is interpreted as low or indeterminate risk.
- A negative device output does not exclude an acute coronary syndrome, STEMI, STEMI-equivalent, acute coronary occlusion, or other cardiac conditions, and the standard of care should not be modified in response to a negative device output.
- No therapeutic decisions should be made based solely on the device output, and device findings should be reviewed by a qualified cardiology specialist.
# LIMITATIONS
The sale, distribution, and use of the ACS AI ECG Model are restricted to prescription use in accordance with 21 CFR 801.109.
- A negative interpretation result does not rule-out follow-up testing.
- A negative interpretation result does not guarantee that the patient is not experiencing a cardiovascular event, abnormal rhythm, or other health conditions when no serious diagnoses are detected.
- The device output does not replace a full clinical evaluation of the patient and the output may not be sufficient as the sole basis for further therapeutic or diagnostic decisions.
- The device does not detect all types of acute myocardial infarctions, it is only designed to detect myocardial infarction due to an acutely occlusive or flow limiting culprit lesion. The device is not intended for the detection of chronic total occlusions.
- DO NOT use the device for analyzing ECG recordings of low quality or if it does not satisfy the device's input requirements. Low quality includes, but is not limited to, missing/disconnected leads, lead reversal, movement artifact, pulse tapping artifact, and any other feature lowering the quality of human ECG interpretation. As a rule of thumb, if any aspect of ECG quality prohibits the physician from being confident in their human interpretation, then the physician should not rely on the quality of the device
- DO NOT use the device for analyzing exercise stress test ECG recordings.
- DO NOT use the device for analyzing ambulatory/Holter ECG recordings.
- The device does not achieve 100% accuracy and may occasionally misidentify cardiovascular conditions. As with any medical device, experienced clinician judgment supersedes the Model in the event of disagreement.
PLEASE REFER TO THE LABELING FOR A COMPLETE LIST OF WARNINGS, PRECAUTIONS AND CONTRAINDICATIONS.
# DEVICE DESCRIPTION
De Novo Summary (DEN250044)
Page 2 of 12
{3}
The Queen of Hearts™ (ACS AI ECG Model) software analyzes 12-lead resting ECG data from adult patients presenting with a clinical suspicion of acute coronary syndrome using an artificial intelligence-based algorithm. The device is a decision support tool intended for healthcare professionals to help with the early identification of suspected ACS patients who may benefit from early cardiology consultation and subsequent clinical evaluation. The device is a stand-alone software module that integrates with existing medical systems (e.g., electronic health record systems, hospital information systems, or other medical device platforms). This device only analyzes ECG recordings from compatible ECG acquisition systems transferred in digital waveform format, does not analyze ECGs acquired from images of ECG tracings. The device is not intended to be the sole means of diagnosis, and no therapeutic decisions are made based solely on the device output. A negative interpretation result does not rule-out follow-up testing. Note: The device has not been tested for pediatric patients or patients under the age of 22.
## SUMMARY OF NONCLINICAL/BENCH STUDIES
### SOFTWARE
The software was reviewed according to the “Content of Premarket Submissions for Device Software Functions” dated June 24, 2023. Appropriate software documentation consistent with a “Enhanced” documentation level was provided. The software documentation included a detailed description of the machine learning model, the model inputs and outputs, and the supported patient population. Integration testing was conducted in the intended software system. Functional test cases, including end-to-end tests, unit-level tests, and verification by inspection, were designed to demonstrate that the software properly implements the requirements. This included signal quality assessments for data inputs. Data input requirements are specified by 12-lead ECG parameters. Input validation was assessed for missing fields, format, sampling frequency, appropriate ECG leads, etc. The algorithm will not compute an output in the event of input failures.
The algorithm was developed on 12-lead ECGs collected from US and OUS sources. The development dataset consists of both training and tuning sets. The reference standard of the development data was established using angiographic and laboratory outcome data. Across all development data the reference standard was an algorithmic decision based on angiographic outcome.
