Polygenic Scoring and Translational Health Data Science: A Strategic Analysis of Clinical Readiness and Equity Barriers
- Nelson Advisors
- Oct 30
- 17 min read
Foundational Principles of Polygenic Risk Scoring
The implementation of Polygenic Risk Scores (PRS) represents a fundamental advancement in quantifying the genetic liability for complex diseases. This approach acknowledges that common traits and disorders are influenced not by single variants, but by the cumulative effect of thousands of genetic markers, each contributing an attenuated but measurable impact. Understanding the mechanisms by which these scores are constructed, from basic summation models to sophisticated machine learning frameworks, is essential for evaluating their translational potential.
Defining Polygenicity and the Standard Construction Pipeline
The core methodology for calculating a standard PRS involves aggregating the effects of risk-modifying alleles across a vast number of Single Nucleotide Polymorphisms (SNPs).This process relies heavily on statistical outputs from large-scale Genome-Wide Association Studies (GWAS).
The standard calculation involves multiplying the number of risk-increasing alleles for each SNP by a specific weight, which represents the estimated effect size for that SNP derived from the GWAS. Mathematically, the estimated PRS is obtained as a weighted sum of allele counts across the chosen set of $M$ SNPs. This technique is rooted in regression analysis, effectively summarising an individual's total genetic load for a given phenotype. Critically, the aggregate nature of PRS confers superior predictive performance. Studies have demonstrated that this comprehensive approach yields substantially greater predictive power for complex traits compared to models that rely only on a small subset of genome-wide significant SNPs.
Advanced Predictive Modelling Techniques
The standard linear summation model, while foundational, operates under the simplifying assumption that genetic variants act independently. This assumption often proves inaccurate in biological systems and fails to adequately account for Linkage Disequilibrium (LD) the correlation between alleles at different loci within a population. The limitations of linear modelling necessitate a shift toward more sophisticated, LD-aware methodologies to maximise the predictive signal extracted from genomic data.
Bayesian and Penalised Regression Methods
To overcome the challenges posed by LD and the reliance on simple P-value thresholding (P+T), advanced statistical genetics methods have been developed. Methods such as LDpred and lassosum utilise penalised regression frameworks to refine SNP weights by incorporating LD information derived from a reference population panel. Lassosum, for instance, employs an L1 penalty within its function, enabling a more accurate estimation of effect sizes. The strategic adoption of these methods is paramount because practical application of PRS globally often relies on publicly shared GWAS summary statistics rather than restricted individual-level genotype data.
Since access to individual biobank data is limited by logistical, security and ethical constraints, methods that optimise prediction using readily available summary statistics provide the most scalable pathway for clinical deployment and continuous improvement. For instance, simulations have indicated that lassosum not only achieves better prediction accuracy than P+T but is also substantially faster and more accurate than the previously proposed LDpred. This strategic focus on LD-aware methods is critical for maximising the utility of existing GWAS data.
Integration of Machine Learning (ML) and Deep Learning (DL)
To address the inherent non-linear nature of complex trait etiology, a further methodological refinement involves integrating advanced computational techniques. Machine learning algorithms, including random forests, neural networks, and gradient boosted trees, are explicitly non-linear and capable of modelling interaction effects, such as Gene-Gene (Epistasis) and Gene-Environment (GxE) interactions.
Deep learning neural networks, in particular, are being extensively investigated for their capacity to model non-linear relationships and their potential to capture the full extent of genetic variability that simple linear summation models fail to explain. The incorporation of a standard PRS as a feature within a larger ML model can significantly increase the percentage of variance explained for certain complex traits. This necessary methodological shift away from traditional linear assumptions toward non-linear, interaction-aware modelling is driven by the mandate to capture "missing heritability" and significantly improve predictive power.
However, a major barrier to the clinical translation of these highly accurate DL models is the lack of standardized reporting and interpretability. While DL models show promising signs for improved predictive power, they often function as "black boxes," making the inference of causation difficult and challenging for clinical trust and regulatory validation. This technical hurdle creates a paradox: the most accurate predictive models are currently the most difficult to translate into clinical practice because of the inherent complexity in explaining their predictive basis to clinicians and patients. Consequently, the industry faces the challenge of establishing clear benchmarks on consistent datasets to ensure that improvements derived from advanced architectures are quantifiable and reproducible for clinical acceptance.
