Hi there! 👋🏼

I love turning data into stories that matter, especially in fields where sensitive information needs thoughtful handling and creative solutions. I’m driven by the challenge of making data useful while working through the ethical, legal, and practical complexities that come with it.

Currently, I am a Postdoctoral Researcher affiliated with both KU Leuven and Hasselt University in Belgium 🇧🇪, where I continue to work on federated learning and real-world data research in healthcare. Previously, I completed a double PhD (one from each university) at these institutions, working on making real-world data more accessible and useful for getting meaningful insights.

The Challenge 🧩

But let me tell you, working with data isn’t as straightforward as it sounds. It often feels like trying to solve a jigsaw puzzle, but this isn’t your typical puzzle.

Imagine the pieces are scattered across different locations, some of them locked away in drawers labeled “private”, and others so rare they might as well be one-of-a-kind. Now, add a rule: you’re not allowed to move any pieces from their original location.

That’s what working with data in fields like rare diseases or low-prevalence conditions feels like. Every single piece matters, but putting them together is far from simple.

You'd think the solution would be straightforward: gather all the data into one place and get to work.

But in reality, centralizing data is often out of the question. Privacy concerns, ethical dilemmas, data ownership rules, and layers of administrative hurdles stand in the way. It’s like trying to build the puzzle while someone keeps reminding you, “You can’t touch the pieces.”

The Solution: Federated Learning 🚀

So, how do we solve this? That’s where Federated Learning (FL) comes in.

Think of FL as a way to build the puzzle together, but with a twist: everyone works on their pieces separately, sharing just enough hints to complete the bigger picture. With FL, machine learning models can be trained on `decentralized data` without ever moving the data itself. `Collaborating` while keeping what's personal safe and secure. This approach doesn't just safeguard privacy; it opens doors. Organizations can work together without worrying about sensitive information leaving their walls, unlocking insights that were once `impossible to achieve`. My research focused on designing and implementing FL techniques to tackle exactly these kinds of challenges, bringing together `innovation` and `practicality` to make this idea work in practice.

Why I Do What I Do 👨🏻‍💻

The world is full of complex problems, and I believe the best solutions come from collaboration. But collaboration isn’t always easy, especially when sensitive data is involved.

That’s why I love combining technical innovation with practical problem-solving to create tools that make working together possible, even when the odds are stacked against us. Whether it’s improving global data-sharing pipelines or designing tools for personalized predictions, I focus on creating solutions that are inclusive, accessible, and meaningful.

What I’m Working On Now 🔬

Right now, I'm leading the development of **FLkit**, a toolkit that helps researchers and practitioners get started with `federated analytics` and learning. I'm also continuing to explore `personalized federated learning` approaches, especially for healthcare applications where `privacy` really matters.

Highlights of My Work

Here are some highlights of my work. Check out more details on the projects page.

Research Publications

📄

Personalized Federated Learning for MS

Published in npj Digital Medicine on `personalized federated learning` for predicting disability progression in MS. First federated learning study in multiple sclerosis using `routine clinical data` with 26,000+ patients across 146 centers in 32 countries.

Achieved ROC-AUC scores around 0.84 while `preserving privacy`.

🌐

Global Data Sharing Initiative

Built a `federated data analysis pipeline` for COVID-19 and MS research. This work helped create one of the largest datasets for these conditions, enabling `global collaboration` while respecting privacy constraints. Published in JMIR Medical Informatics.

Contributed to multiple publications and `global treatment guidelines`.

Tools & Frameworks

🛠️

FLkit

Leading the development of the Federated Learning toolkit for Health & Life Sciences, a `structured entry point` into federated analytics and learning with 29+ contributors.

🔗

FL4E Framework

Created FL4E (Federated Learning For Everyone), a framework that `bridges centralized and decentralized` data analysis for `real-world healthcare` applications. Published in JMIR Formative Research.

Recognition & Awards

🏆

ECTRIMS 2025 Grant

Received a personal grant for scientific quality for our work on "LoRank: Adaptive Structured Personalization in Federated Learning for Predicting Disability Progression in Multiple Sclerosis". Published in Multiple Sclerosis Journal.