### CYBERSECURITY
Cybersecurity was reviewed in accordance with Section 524B of the Food, Drug, & Cosmetic Act, effective March 29, 2023 and FDA guidance document “Cybersecurity in Medical Devices: Quality Management System Considerations and Content of Premarket Submissions” dated February 3, 2026.
De Novo Summary (DEN250044)
Page 3 of 12
{4}
## **USABILITY**
The usability of the ACS AI ECG Model was assessed per EN 62366-1 “Medical Devices Part 1: Application of usability engineering to medical devices” dated 2015 + A1:2020 and the FDA guidance document “Applying Human Factors and Usability Engineering to Medical Devices” dated February 3, 2016. The study was designed and executed based on a Use-Related Risk Analysis (URRA) that identified critical tasks – those that, if performed incorrectly, could lead to harm of moderate severity or greater. Identified critical tasks centered on locating the device output in the user interface and correctly interpreting the output in clinical context. All participants successfully completed all critical tasks, with a 100% pass rate across every critical task. The results of the human factors validation study support that the device is safe and effective for the intended users, uses, and use environments.
## **SUMMARY OF CLINICAL INFORMATION**
A multi-center clinical validation study was conducted to evaluate the safety and effectiveness of the subject device for its intended use. The study was designed to assess device performance and standard of care performance against a clinically accepted reference standard in the intended use population. Device output and standard of care determination were compared against a reference standard of type 1 myocardial infarction including pre-specified coronary angiographic criteria, which are commonly used in clinical practice. The reference standard was established independently of the subject device output to preserve the integrity of the performance assessment. In the validation study, patients were classified as reference-standard positive (STEMI/STEMI-equivalent) when the patient had an angiographic culprit lesion together with an adjudicated diagnosis of type 1 MI. Patients not meeting both criteria were classified as reference-standard negative. All reference standard positive patients underwent diagnostic cardiac catheterization. For all reference standard negative patients, the reference standard was established by troponin testing, the guideline-defined standard for establishing whether an acute MI occurred.
The overall validation dataset included 2,001 patients from 4 US sites. Patient care occurred between 2021 and 2024. Mean age was 54 years, 44% were female, and 47% were non-Caucasian, reflecting substantial representation of female patients and of racial and ethnic minority patients.
The study datasets used for algorithm validation were not used for training or tuning purposes. The validation dataset was firewalled during the algorithm development and separated from the training set and came from different clinical sites.
The primary analysis population included 1,805 patients (mean age 54 years; 44% female; 47% non-White among patients with recorded race; 11% Hispanic or Latino).
The clinical validation employed a pre-specified statistical analysis plan. The sample size was determined to provide adequate statistical power to evaluate the endpoints of sensitivity and specificity, with appropriate margins defined a priori. Enrollment spanned multiple clinical sites
*De Novo Summary (DEN250044)*
Page 4 of 12
{5}
in the US representative of the intended use environment, and the study population reflected the demographic and clinical characteristics of patients for whom the device is indicated. Statistical comparisons were pre-specified to evaluate both superiority and non-inferiority endpoints as appropriate. The subject device met all pre-specified primary endpoints. Specifically, the device demonstrated:
- Statistically significant superior sensitivity compared to the standard of care, with the lower bound of the confidence interval exceeding the pre-specified superiority margin.
- Non-inferior specificity compared to the standard of care, with the lower bound of the confidence interval falling within the pre-specified non-inferiority margin.
