Cardiomyopathy, Pregnancy Related
Conditions
Keywords
Cardiomyopathy, Artificial Intelligence, Women's Health, Digital Health
Brief summary
This study will evaluate the effectiveness of an artificial intelligence-enabled ECG (AI-ECG) for cardiomyopathy detection in an obstetric population in Nigeria.
Interventions
OTHERDigital stethoscope electrocardiogram
Digital stethoscope artificial intelligence enabled electrocardiogram (AI-ECG). An artificial intelligence algorithm which analyses ECG data and generates prediction probabilities for a diagnosis of cardiomyopathy.
Sponsors
Eunice Kennedy Shriver National Institute of Child Health and Human Development (NICHD)
National Center for Advancing Translational Sciences (NCATS)
National Institute of Arthritis and Musculoskeletal and Skin Diseases (NIAMS)
Mayo Clinic
Study design
Allocation
RANDOMIZED
Intervention model
PARALLEL
Primary purpose
SCREENING
Masking
NONE
Eligibility
Sex/Gender
FEMALE
Age
18 Years to 49 Years
Healthy volunteers
Yes
Inclusion criteria
* Currently pregnant or within 12 months postpartum * Willing and able to provide informed consent
Exclusion criteria
* Complex congenital heart disease (single ventricle physiology or significant shunts with cardiac structural changes) * Significant conduction abnormalities (ventricular pacing on recorded ECG, pacemaker dependence, or severely abnormal/bizarre QRS morphology on ECG tracings) * Unable or unwilling to provide consent
Design outcomes
Primary
| Measure | Time frame | Description |
|---|---|---|
| Left Ventricular Ejection Fraction (LVEF) <50% | 18 months | Number of participants diagnosed with left ventricular ejection fraction (LVEF) \<50% by echocardiography during pregnancy or within 12 months postpartum. |
Secondary
| Measure | Time frame | Description |
|---|---|---|
| Effectiveness AI-ECG for Cardiomyopathy Detection in the Intervention Arm in LVEF < 40% | 18 months | This is defined as a positive point-of-care AI prediction for LVEF \< 40% (maximum prediction across all stethoscope recording locations) confirmed with echocardiography |
| Effectiveness AI-ECG for Cardiomyopathy Detection in the Intervention Arm in LVEF < 45% | 18 months | This is defined as a positive point-of-care AI prediction for LVEF \<45% (maximum prediction across all stethoscope recording locations) confirmed with echocardiography |
| Effectiveness AI-ECG for Cardiomyopathy Detection in the Intervention Arm in LVEF < 50% | 18 months | This is defined as a positive point-of-care AI prediction for LVEF \<50% (maximum prediction across all stethoscope recording locations) confirmed with echocardiography |
| Effectiveness of AI-ECG for Cardiomyopathy Detection in the Intervention Arm for Left Ventricular Ejection Fraction (LVEF) ≤ 35% | 18 months | This is defined as a positive point-of-care AI prediction for LVEF ≤ 35% (maximum prediction across all stethoscope recording locations) confirmed with echocardiography |
Other
| Measure | Time frame | Description |
|---|---|---|
| Echocardiography Utilization | 18 months | Determine the impact of an AI-ECG on echocardiography utilization |
| Effectiveness of AI Point of Care Tools for Cardiomyopathy Detection in the Intervention Arm | 18 months | Develop and evaluate the diagnostic performance of an AI-enhanced point of care screening tool |
| Composite Adverse Cardiovascular Events | 18 months | The number of subjects to experience composite cardiovascular events with include any of the following: diastolic heart failure, gestational hypertension, pre-eclampsia, eclampsia, valvular heart disease, atrial arrhythmias and sustained ventricular arrhythmias. |
Countries
Nigeria
Participant flow
Participants by arm
| Arm | Count |
|---|---|
| Intervention Participants had ECGs analyzed with artificial intelligence for cardiomyopathy detection.