Clinical Utility, Risk Stratification, and Performance Validation
The true value of polygenic scoring is measured not just by statistical power, but by its ability to inform preventative strategies and refine risk management in clinical settings. PRS functions most powerfully as a biomarker for genetic liability, capable of identifying individuals at the extremes of the population risk distribution.
Quantifying Clinical Impact: Case Studies
PRS methodologies have proven highly generalisable across a variety of complex traits, particularly in conditions where early intervention can significantly alter disease trajectory.
Cardiovascular Disease and Metabolic Conditions
In cardiovascular disease (CAD), PRS offers significant potential for lifetime risk stratification. Large-scale cohort studies, such as the UK Biobank, demonstrate that individuals in the highest PRS percentiles face a $3$- to $5$-fold higher risk of CAD compared to the general population. This stratification is highly actionable: individuals identified as having high genetic risk may experience the greatest therapeutic benefit from interventions like statin therapy. Furthermore, PRS can be utilised to reclassify younger adults categorised as having borderline or intermediate risk based on traditional factors and it can complement established diagnostic tools such as coronary artery calcium (CAC) scoring. Beyond CAD, PRS identifies significant proportions of European individuals at greater than threefold risk for other common diseases, including atrial fibrillation (up to 6.1%), Type 2 Diabetes (T2D) (3.5%), and breast cancer (1.5%).
This evidence demonstrates that the immediate strategic value of PRS is not in broad diagnostic screening, but in identifying the genetically highest-risk individuals early in life. This allows for the implementation of intensified, pre emptive interventions before the onset of clinical symptoms, maximising the opportunity for primary prevention.
Pharmacogenomics and Trial Enrichment
The utility of PRS extends into drug development, where scores are increasingly employed to enrich clinical trials for individuals more likely to respond to treatment or to predict adverse events. A review of documents submitted to the Food & Drug Administration (FDA) reveals increasing use of PRS, particularly in therapeutic areas such as neurology, psychiatry and oncology.
Most PRS applications in this context support secondary or exploratory analyses within early drug development phases (Phase 1, 1/2, or 3). This utilisation creates a foundational mechanism for future regulatory approval. If a drug's efficacy is demonstrated specifically within a high-PRS sub-cohort, it validates the PRS as an effective companion diagnostic or biomarker for therapeutic targeting. This approach allows PRS to bypass initial regulatory demands for high population-level discriminative accuracy (as measured by AUC) and instead focuses on establishing predictive validity within therapeutically defined subsets, accelerating the pace of personalised medicine.
Essential Metrics for Translational Evaluation
Accurate assessment of clinical readiness requires metrics that move beyond simple statistical fit to quantify genuine clinical actionability and cost effectiveness.
Limitations of Traditional Metrics
The Area Under the Receiver Operating Characteristic curve (AUC) is a standard statistical measure used to evaluate prediction model performance when genotypic and phenotypic data are available. However, the AUC, as well as the Subset Relative Risk (SRR), which has been shown to be essentially a relabelling of AUC, may not correlate directly with expected clinical utility. The evaluation of PRS depends critically on the metric chosen for interpretation.
Focusing on Actionability and Utility
A metric directly related to the expected clinical utility is essential for translating genetic findings into practice. The Minimum Test Trade off is recommended as a preferred metric, as it calculates the minimum number of genetic variant ascertainment’s (one per person) required for each correct prediction of disease (eg. cancer) to achieve a positive expected clinical utility.This provides a practical, decision-focused measure tied to the anticipated benefits and costs of intervening based on the PRS result.
Economic Evaluation and Real-World Data
Analysis of the cost-effectiveness of PRS-based interventions shows a general positive trend toward utility, particularly in optimising screening programs for cancer and refining eligibility for preventive therapies in cardiovascular disease. However, the prevailing optimistic view of PRS cost-effectiveness is often based on fragile foundations. Systematic reviews highlight that economic studies frequently rely on hypothetical cohorts, have limited generalisability and pay insufficient attention to critical implementation costs, such as the delivery model and workflow reconfiguration.
Realised cost-effectiveness will be significantly lower if institutions fail to account for the substantial implementation costs associated with EHR integration, provider education and process optimisation. Further research and prospective pilot studies utilising real-world data across diverse populations are necessary to rigorously evaluate the total costs and benefits of PRS applications.