The device outputs a positive finding when STEMI or STEMI-equivalent characteristics are identified in the ECG. The clinical utility of this device output includes use during the initial evaluation by health care providers (non-cardiologist, often non-physician) to support referral of patients for a full evaluation by a cardiologist and for early invasive diagnostic testing and coronary intervention, where clinically indicated. The current standard of care for a patient presenting with chest pain in the emergency department setting is a 12-lead ECG with a manual ECG read by a health care provider, which has limited inter-rater reliability for detecting ACS related signals. This software helps expedite follow up care for suspected acute coronary events, which may include additional clinical evaluation and/or testing, such as imaging (e.g., echocardiography), cardiac biomarkers (troponins), serial ECGs, etc. Performance was consistent across pre-specified subgroups, and no clinically meaningful differences were observed across relevant patient subpopulations. The study was not specifically powered for sex-, age-, race-, ethnicity-, or any other relevant characteristic- specific subgroups. The table below outlines the baseline demographics and clinical parameters.
Table 1: Baseline Characteristics
| | | Overall | Reference Standard STEMI/STEMI equivalent (+) | Reference Standard STEMI/STEMI equivalent (-) | P-Value (Test name) |
| --- | --- | --- | --- | --- | --- |
| Unique Patients: n | | 1805 | 293 | 1512 | - |
| Unique hospitalizations: n | | 1805 | 293 | 1512 | - |
| Unique ECGs: n | | 1805 | 293 | 1512 | - |
| Age: Mean ± SD | | 54 ± 17 | 63 ± 13 | 52 ± 17 | <0.001 (One-way ANOVA) |
| Sex: n (%) | | 1805 | 293 | 1512 | 0.01 (Chi-squared) |
| | Female | 791 (43.82%) | 108 (36.86%) | 683 (45.17%) | - |
| | Male | 1,014 (56.18%) | 185 (63.14%) | 829 (54.83%) | - |
| Race: n (%) | | 1628 | 287 | 1341 | 0.157 |
De Novo Summary (DEN250044)
Page 5 of 12
{6}
| | | | | | (Chi-squared) |
| --- | --- | --- | --- | --- | --- |
| | White | 863(53.01%) | 170 (59.23%) | 693 (51.68%) | - |
| | Black or African American | 666(40.91%) | 101 (35.19%) | 565 (42.13%) | - |
| | Native Hawaiian or other Pacific Islander | 8 (0.49%) | 2 (0.70%) | 6 (0.45%) | - |
| | American Indian | 3 (0.18%) | 1 (0.35%) | 2 (0.15%) | - |
| Ethnicity: n (%) | | 1791 | 293 | 1498 | 0.179(Chi-squared) |
| | Hispanic or Latino | 196(10.94%) | 25 (8.53%) | 171 (11.42%) | - |
| | Not Hispanic or Latino | 1,595(89.06%) | 268 (91.47%) | 1,327 (88.58%) | - |
| Presentation type: n (%) | | 1805 | 293 | 1512 | <0.001(Chi-squared) |
| | EMS | 165(9.14%) | 94 (32.08%) | 71 (4.70%) | - |
| | Referral (e.g. from GP, Spoke hospital) | 250(13.85%) | 16 (5.46%) | 234 (15.48%) | - |
| | Walk-in | 1,390(77.01%) | 183 (62.46%) | 1,207 (79.83%) | - |
| Discharge diagnosis: n (%) | | 1805 | 293 | 1512 | <0.001(Chi-squared) |
| | MI ruled-out | 1,326(73.46%) | 1 (0.34%) | 1,325 (87.63%) | - |
| | NSTEMI | 233(12.91%) | 134 (45.73%) | 99 (6.55%) | - |
| | Other | 60(3.32%) | 5 (1.7%) | 55 (3.64%) | - |
| | STEMI | 186(10.30%) | 153 (52.22%) | 33 (2.18%) | - |
| Year of patient encounter: n (%) | | 1805 | 293 | 1512 | <0.001(Chi-squared) |
| | 2021 | 990(54.85%) | 45 (15.36%) | 945 (62.50%) | - |
| | 2022 | 246(13.63%) | 171 (58.36%) | 75 (4.96%) | - |
De Novo Summary (DEN250044)
Page 6 of 12
{7}
| | 2023 | 537 (29.75%) | 64 (21.84%) | 473 (31.28%) | - |
| --- | --- | --- | --- | --- | --- |
| | 2024 | 32 (1.77%) | 13 (4.44%) | 19 (1.26%) | - |
| LBBB: n (%) | | 68 (3.77%) | 25 (8.53%) | 43 (2.84%) | <0.001 (Chi-squared) |
| RBBB: n (%) | | 145 (8.03%) | 15 (5.12%) | 130 (8.60%) | 0.059 (Chi-squared) |
| Rhythm: n (%) | | 1805 | 293 | 1512 | 0.096 (Chi-squared) |
| | Sinus | 1,693 (93.80%) | 267 (91.13%) | 1,426 (94.31%) | - |
| | AF | 57 (3.16%) | 12 (4.10%) | 45 (2.98%) | - |
| | Other | 55 (3.05%) | 14 (4.78%) | 41 (2.71%) | - |
Both pre-specified primary endpoints were met in the primary analysis population. Device sensitivity exceeded standard-of-care sensitivity by 11.6 percentage points, meeting the pre-specified superiority criterion, and device specificity was non-inferior to standard of care (difference +0.5 percentage points). Performance is summarized in the table below.
