Not provided
Not provided
Not provided
Not provided
Not provided
Not provided
Not provided
Not provided
Not provided
Not provided
Not provided
Not provided
Not provided
Not provided
Not provided
| Name | Class |
|---|---|
| VISUAL INTERACTION & COMMUNICATION TECHNOLOGIES - VICOMTECH | UNKNOWN |
| Clinic for Cardiovascular Diseases Magdalena | NETWORK |
| Biokeralty Research Institute | INDUSTRY |
Not provided
Not provided
Not provided
Not provided
This retrospective observational study, part of the EU-funded CARAMEL project, aims to develop and validate personalized cardiovascular disease (CVD) risk assessment models specifically designed for menopausal and perimenopausal women (ages 40-60). The study leverages Real World Data (RWD) collected from multiple international clinical partners, including electronic health records (EHR), diagnostic imaging data, and signal data.
The main objective is to improve the prediction of CVD precursors such as hypertension and dyslipidemia, as well as mid- and long-term risk of CVD events, through advanced artificial intelligence (AI) models. These models will be trained on multimodal data to capture complex, individualized risk trajectories that current risk calculators fail to address, particularly in women. Special focus is placed on under-researched, women-specific risk factors and their interactions with traditional predictors.
The study includes several research objectives: (1) predicting the onset of hypertension and dyslipidemia using EHR data; (2) modeling the long-term risk of fatal and non-fatal cardiovascular events and disease trajectories; (3) identifying novel imaging biomarkers from routine screening tests such as mammography, DXA, ultrasound, and cardiac MRI; (4) developing multimodal prediction models combining imaging and clinical data; (5) creating automated AI tools for imaging biomarker extraction; and (6) using signal data from cardiac devices to predict disease progression and events.
The study population consists of middle-aged women with retrospective data available across different health systems. The expected outcome is a validated set of stratified, personalized CVD risk models that can support targeted prevention strategies and enable more equitable, sex-specific care. This will contribute to reducing the burden of CVD in women and addressing critical gaps in early detection, clinical decision-making, and health policy.
This project has received funding from the European Union's Horizon Europe Research and Innovation Programme under Grant Agreement No 101156210.
Not provided
Not provided
Not provided
Not provided
Not provided
Not provided
| Label | Type | Description | Intervention Names |
|---|---|---|---|
| ASCIRES IMAGE DATABASE | Digital imaging biobank 10y long from several manufact 1,000 cMRI; 500 cardiac CT; 500 coronary artery calcification; 1,000 DXA From women 40- 60y urers / modalities | ||
| Basque Health Service Database | Longitudinal EHR data up to 15y including diagnosis, procedures, prescriptions, lab tests, visits, imaging, etc. ~128,00 women 40-60 14,880 DM, 3,124 DXA, 332 carotid US | ||
| Clalit Primary Prevention Database | Manually curated DB of structured EHR data ~750,000 middleaged women | ||
| Irish Implant Devices Registry | Irish Implant Devices Registry (REG) (HRI) 15y of data for implant procedures and follow-ups (pacemakers, ICD's, loop recorders) ~85,000 implant (pacemaker) proced ures ~700,000 follow-up w. indications & diagnosis | ||
| Keralty Colombia Database | EHR data from primary/specialised care centres. Longitudinal EHR data up to 5-10y Including diagnosis, procedures, prescriptions, lab tests, visits, etc. ~85,593 women 40-60y ~25,000 women with CVD problems | ||
| Andalusian Health Population Database & Macarena University Hospital EHR | Longitudinal EHR data up to 15y including diagnosis, clinical procedures, prescriptions, lab tests, visits, etc. The hospital Dataset is OMOP CMD mapped ~700,000 middleaged women |
Not provided
| Measure | Description | Time Frame |
|---|---|---|
