Highlights

2020-2025

INTERVENE set out to advance the AI-facilitated analyses of complex medical data to develop genetic risk scores, which would improve our understanding of diseases and treatment options tailored to individuals.

The core aims of the project were to:

Harmonise data on more than 1.7 million genotyped / sequenced genomes with longitudinal clinical data and facilitating access to summary statistics for comparative meta-analysis.

Develop AI-based integrative risk scores for the next generation of predictive and personalised medicine with continuous improvement and disease coverage of these scores
Test the utility of novel risk scores in clinical settings and developing a strategy for their implementation in Europe and beyond.

During five years of enthusiastic high-quality research and active collaboration INTERVENE has succeeded in testing and producing genetic risk scores that condense information from genetic variants and other health data into a number that reflects a person’s inherited susceptibility towards a disease. Furthermore, INTERVENE has developed tools for calculating risk scores in international large-scale, multi-ethnic biobank projects and piloted their use in real-world clinical settings. Throughout the project INTERVENE has strived to communicate its research widely and has paid close attention to the ethical and societal considerations connected to using genetic information and artificial intelligence to support decision making.

This article showcases the highlights of INTERVENEs achievements:

Papers that push forward the frontiers of predictive genomics

INTERVENE has published its scientific work in numerous high-quality journals including Nature Communications, Nature Aging, Nature Medicine and Nature Genetics.

The project’s Flagship paper “A unified framework for estimating country-specific cumulative incidence for 18 diseases stratified by polygenic risk” authored by Bradley Jermy, Kristi Läll and Brooke Wolford, was published in Nature Communications in 2024.

The paper describes a new framework for estimating cumulative disease incidences over the life course and allows stratification for risk-based prevention and screening using PGS accounting for country, age and sex for 18 high-burden diseases. The ambitious undertaking of combining 1.2 million biobanked samples across Europe revealed that in many diseases the effects of polygenic risk scores are much higher in young individuals compared to older participants and some effects are also different for males and females. https://doi.org/10.1038/s41467-024-48938-2

Other scientific highlights include:

“Identifying interpretable gene-biomarker associations with functionally informed kernel-based tests in 190,000 exomes” by Monti et al, published in Nature Communications in 2022, which is an exome-wide rare genetic variant association study for 30 blood biomarkers demonstrating the benefits of investigating different functional mechanisms when performing rare-variant association tests. https://doi.org/10.1038/s41591-022-01957-2

And “Cross-biobank generalizability and accuracy of electronic health record-based predictors compared to polygenic scores”, by Kira Detrois, Tuomo Hartonen et al., published in Nature Genetics in 2025, which highlights the benefits of combining health records and genetics to predict disease risk. https://doi.org/10.1038/s41588-025-02298-9

INTERVENE researchers have presented their work at many prestigious scientific forums, including the annual meetings of the American Society of Human Genetics and the European Society of Human Genetics.

Clinical pilots that inform practice and policy

To validate the use of integrative genetic risk scores in a real-world setting INTERVENE carried out clinical pilot studies targeting major diseases with high public health impact, including breast cancer and cardiovascular disease.

Breast cancer pilot (Helsinki, Sienna, Turin)

The pilot study first recruited a cohort of 900 female breast cancer patients, including individuals with invasive disease as well as carcinoma in situ. Each participant underwent a comprehensive genetic risk assessment encompassing pathogenic variants in known high- and moderate ]penetrance breast cancer susceptibility genes, as well as a polygenic risk score. Based on these assessments, healthy female first-degree relatives of patients with elevated genetic risk were subsequently recruited to evaluate the feasibility and clinical utility of PRS in genetic testing and counselling.

The findings made in the pilot confirm that PRS makes an important contribution to breast cancer risk assessment: 25% of enrolled patients had elevated PRS, and integrated risk assessment including PRS identified 30% of relatives as being at elevated risk. At the same time, having a close relative with genetic risk for breast cancer was associated with substantial worry, and the overwhelming majority of relatives expressed a desire to better understand their own risk.