Table 2: Overall Algorithm Performance
| Comparator | | Ref + | Ref - | Accuracy | Sensitivity | Specificity | PPV | NPV | F1 |
| --- | --- | --- | --- | --- | --- | --- | --- | --- | --- |
| STEMI AI ECG Model | + | 170 | 50 | 90.4% | 58% | 96.7% | 77.3% | 92.2% | 66.3% |
| | - | 123 | 1462 | (89.1-91.7) | (52.5-63.6) | (95.8-97.6) | (71.7-82.6) | (90.9-93.5) | (61.5-70.9) |
| Standard of care | + | 136 | 57 | 88.1% | 46.4% | 96.2% | 70.5% | 90.3% | 56% |
| | - | 157 | 1455 | (86.6-89.6) | (40.8-52) | (95.2-97.2) | (63.8-76.7) | (88.8-91.7) | (50.5-61) |
Subgroup analysis by ECG manufacturer demonstrated acceptable and consistent device performance on ECGs acquired using different acquisition systems.
The performance of the algorithm will continue to be monitored postmarket to ensure the generalizability and device performance in the intended patient population in real-world use. This will include performance metrics as well as clinical outcomes and aims to detect changes in performance across diverse sites, including community-based sites and tertiary care centers. Additionally, results will be communicated via updated labeling.
De Novo Summary (DEN250044)
Page 7 of 12
{8}
## Pediatric Extrapolation
In this De Novo request, existing clinical data were not leveraged to support the use of the device in a pediatric patient population.
## LABELING
The labeling includes the following considerations, as expected for this class of devices:
a. A summary of the performance testing methods, tested hardware, tested/supported patient population, results of the performance testing for tested performance measures/metrics, summary-level descriptions of patient demographics and associated subgroup analyses for training and test datasets;
b. Device limitations or subpopulations for which the device may not perform as expected;
c. A statement that the device output should not replace a full clinical evaluation of the patient and that the output may not be sufficient as the sole basis for further testing or treatment;
d. Warnings identifying sensor acquisition factors that may impact results;
e. An integration guide with examples and recommendations regarding UI implementation of warnings, cautions and device limitations;
f. The type(s) of hardware sensor data used, including specification of compatible sensors for data acquisition.
## PREDETERMINED CHANGE CONTROL PLAN
The submission contains a Predetermined Change Control Plan (PCCP), which complies with the Agency's requirements and Section 3308 of the Food and Drug Omnibus Reform Act (FDORA) of 2022, enacted on December 29, 2022.