Digital stethoscope electrocardiogram: Digital stethoscope artificial intelligence enabled electrocardiogram (AI-ECG). An artificial intelligence algorithm which analyses ECG data and generates prediction probabilities for a diagnosis of cardiomyopathy. | 587 |
| Control Participants had standard clinical ECGs acquired. | 608 |
| Total | 1,195 |
Withdrawals & dropouts
| Period | Reason | FG000 | FG001 |
|---|---|---|---|
| Overall Study | Death | 1 | 1 |
| Overall Study | Did not complete baseline testing | 22 | 4 |
| Overall Study | Not Eligible | 2 | 1 |
| Overall Study | Withdrawal by Subject | 4 | 2 |
Baseline characteristics
| Characteristic | Control | Total | Intervention |
|---|---|---|---|
| Age, Continuous | 31 years | 31 years | 31 years |
| Race/Ethnicity, Customized Ethnicity - Hausa | 174 Participants | 337 Participants | 163 Participants |
| Race/Ethnicity, Customized Ethnicity - Igbo | 68 Participants | 129 Participants | 61 Participants |
| Race/Ethnicity, Customized Ethnicity - other | 41 Participants | 76 Participants | 35 Participants |
| Race/Ethnicity, Customized Ethnicity - Yoruba | 325 Participants | 653 Participants | 328 Participants |
| Race (NIH/OMB) American Indian or Alaska Native | 0 Participants | 0 Participants | 0 Participants |
| Race (NIH/OMB) Asian | 0 Participants | 0 Participants | 0 Participants |
| Race (NIH/OMB) Black or African American | 608 Participants | 1195 Participants | 587 Participants |
| Race (NIH/OMB) More than one race | 0 Participants | 0 Participants | 0 Participants |
| Race (NIH/OMB) Native Hawaiian or Other Pacific Islander | 0 Participants | 0 Participants | 0 Participants |
| Race (NIH/OMB) Unknown or Not Reported | 0 Participants | 0 Participants | 0 Participants |
| Race (NIH/OMB) White | 0 Participants | 0 Participants | 0 Participants |
| Region of Enrollment Nigeria | 608 participants | 1195 participants | 587 participants |
| Sex: Female, Male Female | 608 Participants | 1195 Participants | 587 Participants |
| Sex: Female, Male Male | 0 Participants | 0 Participants | 0 Participants |
Adverse events
| Event type | EG000 affected / at risk | EG001 affected / at risk |
|---|---|---|
| deaths Total, all-cause mortality | 12 / 587 | 3 / 608 |
| other Total, other adverse events | 1 / 587 | 4 / 608 |
| serious Total, serious adverse events | 56 / 587 | 53 / 608 |
Outcome results
Primary
Left Ventricular Ejection Fraction (LVEF) <50%
Number of participants diagnosed with left ventricular ejection fraction (LVEF) \<50% by echocardiography during pregnancy or within 12 months postpartum.
Time frame: 18 months
Population: Analysis was categorized by pregnant and post-partum participants. Total of each group equals the overall number of participants analyzed per arm.
| Arm | Measure | Group | Value (COUNT_OF_PARTICIPANTS) |
|---|---|---|---|
| Intervention | Left Ventricular Ejection Fraction (LVEF) <50% | Pregnant | 3 Participants |
| Intervention | Left Ventricular Ejection Fraction (LVEF) <50% | Post partum | 21 Participants |
| Control | Left Ventricular Ejection Fraction (LVEF) <50% | Pregnant | 1 Participants |
| Control | Left Ventricular Ejection Fraction (LVEF) <50% | Post partum | 11 Participants |
p-value: 0.03295% CI: [1.05, 4.27]Regression, Logistic
Secondary
Effectiveness AI-ECG for Cardiomyopathy Detection in the Intervention Arm in LVEF < 40%
This is defined as a positive point-of-care AI prediction for LVEF \< 40% (maximum prediction across all stethoscope recording locations) confirmed with echocardiography
Time frame: 18 months
Population: To explain the differences in number analyzed for each section below, details are provided: Sensitivity, is calculated as True Positive (TP) / (TP + FN), as such the denominator in this case is 20. Specificity is calculated as True Negatives (TN) divided by True Negatives and False Positives (TN + FP), and the denominator is 560. Positive Predictive Value (PPV) = TP / (TP + FP), with a denominator of 122 and Negative Predictive Value (PPV) = TN / (TN + FN) with a denominator of 458.
| Arm | Measure | Group | Value (COUNT_OF_PARTICIPANTS) |
|---|---|---|---|
| Intervention | Effectiveness AI-ECG for Cardiomyopathy Detection in the Intervention Arm in LVEF < 40% | LVEF < 40% (Sensitivity) | 20 Participants |
| Intervention | Effectiveness AI-ECG for Cardiomyopathy Detection in the Intervention Arm in LVEF < 40% | LVEF < 40% (Specificity) | 458 Participants |
| Intervention | Effectiveness AI-ECG for Cardiomyopathy Detection in the Intervention Arm in LVEF < 40% | LVEF < 40% (PPV) | 20 Participants |
| Intervention | Effectiveness AI-ECG for Cardiomyopathy Detection in the Intervention Arm in LVEF < 40% | LVEF < 40% (NPV) | 458 Participants |
Secondary
Effectiveness AI-ECG for Cardiomyopathy Detection in the Intervention Arm in LVEF < 45%
This is defined as a positive point-of-care AI prediction for LVEF \<45% (maximum prediction across all stethoscope recording locations) confirmed with echocardiography
Time frame: 18 months
Population: To explain the differences in number analyzed for each section below, details are provided: Sensitivity, is calculated as True Positive (TP) / (TP + FN), as such the denominator in this case is 23. Specificity is calculated as True Negatives (TN) divided by True Negatives and False Positives (TN + FP), and the denominator is 557. Positive Predictive Value (PPV) = TP / (TP + FP), with a denominator of 122 and Negative Predictive Value (PPV) = TN / (TN + FN) with a denominator of 458.