The pilot included the development of an accessible web-based risk calculator including graphical elements, which is expected to help women understand their risk. When used in counselling sessions, the tool may enhance personal interaction between health care professionals and patients or their relatives and help make informed decisions.

CVD pilot (Tartu)

The pilot study was a randomized controlled trial of 1020 young overweight individuals and was conducted to evaluate whether integrating polygenic risk scores (PRS) for coronary artery disease (CAD) into cardiovascular risk communication could effectively influence health behavior and improve clinical outcomes.

The results showed that communicating high coronary artery disease (CAD) polygenic risk scores (PRS) did not lead to behavior changes sufficient to reduce cardiovascular risk factors over 12 months in young overweight, but otherwise healthy adults. This indicates that a resource-intensive intervention does not produce measurable benefits to the health care ecosystem compared to current approaches for cardiovascular risk assessment in young adults. Although knowledge of polygenic risk does not seem to motivate behavioral risk in citizens, the pilot increased the initiation of preventive treatments. This indicates that clinicians became more aware of genetic factors underlying the risk for CVD, which is an important dimension in the implementation of health care programmes that are based on genetic risk.

A platform for calculating polygenic risk scores

A major achievement of INTERVENE is the development a federated AI-enabled data analysis platform that enables secure, distributed computation across partner biobanks, integrating genomic, omics, and longitudinal health data to generate advanced integrative genetic scores.

GeneticScores.org allows researchers to generate and interpret PGSs through two complementary modes of operation:

1. An online, cloud-hosted PGS calculation service, enabling users to upload encrypted genomic data, select polygenic scores from the PGS Catalog, and run analyses through a validated bioinformatic pipeline.

2. The PGS Catalog Calculator (pgsc_calc), an open-source, portable workflow implementing the same validated computation logic, designed for use within institutional clusters and national Trusted Research Environments (TREs), where most sensitive genomic data are now processed.

Together, these components deliver a unified, reproducible, and FAIR-aligned framework for PGS analysis across a wide range of computational and governance contexts.

GeneticScores.org and pgsc_calc now form a trusted foundation for reproducible, equitable access, and scalable PGS analysis across the international research ecosystem, ensuring continued impact beyond the INTERVENE funding period.

GeneticScores.org

Training early career researchers

INTERVENE offered several talented early career researchers in genetics and AI to develop their skills by working in the different teams of the consortium. Brooke Wolford (NTU), Kristi Läll (Tartu), Sophie Warrie (Aalto), Remo Monti (HPI), Tuomi Hartonen (FIMM), Lisa Eick (FIMM), Stavroula Kanoni (QUML), Fiona Hagenbeek (FIMM), Aoife McMahon (EBI), Benjamin Wingfield (EBI) not only contributed to the research in their respective groups and institutes, but met regularly in analyst meetings to promote collaboration and synergy within the consortium. Their individual and joint efforts led to many high-impact scientific papers and were crucial for the development and implementation of the GeneticScores.orgv and the PGS Catalog Calculator. After having been trained in a multidisciplinary international consortium each ECR can be expected to flourish in a career in genomics and contribute to the field in the future.

Outreach and engagement to inform and impact

INTERVENE has built a comprehensive ethical, legal, and social (ELSI) framework and made sure it addressed issues such as data privacy, regulatory compliance, patient consent, public acceptance, and equitable access to genetic risk tools. This work was done in close collaboration between researchers in genetics and AI and researchers in ethics. Public outreach and engagement were facilitated by partnering with patient organisations, who were able to condense the often complex concepts of the research and results into comprehensible summaries, which they disseminated through their networks to patients and their families. The consortium reached out to the wider public by creating short, accessible videos and speaking on podcasts and to high-school students. The researchers wrote opinion pieces and presented at conferences of medical societies to connect with health care professionals and policy makers.

Five years of EU funded research and innovation have laid a solid foundation for AI-aided predictive genomics and personalized health care. For a look at the long-term impact of INTERVENE, read this article written by INTERVENE partner, PNO. How INTERVENE shapes the future of predictive genomics | PNO Innovation