The PCCP includes a modification to establish compatibility of the ACS AI ECG Model with additional ECG sources. An ECG source is defined as an acquisition and conversion pathway, identified by the ECG acquisition device manufacturer and the acquisition setting used to record the ECG, that delivers a standard 12-lead ECG signal conforming to the device input specification below:
| Parameter | Specification |
| --- | --- |
| Leads | Simultaneous standard 12-lead ECG (I, II, III, aVR, aVL, aVF, V1–V6) |
| Sampling frequency | \( \geq 200 \text{ Hz}; \leq 2000 \text{ Hz} \) |
| Duration | \( \geq 2.5 \text{ seconds}; \leq 15 \text{ seconds} \) |
| Per-lead completeness | All 12 leads present with valid signal for the required duration |
De Novo Summary (DEN250044)
Page 8 of 12
{9}
The PCCP includes specific test protocols, including methods of implementation and acceptance criteria, providing sufficient means to implement the outlined modifications without requiring an additional marketing submission prior to implementation. Algorithm performance on each additional ECG source will be evaluated against pre-specified acceptance criteria referenced to the performance demonstrated in the clinical validation study.
Released modifications will be communicated to users through software update notes and labeling.
## RISKS TO HEALTH
The table below identifies the risks to health that may be associated with use of acute cardiovascular event detection and notification software.
| Risks to Health | Mitigation Measures |
| --- | --- |
| False positive or false negative leading to incorrect treatment or diagnosis | Clinical performance testing Labeling |
| Incorrect treatment or diagnosis due to model bias or failure to adequately generalize to the intended use population | Clinical performance testing Labeling |
| Device used in unsupported patient population or with unsupported input/hardware | Labeling Human factors assessment Software verification, validation, and hazard analysis |
| Overreliance on device output for follow-up | Labeling Human factors assessment |
## SPECIAL CONTROLS
In combination with the general controls of the FD&C Act, the acute cardiovascular event detection and notification software is subject to the following special controls:
(1) Clinical performance testing must demonstrate that the device performs as intended under anticipated conditions of use. The following must be met:
(i) Clinical validation must use a test dataset acquired from a representative patient population. Data must 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. The test dataset must be independent from data that was used in training/development and contain sufficient numbers of cases from important cohorts (e.g., demographic populations, subsets defined by clinically relevant confounders, comorbidities, and subsets defined by hardware and acquisition characteristics) such that the performance estimates and
De Novo Summary (DEN250044)
Page 9 of 12
{10}
confidence intervals of the device for these individual subsets can be characterized for the intended use population and acquisition systems (e.g., acquisition hardware or preprocessing software). Study protocols must include a description of the adjudication process(es) for determining ground truth of training and test datasets;
(ii) Data must demonstrate consistency of the output over the full range of inputs;
(iii) A justification for the performance goals must be provided;
(iv) The test dataset must include a minimum of 3 geographically diverse sites, separate from sites used in training of the model; and
(v) Justification for the clinical utility of the device output must include a discussion of how device output would be used in patient management to improve care in the acute setting (e.g., reference to clinical guidelines or outcomes study).
(2) The device manufacturer must develop and implement a postmarket performance management plan that ensures regular assessment of the generalizability and device performance in the intended patient population in real-world use. The plan must include:
(i) Data collection, analysis methods, and procedures for:
(A) Monitoring relevant performance characteristics and detecting changes in performance;
(B) Identifying sources of performance changes between validation and real-world environment over time; and
(C) Assessing the results from the performance testing on safety and effectiveness;
(ii) Procedures for communicating the device's current performance to the users.
(3) Software verification, validation, and hazard analysis must be performed. Software documentation must include:
(i) A full characterization of technical parameters of the software, including any algorithms or models used, all inputs and outputs for the software, and the supported patient population;
(ii) Description of the expected impact of all applicable signal acquisition hardware characteristics on performance and any associated hardware specifications;
(iii) Specification of acceptable incoming signal data quality control measures; and
(iv) Data documentation describing any training, tuning, or validation datasets used in algorithm development.
(4) Human factors/usability assessment must be provided to mitigate the risk of misinterpretation of the device output.