The current understanding of clinical risk stratification based on PRS is summarised below:
Observed Clinical Risk Stratification for Select Diseases
The Digital Nexus: Integrating PRS into the Healthcare Ecosystem
The transition of PRS from research utility to clinical practice is predicated upon the development of a robust, interoperable data infrastructure that integrates genomic results with longitudinal patient health data.
Biobanks and Electronic Health Records (EHRs)
The primary engine for advancing PRS development and application is the linkage of large-scale genetic biobanks with Electronic Health Records (EHRs). This convergence provides unparalleled opportunities to systematically examine patient disease susceptibilities and enables genomic discovery by leveraging vast amounts of phenotype data. Linked data allows researchers to study the genetic basis of common and rare disorders, assess the pathogenicity of novel variants, and rapidly assemble cohorts for genomic medicine clinical trials, establishing a "virtuous cycle" of the learning healthcare system.
However, the quality of PRS derived from this data is critically dependent on the accuracy and completeness of the EHR phenotyping. The heterogeneous nature of EHR data, often collected primarily for medical documentation and billing purposes (eg. using International Classification of Diseases codes), presents significant practical challenges. Inconsistent phenotyping or missing data limits the power and specificity of derived PRS models. The integrity of this EHR-PRS feedback loop is vital: if EHR phenotyping is incomplete or unreliable, the resulting PRS will be inaccurate, leading to reduced clinical utility and decreased incentive to improve EHR data quality, creating a detrimental feedback cycle that hinders progress.
Informatics Standards and Interoperability
To ensure that genomic data can be utilized dynamically at the point of care, standardized data exchange formats are necessary. Historically, most genetic test results entered into the EHR were static PDF reports, severely limiting their use in computational tools.
The development of the Health Level 7 Fast Healthcare Interoperability Resources (FHIR) standard, specifically FHIR Genomics, is the leading informatics solution for this challenge. FHIR Genomics is designed to facilitate the feasible and efficient exchange of complex clinical genomic data and interpretations. The FHIR framework consists of linkable and extendable data structure specifications, modelling essential healthcare concepts (eg. patients, conditions, clinical observations). By enabling apps to be launched directly from within the EHR (via SMART on FHIR platforms), this standard allows for the tailoring of genomic resources to use cases like integrating PRS results, which is indispensable for realising precision medicine.
Workflow Integration and Actionability
The technical feasibility provided by FHIR only addresses how the data is exchanged; the challenge of what to do with the result, actionability, must be solved at the clinical workflow level. Implementing a new system, even an EHR-integrated Patient-Reported Outcomes (PRO) system, necessitates a change in clinical workflow and organisational culture.
Providers, including physicians, medical assistants and psychosocial providers, frequently face barriers stemming from a lack of actionable data, technical issues, and workflow disruption. For PRS to be effective, it must facilitate targeted conversation with patients and provide automated triage for specialized care. This requires strategic investment that is split between developing informatics standards (FHIR) and designing highly intuitive Clinical Decision Support (CDS) interfaces. These interfaces must translate complex PRS output (such as a percentile risk score) directly into clear, next-step clinical guidance, such as auto-referral for genetic counselling or initiating preventative screening.
The adoption of PRS thus mandates a substantial, non-trivial investment in interoperability infrastructure and workflow reconfiguration across healthcare institutions. This cost, which includes provider training and optimisation of referral processes, may surpass the cost of the genetic testing itself, representing a critical, often hidden, barrier to achieving positive cost-effectiveness in the real world. The success of PRS hinges on this seamless, friction-free integration into the existing clinical framework.
The Equity Imperative: Addressing Ancestral Bias and Generalisability
The most significant scientific and ethical hurdle confronting the widespread clinical use of PRS is the systematic disparity in predictive accuracy across human populations. This limitation stems directly from biases in the foundational research data.
The Eurocentric Bias Crisis
The development of GWAS has historically been heavily skewed toward European ancestry populations. Roughly 78% of individuals included in all GWAS are of European ancestry, leading to an oversampling that generates inherent bias.
Quantifying the Disparity
This Eurocentric bias means that current PRS are "many-fold more accurate" and perform significantly better in individuals of European descent. This poor generalisability across diverse ancestries and cohorts is the biggest current limitation of PRS. The decreased predictive performance of European ancestry-derived PRS in non-European samples is a consequence of fundamental differences in linkage disequilibrium (LD) patterns and variant frequencies between ancestral groups.