Key Contributions

⚖️

Degree of Federation

Introduced the concept of `"degree of federation"`, which gives organizations a `flexible way` to balance privacy, practicality, and collaboration based on their specific needs.

🤖

Personalized FL Approaches

Developed `personalized federated learning` approaches that help machine learning models `adapt to local datasets`, improving predictions for disability progression in MS while `keeping data private`.

Let’s Connect! 🌟

I’m always excited to collaborate with people who share my passion for turning data into insights that make a difference.

If you’re working on something where real-world data and AI or machine learning could play a role, whether it’s in healthcare, finance, or beyond, I’d love to chat and explore ways we can team up to create meaningful impact.

Feel free to reach out through LinkedIn or check out my projects and publications to learn more about what I’m working on.

news

Dec 18, 2025	KU Leuven Department of Electrical Engineering When: December 18th, 2025 Where: KU Leuven, Belgium Title: “FedMinibatch: Federated Synchronous SGD and Its Equivalence to Centralized Training” Focus: Presented federated learning concepts, architectures, and real-world applications in healthcare, connecting theoretical foundations with practical implementations in MS research.
Dec 15, 2025	Medical Informatics Europe 2026 - Two Papers Submitted Submission Date: December 2025 Conference: MIE 2026 (Medical Informatics Europe) Conference Date: May 2026 Location: Genoa, Italy Papers Submitted: “Good for All, Not Good Enough for One: Reuse Dilemma in Federated Learning” Explores the challenges and opportunities in reusing federated learning models across different contexts and institutions. “Client Participation per Round in Federated Learning for Multiple Sclerosis with Real-World Data” Investigates optimal client participation strategies in federated learning workflows for MS research using real-world clinical data. Both papers address important questions in federated learning for healthcare applications, building on our ongoing research in privacy-preserving machine learning for medical data.
Dec 03, 2025	Flower Monthly Meeting When: December 3rd, 2024 Format: Virtual Meeting Title: “Transforming Multiple Sclerosis Research: Advancing Disability Progression Insights through Practical and Precise Federated Learning using Real-World Data” Focus: Shared insights on personalized federated learning approaches for healthcare applications, drawing from our recent work on MS disability progression prediction and the FLkit project.
Nov 05, 2025	Roche Federated Learning Team for Multiple Sclerosis When: November 5th, 2025 Where: Roche (Virtual/In-Person) What: Presentation to the Federated Learning Team working on Multiple Sclerosis Focus: Shared research findings and practical insights on the latest federated learning approaches for MS research, including our work on personalized federated learning and the FLkit framework.
Nov 04, 2025	North American Society of AI in MS When: November 4th, 2025 Where: Virtual/In-Person Title: “Personalized federated learning for predicting disability progression in multiple sclerosis using real-world routine clinical data” Focus: Discussed the potential of federated learning to advance MS research while preserving patient privacy, highlighting our work with the MSBase registry and personalized federated learning approaches.

selected publications

Federated Learning

Accessible ecosystem for clinical research (federated learning for everyone): development and usability study