(5) Labeling must include the following:
(i) A summary of the performance testing methods, tested hardware, tested/supported patient population, results of the performance testing for tested performance measures/metrics, summary-level descriptions of patient demographics and associated subgroup analyses for training and test datasets;
(ii) Device limitations or subpopulations for which the device may not perform as expected;
(iii) A statement that the device output should not replace a full clinical evaluation of the patient and that the output may not be sufficient as the sole basis for further testing or treatment;
De Novo Summary (DEN250044)
Page 10 of 12
{11}
(iv) Warnings identifying sensor acquisition factors that may impact results;
(v) An integration guide with examples and recommendations regarding UI implementation of warnings, cautions and device limitations; and
(vi) The type(s) of hardware sensor data used, including specification of compatible sensors for data acquisition.
## BENEFIT-RISK DETERMINATION
Real-world performance will continue to be characterized after marketing. The indications for use specify the population in which premarket data support that the probable benefits outweigh the probable risks, and the postmarket performance management plan addresses any residual uncertainty regarding the generalizability of device performance in the intended use population.
The risks of the device are based on data collected in a clinical study described above.
The primary risk of using the device is false negatives where a subject with acute coronary stenosis or occlusion or a STEMI/STEMI equivalent would be said to not likely have the disease. Misclassification could lead to inappropriate treatment decisions, resulting in under- treatment of high-risk patients or over-treatment of low-risk patients. This could result in preventable cardiac events, including myocardial infarctions or cardiac deaths. Inappropriate medication prescribing based on inaccurate risk stratification could expose patients to unnecessary side effects from medications they don't need, or conversely, deny them beneficial treatments that could prevent cardiovascular events. The risks associated with false positives results in overestimation, which could lead to unnecessary anxiety in patients and additional catheterization procedures.
The risk of false negatives is mitigated in two ways: the device demonstrated higher sensitivity than the standard of care in the pivotal study, and its intended use specifies no change to patient management in response to a negative result – a negative device output does not modify the standard of care. The risk associated with false positives is low, as a false positive leads only to earlier full patient evaluation by a cardiologist.
The probable benefits of the device are also based on data collected in a clinical study as described above.
The primary benefit of using the device is the aid in prioritization of patients who may benefit from early cardiology consultation. This device can help physicians make more timely and informed decisions about treatment options as an additional input to their decision-making process, along with data from other standard of care options available to them. The indications for use specify the population where premarket data supports that probable benefits outweigh the risks.
## Patient Perspectives
This submission did not include specific information on patient perspectives for this device.
De Novo Summary (DEN250044)
Page 11 of 12
{12}
# Benefit/Risk Conclusion
In conclusion, given the available information above, for the following indication statement:
Queen of Hearts™ (ACS AI ECG Model) is a software as a medical device (SaMD) intended for analyzing resting 12-lead ECGs in patients (age ≥22 years) presenting with symptoms suspicious for an acute coronary syndrome in whom referral to a cardiologist for early invasive diagnostic testing and clinically indicated coronary intervention may be considered. When using the device, healthcare professionals should take the following factors into account:
- The device should be used in conjunction with the standard of care in ED and EMS settings for improved evaluation of patients with potential acute coronary syndromes.
- Device use is targeted for potential acute coronary syndrome patients based on presenting clinical characteristics in whom the initial ECG is interpreted as low or indeterminate risk.
- A negative device output does not exclude an acute coronary syndrome, STEMI, STEMI-equivalent, acute coronary occlusion, or other cardiac conditions, and the standard of care should not be modified in response to a negative device output.
- No therapeutic decisions should be made based solely on the device output, and device findings should be reviewed by a qualified cardiology specialist.
The probable benefits outweigh the probable risks for the Queen of Hearts™ ACS AI ECG Model. The device provides benefits and the risks can be mitigated by the use of general controls and the identified special controls.