| Arm | Measure | Group | Value (COUNT_OF_PARTICIPANTS) |
|---|---|---|---|
| Intervention | Effectiveness AI-ECG for Cardiomyopathy Detection in the Intervention Arm in LVEF < 45% | LVEF < 45% (Sensitivity) | 22 Participants |
| Intervention | Effectiveness AI-ECG for Cardiomyopathy Detection in the Intervention Arm in LVEF < 45% | LVEF < 45% (Specificity) | 457 Participants |
| Intervention | Effectiveness AI-ECG for Cardiomyopathy Detection in the Intervention Arm in LVEF < 45% | LVEF < 45% (PPV) | 22 Participants |
| Intervention | Effectiveness AI-ECG for Cardiomyopathy Detection in the Intervention Arm in LVEF < 45% | LVEF < 45% (NPV) | 457 Participants |
Secondary
Effectiveness AI-ECG for Cardiomyopathy Detection in the Intervention Arm in LVEF < 50%
This is defined as a positive point-of-care AI prediction for LVEF \<50% (maximum prediction across all stethoscope recording locations) confirmed with echocardiography
Time frame: 18 months
Population: To explain the differences in number analyzed for each section below, details are provided: Sensitivity, is calculated as True Positive (TP) / (TP + FN), as such the denominator in this case is 23. Specificity is calculated as True Negatives (TN) divided by True Negatives and False Positives (TN + FP), and the denominator is 557. Positive Predictive Value (PPV) = TP / (TP + FP), with a denominator of 122 and Negative Predictive Value (PPV) = TN / (TN + FN) with a denominator of 458.
| Arm | Measure | Group | Value (COUNT_OF_PARTICIPANTS) |
|---|---|---|---|
| Intervention | Effectiveness AI-ECG for Cardiomyopathy Detection in the Intervention Arm in LVEF < 50% | LVEF < 50% (Sensitivity) | 22 Participants |
| Intervention | Effectiveness AI-ECG for Cardiomyopathy Detection in the Intervention Arm in LVEF < 50% | LVEF < 50% (Specificity) | 457 Participants |
| Intervention | Effectiveness AI-ECG for Cardiomyopathy Detection in the Intervention Arm in LVEF < 50% | LVEF < 50% (PPV) | 22 Participants |
| Intervention | Effectiveness AI-ECG for Cardiomyopathy Detection in the Intervention Arm in LVEF < 50% | LVEF < 50% (NPV) | 457 Participants |
Secondary
Effectiveness of AI-ECG for Cardiomyopathy Detection in the Intervention Arm for Left Ventricular Ejection Fraction (LVEF) ≤ 35%
This is defined as a positive point-of-care AI prediction for LVEF ≤ 35% (maximum prediction across all stethoscope recording locations) confirmed with echocardiography
Time frame: 18 months
Population: To explain the differences in number analyzed for each section below, details are provided: Sensitivity, is calculated as True Positive (TP) / (TP + FN), as such the denominator in this case is 17. Specificity is calculated as True Negatives (TN) divided by True Negatives and False Positives (TN + FP), and the denominator is 563. Positive Predictive Value (PPV) = TP / (TP + FP), with a denominator of 122 and Negative Predictive Value (PPV) = TN / (TN + FN) with a denominator of 458.
| Arm | Measure | Group | Value (COUNT_OF_PARTICIPANTS) |
|---|---|---|---|
| Intervention | Effectiveness of AI-ECG for Cardiomyopathy Detection in the Intervention Arm for Left Ventricular Ejection Fraction (LVEF) ≤ 35% | LVEF ≤ 35% (Sensitivity) | 17 Participants |
| Intervention | Effectiveness of AI-ECG for Cardiomyopathy Detection in the Intervention Arm for Left Ventricular Ejection Fraction (LVEF) ≤ 35% | LVEF ≤ 35% (Specificity) | 458 Participants |
| Intervention | Effectiveness of AI-ECG for Cardiomyopathy Detection in the Intervention Arm for Left Ventricular Ejection Fraction (LVEF) ≤ 35% | LVEF ≤ 35% (PPV) | 17 Participants |
| Intervention | Effectiveness of AI-ECG for Cardiomyopathy Detection in the Intervention Arm for Left Ventricular Ejection Fraction (LVEF) ≤ 35% | LVEF ≤ 35% (NPV) | 458 Participants |
Other Pre-specified
Composite Adverse Cardiovascular Events
The number of subjects to experience composite cardiovascular events with include any of the following: diastolic heart failure, gestational hypertension, pre-eclampsia, eclampsia, valvular heart disease, atrial arrhythmias and sustained ventricular arrhythmias.
Time frame: 18 months
| Arm | Measure | Value (COUNT_OF_PARTICIPANTS) |
|---|---|---|
| Intervention | Composite Adverse Cardiovascular Events | 56 Participants |
| Control | Composite Adverse Cardiovascular Events | 53 Participants |
p-value: 0.62195% CI: [0.74, 1.64]Regression, Logistic
Other Pre-specified
Echocardiography Utilization
Determine the impact of an AI-ECG on echocardiography utilization
Time frame: 18 months
Population: Data was not collected nor analyzed for this outcome measure
Other Pre-specified
Effectiveness of AI Point of Care Tools for Cardiomyopathy Detection in the Intervention Arm
Develop and evaluate the diagnostic performance of an AI-enhanced point of care screening tool
Time frame: 18 months
Population: Data was not collected nor analyzed for this outcome measure