For example, studies analysing coronary heart disease (CHD) show a stark drop-off in association strength in non-European ancestries. While the predictive power (odds ratio) remains high in European and South Asian ancestries, the association is markedly lower in East Asian ancestry and weakest in Hispanic/Latino and African ancestry groups.
Exacerbating Health Disparities
The clinical use of these skewed PRS poses a direct ethical risk: it threatens to exacerbate existing health disparities. Since the predictive power is reduced for non-European populations, the benefit of early risk stratification and targeted prevention will systematically afford greater improvement to European-descent populations, even though minority groups often bear a higher burden of disease incidence and mortality (eg. prostate cancer in men of African ancestry). The generation of "false results" (ie. less accurate risk estimates) for ancestrally diverse individuals means that an inequitable genetic tool risks widening societal gaps, even though health disparities primarily stem from systemic factors, not genetics.
Strategies for Multi-Ancestry PRS Development
Addressing the generalisability crisis requires a dual approach focusing on radical data diversification and refined methodological techniques.
Data Diversification and Strategic Focus
Major data generation initiatives are underway globally to improve generalizability, often leveraging ancestrally diverse genomic data linked to health records, such as those within the All of Us Research Program. The analytical evidence suggests that while differences in heritability across populations contribute to generalisability challenges, the massive disparity in sample sizes across ancestries currently plays a much larger and more actionable role in PRS accuracy differences. Therefore, the most immediate and impactful mitigation strategy is aggressive funding and strategic execution of large-scale, high-quality, diverse genomic data generation.
Early efforts confirm that diversification shows promise in levelling this imbalance, even when non-European sample sizes remain smaller than the largest European studies. Furthermore, researchers must prioritise the public dissemination of GWAS summary statistics for non-European cohorts to ensure improved scores are universally available.
Methodological Advances for Portability
To decouple PRS performance from population-specific LD patterns, new methodological frameworks are essential:
Trans-Ethnic Meta-Analysis: Combining data from multiple ancestries allows for the discovery of new genetic loci and the refinement of existing risk regions, leading to the development of risk scores that are more transferable across population groups.
Leveraging Functional Annotations: Methods that partition SNP heritability by functional annotations (eg. focusing on variants in regulatory or coding regions) help differentiate SNPs that are potentially causal and explain a larger portion of heritability. By prioritising causal variants over proxy variants associated via LD, the derived score becomes inherently more portable across diverse genetic backgrounds.
Deep Learning Latent Representations: Advanced deep learning architectures that utilise latent representations in autoencoders have shown potential for improving predictive performance across diverse ancestries by extracting underlying biological signals that are less dependent on population structure.
Regulatory Frameworks, Ethical Implications, and Professional Governance
The safe, equitable, and widespread adoption of PRS demands clear regulatory oversight and careful management of complex Ethical, Legal, and Social Implications (ELSI).
Regulatory Landscape and Clinical Readiness
The use of PRS algorithms in clinical care is currently subject to rigorous debate as regulatory bodies attempt to establish appropriate validation standards.
FDA Precedent and Utility Thresholds
The FDA’s approval of a genetic risk algorithm for Opioid Use Disorder (OUD), despite comprising only 15 candidate variants, has prompted significant debate regarding the overall readiness of genetic scores for clinical use. A major point of contention is that PRS for highly polygenic conditions, such as Substance Use Disorders (SUDs), typically explain only a small proportion of trait variation (usually $2\%$ to $10\%$), raising questions about whether this translates into clinically significant effects.
However, the regulatory approval for OUD suggests a strategic tension between population-wide efficacy and targeted clinical utility. Given the high mortality and societal cost of OUD, regulators may implicitly accept that even a modest predictive lift for a severe condition is preferable to no genetic intervention. This precedent means that relative risk improvement in targeted high-risk groups may outweigh low population-wide variance explained, provided sufficient evidence of clinical benefit is demonstrated.
Medical Device Classification and Validation Requirements
Regulatory classification dictates the required evidence base. Polygenic score software or products intended for 'human genetic testing' under rules set by bodies like the European Medicines Agency (EMA) and similar guidelines followed by the FDA are likely to fall into a higher-risk category. This classification necessitates stringent evidence requirements, including confirmed scientific validity, analytical performance, and clinical performance.