| Occurrence and Predicted Risk of Cardiovascular Disease (CVD) Events (fatal and non-fatal) | The study will retrospectively evaluate the occurrence of cardiovascular disease (CVD) events and develop predictive models to estimate individual risk profiles for such events. CVD events include both fatal and non-fatal occurrences such as myocardial infarction, stroke, heart failure, arrhythmias, and atherosclerotic disease. Events will be identified using structured electronic health records (EHR) and coded using ICD-10 classifications. Risk will be modeled using multimodal data sources (EHR, imaging, and signals) to predict short- and long-term outcomes, stratified by individual characteristics. The outcome integrates: Event-based measures: Time to first fatal or non-fatal CVD event. Risk-based measures: Individual predicted probabilities of experiencing a CVD event or precursor condition (e.g., hypertension, dyslipidemia) over different time frames. | up to 10 years |
| Measure | Description | Time Frame |
|---|---|---|
| RO1. Personalized risk prediction of CVD precursors | First observation of HT or DY registered in the EHR, registered as a diagnostic code, or as a laboratory result or test. These include:
|
Not provided
Inclusion Criteria:
Self-identified as female in the electronic health record (EHR). Age between 40 and 60 years at the time of data collection/index date. Availability of at least 5-6 years of retrospective data in the EHR, depending on the research objective.
At least one healthcare encounter (visit, imaging, lab test, diagnosis, etc.) within the defined age range.
For imaging substudies (e.g., RO3-RO5): availability of at least one relevant imaging test (e.g., DXA, digital mammography, cMRI, CCTA, US) during the age range.
For signal-based analysis (RO6): presence of ECG monitoring data from implanted devices and at least 2 years of follow-up.
Exclusion Criteria:
Prior diagnosis of cardiovascular disease before the observation window (only applicable to specific ROs, e.g., RO2, RO4).
Insufficient data quality or missing key variables needed for modeling (e.g., absence of blood pressure or lipid profile).
Patients with incomplete or inconsistent records (e.g., duplicate IDs, mismatched time frames).
For signal-based RO6: hospitalizations or diagnoses unrelated to cardiovascular health that may bias AI model training.
The study intentionally includes only biological women aged 40-60 to address a well-documented gender gap in cardiovascular research. Women, particularly during menopause, are underrepresented in clinical studies and underserved by existing CVD risk models, which are largely based on male populations. This gender-specific focus aims to develop tailored risk prediction tools that reflect the unique physiological, hormonal, and clinical characteristics of women during this high-risk transition period.
Participants are identified retrospectively from electronic health records, imaging archives, and device registries across multiple healthcare systems and countries
Not provided
Not provided
Not provided
Not provided
Not provided
Not provided
| Keralty SAS. Colombia |
| OTHER |
| ETHNIKO KAI KAPODISTRIAKO PANEPISTIMIO ATHINON | UNKNOWN |
| Fundación Pública Andaluza para la gestión de la Investigación en Sevilla | OTHER |
| University of Dublin, Trinity College | OTHER |
| TREE Technology S.A. | UNKNOWN |
| Dublin City University | OTHER |
| Tampere University | OTHER |
| Ben-Gurion University of the Negev | OTHER |
| Biogipuzkoa Health Research Institute | OTHER |
| Vilnius University Hospital Santaros Klinikos | OTHER |
Not provided
Not provided
Not provided
| Lithuanian High Cardiovascular Risk (LitHiR) primary prevention programme database | EHR data from primary cardiovascular prevention programme in VULSK (1 centre). Data including demographics, risk factors, lab tests (including lipid profile, renal function, etc.), arterial markers (pulse wave velocity analysis data; CardioAngle Vascular Index data; carotid artery intimamedia thickness data). Some patients have 5-10y longitudinal data with outcomes. ~6000 women 40-65y with high - very high cardiovascular risk, but without overt CVD; |