# CONCLUSION
The De Novo request for the Queen of Hearts™ ACS AI ECG Model is granted and the device is classified as follows:
Product Code: SHS
Device Type: Acute cardiovascular event detection and notification software
Regulation Number: 21 CFR 870.2230
Class: II
De Novo Summary (DEN250044)
Page 12 of 12
Predicate graph will load when search results are available.
Embedding visualization will load when search results are available.
PDF viewer will load when search results are available.
Loading panels...
Select an item from Submissions
Click any panel, subpart, regulation, product code, or device to see details here.
Section Matches
Results will appear here.
Product Code Matches
Results will appear here.
Special Control Matches
Results will appear here.
Loading collections...
Loading
My Alerts
You will receive email notifications based on the filters and frequency you set for each alert.
Sort by:
Guidance alerts
New and changed FDA guidance documents, emailed as a daily digest.
Create Alert
Search Filters
Agent Token
Create a read-only bearer token for Claude, ChatGPT, or other agents that can call HTTP APIs.
Copy this now. It will not be shown again.
Connected apps
Apps you authorized through browser sign-in. Disconnecting revokes their access immediately.
Learn the FDA Browser
Two short videos show you everything — or skip straight to the written tutorial if you'd rather read. You can reopen this any time from the ? button in the top bar.
Part 1 — Search, results, and everyday workflows 16 min
Part 2 — Embeddings: the galaxy map 3 min
1. Search: exact and fuzzy
Type a phrase like "coronary artery calcification" into the search box. You get two kinds of results. Exact results match the literal phrase — prefix searches work ("coronary artery calcificati") but suffix searches do not. Fuzzy results match on the meaning and intent of your phrase rather than the exact words, and are sorted by relevance score. Hover over the Exact or Fuzzy badge on any row to see exactly why it matched.
Use the checkboxes above the results to narrow: SaMD keeps only Software as a Medical Device, AI / ML keeps only devices with AI.
Exact vs. fuzzy search: what's the difference?
Exact matches on the literal phrase (prefix search works, suffix does not). Fuzzy matches on the meaning and intent of the phrase rather than the exact words. Hover over the badge on any row to see why it matched.
You search "coronary artery calcification" and want only software devices with AI. What two filters do you apply?
Narrow by SaMD (Software as a Medical Device), then narrow by AI/ML (devices with AI).
2. The results table
Scroll right in the results table. The intended use is extracted for you — no need to open the PDF. The device story gives a high-level snapshot of what the device does and how it's used. The AI Performance sub-table shows each output name, acceptance criteria, observed values, and development/test dataset descriptions — the same format Innolitics uses for regulatory strategy outputs, and the fastest high-level fingerprint of an AI device. It is AI-generated but has been very reliable in practice.
Where do you find a device's intended use without opening the PDF?
Scroll right in the search results table. The intended use column is extracted for you; no need to dig into the 510(k) summary PDF.
What does the AI Performance sub-table show, and why is it useful?
Output name, acceptance criteria, observed values, development dataset description, and test dataset description. It's the same format we use for regulatory strategy output and Fast 510(k) input, and the fastest high-level fingerprint of an AI device. AI-generated but reliable in practice.
3. Judging fuzzy relevance
Fuzzy results trail off in relevance as you scroll. Use three signals to decide how far down to go: the fuzzy badge explanations, the intended use column, and whether your target output (e.g., Cobb angle) still appears in the AI Performance sub-table. Once it stops appearing, you're past the relevant zone. A top hit with a low score (~0.4) and a stretched explanation is a hint the closest predicates are far away — the project may be headed for De Novo. Note the fuzzy search is a pattern match: it doesn't handle negation ("not") well, and hardware devices can appear — filter by SaMD/AI ML to cut them.
How do you judge how far down fuzzy search results to go?
Use the relevancy signals: the fuzzy badge explanations, the intended use column, and whether the target output (e.g., Cobb angle) still appears in the AI Performance sub-table. Once it stops appearing, results are trailing off in relevancy.