This classification introduces significant regulatory friction and cost, requiring developers to demonstrate transferability across different healthcare settings and validation in populations whose genomic data were not used in the score's discovery. Validation must be accompanied by ongoing surveillance to capture any unanticipated adverse events following clinical uptake, ensuring the risks are appropriately weighed against potential benefits.
Ethical, Legal, and Social Implications (ELSI)
The use of predictive genetic information carries inherent risks related to privacy, autonomy and justice.
Privacy and Data Protection
In the United States, several federal policies govern the protection of genomic data. Federally funded research adheres to the Common Rule (45 CFR 46), which mandates informed consent. The NIH Genomic Data Sharing Policy and Certificates of Confidentiality further protect research participant data from non-research disclosure. In the clinical sphere, the Health Insurance Portability and Accountability Act (HIPAA) protects clinical genetic test results, while the Genetic Information Nondiscrimination Act (GINA) restricts access to genetic information by health insurers and employers.
However, the protections provided by GINA are limited to specific domains. As highly predictive PRS become commonplace, identifying individuals with $3$- to $5$-fold risk for conditions like CAD, there is a looming conflict concerning third-party underwriters (eg. life, disability, or long-term care insurance). If GINA’s scope is not expanded, the deployment of highly predictive PRS may create a class of "genetically high-risk" individuals who face prohibitive rates for critical financial services, posing a significant threat to health equity and social justice.
Informed Consent and Risk Communication
Providing genetic risk information, particularly for socially sensitive conditions like SUDs or psychiatric disorders, can increase patient distress, anxiety, or depression, alongside the potential for social stigma and discrimination. Therefore, ethical deployment requires careful consideration of interpretation and use. Effective risk communication and genetic counselling are essential to ensure patients and providers understand the context, limitations, and probabilistic nature of PRS results, supporting patient autonomy and minimising psychological risk.
Standardisation and Professional Governance
Clinical adoption is currently hampered by operational inconsistencies and knowledge deficits within the healthcare workforce. A lack of standardised reporting and best practices hinders reproducibility and benchmarking, complicating translation.
There are documented significant knowledge gaps among healthcare providers regarding PRS interpretation, which increases the likelihood of miscommunication or misuse. This requires a dedicated professional response. To accelerate the acceptance of advanced PRS models, the establishment and adherence to robust reporting standards and common model benchmarks on consistent datasets are necessary. Professional governance must prioritise resources for mandatory professional development and integrate specialised genetic counselling support to bridge the knowledge gap, ensuring that clinical interpretations are evidence-based and tailored to patient needs.
The critical implementation gaps that must be addressed are summarised below:
Major Barriers to PRS Clinical Translation and Required Mitigation
Strategic Recommendations for Equitable and Responsible Implementation
To realise the full potential of polygenic scoring while navigating the complex methodological and ethical challenges, a multi-pronged strategy encompassing data, technology and governance is required.
Data Diversification and Collaborative Strategy
The single greatest strategic risk to the success of PRS is the current ancestry bias, which is an ethical and scientific failure simultaneously perpetuating the data gap. A radical commitment to data diversification is mandatory.
Health system leaders and funding agencies (eg. NIH) must strategically prioritise and fund global consortia dedicated to generating and sharing large-scale genomic data from ancestrally diverse populations. This includes leveraging existing high-diversity cohorts, like the All of Us Research Program, to systematically recalibrate and optimise existing PRS. Furthermore, the strategic imperative involves shifting funding models to reward multi-ancestry GWAS and mandating the public dissemination of high-quality, non-European ancestry summary statistics to ensure universal access to improved scores, rather than allowing critical predictive technologies to remain proprietary or population limited.
Informatics and Interoperability Roadmap
Adoption depends on seamless, friction-free integration that transforms raw genomic data into clinical intelligence.
It is recommended that institutions mandate the adoption of FHIR Genomics standards for all data systems handling clinical genomic results. Beyond mere compliance, significant investment is needed in user-centred design research to optimise data visualisations and Clinical Decision Support (CDS) interfaces. These systems must translate complex outputs (eg. PRS percentiles) directly into concrete, actionable treatment or referral pathways, minimizing workflow disruption and maximising provider utility. Moreover, the informatics roadmap must be designed for future multi-modal prediction, ensuring the infrastructure is capable of integrating environmental factors (GxE modelling) and Patient-Reported Outcomes (PROs) alongside genetic data to build more comprehensive risk models.