| National and Kapodistrian University of Athens Database - Aretaieion Hospital | EHR data from Menopause clinic of Aretaieion university hospital including blood tests, medication, prescriptions, visits ~4000 middle aged women |
| CoroPrevention - Tampere University (TAU) | Pan-European (25 sites) contemporary prospective CVD prevention cohort from ongoing HEU project it includes clinical data, 3-year CV event data, lifestyle, RFs. Standard + CVD biomarkers (CERT2, hsTNI, NTproBNP, Cystatin C…) N=~3,000 women (subsample of whole cohort) |
| AKRIBEA - Cooperative Research Centre for Biosciences Association (CIC) | Non-oriented 7y follow-up cohort from Basque Country Region. Urine+serum biomarkers and metabolome; serum lipoproteins by NMR; demographics & RFs N=~ 2,500 women (40 to 60 y) |
| MENO - Cooperative Research Centre for Biosciences Association (CIC) | Pre- and post-menopausal women cohort from Basque Country Region. Urine+serum biomarkers and metabolome; serum lipoproteins by NMR; demographics & RFs N =~ 1,700 women |
| UK Biobank - UK Biobank | Largest geno-phenotype-rich population-based study in the world (500K), includes multi-modal imaging data (60K) and eye and vision (67K), biomarkers, demographic data, lifestyle (100K with wearables) and health outcomes. Middle-aged women among:
|
| Qatar Biobank | Population-based with annotated data, biological samples, tests and imaging for 60K participants. It includes Demographics data, lifestyle, biomarkers, weight & body fat, hip&waist, BP, ECG, carotid US, full-body MRI, retinography, DXA Middle-aged women among ~60K total participants |
| International Agency for Research on Cancer (IARC) / EPIC-Europa | Long-term European population-based cohort (520K participants across 10 countries). Includes clinical data, anthropometric measurements, demographic, lifestyle, dietary habits, and socioeconomic data, reproductive history, and biological samples such as serum, plasma and DNA for biochemical data and genotyping data N = ~367k women between 35 to 65 years old (subsample of whole cohort) ~65k CVD cases across the full cohort |
| ILERVAS -Institute for Research in Biomedicine IRB Lleida | Interventional longitudinal study that includes detailed assessments of subclinical atheromatosis in 12 vascular territories using ultrasound, along with clinical, anthropometric, lifestyle, dietary, and biochemical data. N = ~4165 women (50 to 70y) (subsample of whole cohort) |
| up to 8 years |
| RO2. Personalized Risk Prediction of CVD Events and CVD trajectories | The occurrence of CVD events, which will be classified in fatal (if they are registered as the cause of death) or not fatal (if they are not registered as cause of death).
| Up to 16 years |
| RO3. Novel Imaging Biomarkers and Patterns for CVD Risk Assessment | Evaluates the predictive performance of multimodal models combining imaging features (e.g., cardiac MRI, DXA, digital mammography) and electronic health record (EHR) variables to estimate the mid- and long-term risk of cardiovascular events (CVD) in women aged 40-60. The endpoint is the first occurrence of a fatal or non-fatal CVD event after the imaging test, as documented in the EHR. The models will be compared against standard risk assessment tools (e.g., SCORE2). | Baseline |
| RO4. Multimodal EHR and ImageBased CVD Prediction Models | The occurrence of CVD events, which will be classified in fatal (if they are registered as the cause of death) or not fatal (if they are not registered as cause of death).
| Up to 16 years |
| RO5. Automatic imaging marker and pattern extraction | The performance and clinical relevance of AI-based tools for the automatic extraction of cardiovascular imaging biomarkers in women aged 40-60. These tools will be used to segment anatomical regions and calculate quantitative measures from multimodal imaging (e.g., ultrasound, DXA, cardiac CT, cMRI, mammography). | Baseline |
| RO6. Signal-based CVD prediction models | Occurrance of CVD events, which include:
| Up to 16 years |
| ID | Term |
|---|---|
| D002318 | Cardiovascular Diseases |
Not provided
Not provided
Not provided