4. Device detail page: chat and citations
Click a device name to open its detail page: device facts on the left, a chat window on the right. Ask something like "Describe the training data". The answer carries little citation bubbles — click one to jump to the highlighted passage in the source PDF, so you can verify every AI answer against the document. There's also a Download PDF button for sharing.
How do you verify an AI chat answer on the device detail page?
Click the citation bubbles to jump to the relevant highlight in the source document.
Reading rule for every project: how many summaries do you read in full?
At least the three most relevant 510(k) or De Novo summaries, in full. After that, use targeted chat questions to confirm your memory quickly. The tool supports this professional habit — it doesn't replace it.
5. Side-by-side comparison
Select multiple rows in the results table (aim for under ~10), then open the PDF Viewer tab. Ask one question — it goes to all selected devices in parallel, each with citations. This is the fastest way to compare and contrast devices: training data, PCCP scope, how they handled adding new scanners, and so on.
What does the side-by-side PDF viewer mode do?
Select multiple devices, open the PDF viewer tab, and ask one question (e.g., "Describe the training data"). It queries all selected devices simultaneously with citations, so you can compare and contrast quickly.
6. Collections
With rows selected, go to the Collections tab and create a labeled collection (e.g., "Cobb Angle Project"). Reload that selection any time — before a client call, pull up the collection and ask questions across all of its devices at once.
How do you save a set of selected devices for later use?
Select the rows, go to the Collections tab, and create a labeled collection (e.g., "Cobb Angle Project"). You can reload the selection anytime and carry it into the PDF viewer and other tabs that support selections.
7. Product codes and the regulations tree
Click a product code in the results to jump to it in the regulations tree — identification text, sibling product codes, and devices you can open in a PDF viewer on the right. Click a regulation number to see its identification, special controls, and related product codes. You can also search by product code or regulation number at the top of the tree. Always read the special controls if any exist for your device — it broadens your search and sharpens pre-kickoff research.
What can you do from the regulations tree view?
Browse product codes and regulation numbers, read the identification text and special controls, browse sibling product codes, open device PDFs on the right, and search by product code or regulation number at the top of the tree.
8. Chart view
Click Show Chart and segment by regulation number (or product code) to see which regulations dominate your result set. Clicking a regulation takes you into the regulations tree. Great for spotting that most matches are, say, hardware laparoscopic devices — a cue to go back and filter.
How do you see which regulations dominate a search result set?
Click "Show Chart" and segment by Regulation Number. Clicking a regulation takes you to the regulations tree.
9. The predicate graph
Open the Predicates tab for a family-tree view of predicate relationships. Click a node to trace its parents and children; selections from search carry over pre-selected. Commonly predicated devices are worth reading — a lot of people predicated them for a reason. The visual lineage is also handy on client calls, e.g. to show how a predicate family evolved and justify why your predicate still holds.
In the predicate graph, why are commonly predicated devices worth reading?
A lot of people predicated them for a reason. Clicking a node traces parents and children, and selections from search carry over pre-selected.
10. Embeddings: the galaxy map
The Embeddings tab plots every matching document in a 2-D "galaxy map" where semantically similar devices cluster together. Hover or click clusters to explore, and let AI label the clusters for you. Embeddings beat product codes for grouping: two devices can carry different product codes (LLZ vs. QIH) yet do the same thing — the embedding captures the meaning of the intended use and device story. This is also exactly how retrieval-augmented generation (RAG) works under the hood, and it makes a great visual on client calls.
Try it yourself
Head to the search page and work through a few of these AI/ML fuzzy searches to build intuition: perivascular fat on CT · aortic valve calcification opportunistic screening on noncontrast CT · breast cancer prediction on digital pathology slides · autism detection · gestational age prediction · a hearing aid that can also detect a pulse · foundation model based analysis of ECG · large language models · penetration test. Watch how the relevance scores, intended use, and AI Performance tables tell you when results stop being meaningful.