Clinical Validation and Economic Assessment
The standard for clinical readiness must be predicated on demonstrable patient benefit and economic viability, measured in real-world environments.
Regulatory and Health Technology Assessment (HTA) agencies should mandate the transition from isolated statistical metrics like AUC to clinically relevant utility metrics, specifically Net Reclassification Improvement (NRI) and the Minimum Test Trade off, for all clinical validation trials. Prospective, high-quality pilot studies must be required to rigorously evaluate both the costs and benefits of PRS applications across diverse populations, ensuring that economic analyses fully account for non-trivial implementation costs, provider education and system reconfiguration. Such rigorous, real-world economic modelling is necessary to ensure the positive trend in cost-effectiveness demonstrated in hypothetical studies is realised in practice.
Policy and Regulatory Governance
Clear and adaptive governance is essential to maintain safety, trust, and ethical boundaries during rapid technological acceleration.
Explicit regulatory guidelines from bodies such as the FDA and EMA are needed to clearly define the classification and required evidence base for PRS algorithms, particularly addressing when low variance explained by a score is clinically tolerable based on disease severity and the availability of a highly effective, targeted intervention.
Furthermore, policy efforts must address the critical ELSI gaps, including potentially expanding the scope of protection against genetic discrimination beyond current GINA limits to areas like life and disability insurance. Finally, resources must be prioritised for evidence based patient education and mandatory professional development for all providers who will interpret or act upon PRS results, ensuring ethical communication and adherence to established privacy frameworks. The systematic commitment to standardisation, education and ethical governance is the ultimate guarantor of equitable and successful PRS implementation.
Nelson Advisors > MedTech and HealthTech M&A
Nelson Advisors specialise in mergers, acquisitions and partnerships for Digital Health, HealthTech, Health IT, Consumer HealthTech, Healthcare Cybersecurity, Healthcare AI companies based in the UK, Europe and North America. www.nelsonadvisors.co.uk
Nelson Advisors regularly publish Healthcare Technology thought leadership articles covering market insights, trends, analysis & predictions @ https://www.healthcare.digital
We share our views on the latest Healthcare Technology mergers, acquisitions and partnerships with insights, analysis and predictions in our LinkedIn Newsletter every week, subscribe today! https://lnkd.in/e5hTp_xb
Founders for Founders > We pride ourselves on our DNA as ‘HealthTech entrepreneurs advising HealthTech entrepreneurs.’ Nelson Advisors partner with entrepreneurs, boards and investors to maximise shareholder value and investment returns. www.nelsonadvisors.co.uk
#NelsonAdvisors #HealthTech #DigitalHealth #HealthIT #Cybersecurity #HealthcareAI #ConsumerHealthTech #Mergers #Acquisitions #Partnerships #Growth #Strategy #NHS #UK #Europe #USA #VentureCapital #PrivateEquity #Founders #BuySide #SellSide#Divestitures #Corporate #Portfolio #Optimisation #SeriesA #SeriesB #Founders #SellSide #TechAssets #Fundraising#BuildBuyPartner #GoToMarket #PharmaTech #BioTech #Genomics #MedTech
Nelson Advisors LLP
Hale House, 76-78 Portland Place, Marylebone, London, W1B 1NT
Meet Us @ HealthTech events
Digital Health Rewired > 18-19th March 2025 > Birmingham, UK
NHS ConfedExpo > 11-12th June 2025 > Manchester, UK
HLTH Europe > 16-19th June 2025, Amsterdam, Netherlands
Barclays Health Elevate > 25th June 2025, London, UK
HIMSS AI in Healthcare > 10-11th July 2025, New York, USA
Bits & Pretzels > 29th Sept-1st Oct 2025, Munich, Germany
World Health Summit 2025 > October 12-14th 2025, Berlin, Germany
HealthInvestor Healthcare Summit > October 16th 2025, London, UK
HLTH USA 2025 > October 18th-22nd 2025, Las Vegas, USA
Web Summit 2025 > 10th-13th November 2025, Lisbon, Portugal
MEDICA 2025 > November 11-14th 2025, Düsseldorf, Germany
Venture Capital World Summit > 2nd December 2025, Toronto, Canada
Comments