Ashkan Pirmani, Martijn Oldenhof, Liesbet M Peeters, and 2 more authors

JMIR Formative Research, 2024

Abs DOI

Background: The integrity and reliability of clinical research outcomes rely heavily on access to vast amounts of data. However, the fragmented distribution of these data across multiple institutions, along with ethical and regulatory barriers, presents significant challenges to accessing relevant data. While federated learning offers a promising solution to leverage insights from fragmented data sets, its adoption faces hurdles due to implementation complexities, scalability issues, and inclusivity challenges. Objective: This paper introduces Federated Learning for Everyone (FL4E), an accessible framework facilitating multistakeholder collaboration in clinical research. It focuses on simplifying federated learning through an innovative ecosystem-based approach. Methods: The “degree of federation” is a fundamental concept of FL4E, allowing for flexible integration of federated and centralized learning models. This feature provides a customizable solution by enabling users to choose the level of data decentralization based on specific health care settings or project needs, making federated learning more adaptable and efficient. By using an ecosystem-based collaborative learning strategy, FL4E encourages a comprehensive platform for managing real-world data, enhancing collaboration and knowledge sharing among its stakeholders. Results: Evaluating FL4E’s effectiveness using real-world health care data sets has highlighted its ecosystem-oriented and inclusive design. By applying hybrid models to 2 distinct analytical tasks—classification and survival analysis—within real-world settings, we have effectively measured the “degree of federation” across various contexts. These evaluations show that FL4E’s hybrid models not only match the performance of fully federated models but also avoid the substantial overhead usually linked with these models. Achieving this balance greatly enhances collaborative initiatives and broadens the scope of analytical possibilities within the ecosystem. Conclusions: FL4E represents a significant step forward in collaborative clinical research by merging the benefits of centralized and federated learning. Its modular ecosystem-based design and the “degree of federation” feature make it an inclusive, customizable framework suitable for a wide array of clinical research scenarios, promising to revolutionize the field through improved collaboration and data use. Detailed implementation and analyses are available on the associated GitHub repository.
GDSI

The Journey of Data Within a Global Data Sharing Initiative: A Federated 3-Layer Data Analysis Pipeline to Scale Up Multiple Sclerosis Research

Ashkan Pirmani, Edward De Brouwer, Lotte Geys, and 4 more authors

JMIR Medical Informatics, 2023

Abs DOI

Background: Investigating low-prevalence diseases such as multiple sclerosis is challenging because of the rather small number of individuals affected by this disease and the scattering of real-world data across numerous data sources. These obstacles impair data integration, standardization, and analysis, which negatively impact the generation of significant meaningful clinical evidence. Objective: This study aims to present a comprehensive, research question–agnostic, multistakeholder-driven end-to-end data analysis pipeline that accommodates 3 prevalent data-sharing streams: individual data sharing, core data set sharing, and federated model sharing. Methods: A demand-driven methodology is employed for standardization, followed by 3 streams of data acquisition, a data quality enhancement process, a data integration procedure, and a concluding analysis stage to fulfill real-world data-sharing requirements. This pipeline’s effectiveness was demonstrated through its successful implementation in the COVID-19 and multiple sclerosis global data sharing initiative. Results: The global data sharing initiative yielded multiple scientific publications and provided extensive worldwide guidance for the community with multiple sclerosis. The pipeline facilitated gathering pertinent data from various sources, accommodating distinct sharing streams and assimilating them into a unified data set for subsequent statistical analysis or secure data examination. This pipeline contributed to the assembly of the largest data set of people with multiple sclerosis infected with COVID-19. Conclusions: The proposed data analysis pipeline exemplifies the potential of global stakeholder collaboration and underlines the significance of evidence-based decision-making. It serves as a paradigm for how data sharing initiatives can propel advancements in health care, emphasizing its adaptability and capacity to address diverse research inquiries.
GDSI

Updated results of the COVID-19 in MS global data sharing initiative: anti-CD20 and other risk factors associated with COVID-19 severity

Steve Simpson-Yap, Ashkan Pirmani, Tomas Kalincik, and 8 more authors

Neurology: Neuroimmunology & Neuroinflammation, 2022

Abs DOI

Background and Objectives: Certain demographic and clinical characteristics, including the use of some disease-modifying therapies (DMTs), are associated with severe acute respiratory syndrome coronavirus 2 infection severity in people with multiple sclerosis (MS). Comprehensive exploration of these relationships in large international samples is needed. Methods: Clinician-reported demographic/clinical data from 27 countries were aggregated into a data set of 5,648 patients with suspected/confirmed coronavirus disease 2019 (COVID-19). COVID-19 severity outcomes (hospitalization, admission to intensive care unit [ICU], requiring artificial ventilation, and death) were assessed using multilevel mixed-effects ordered probit and logistic regression, adjusted for age, sex, disability, and MS phenotype. DMTs were individually compared with glatiramer acetate, and anti-CD20 DMTs with pooled other DMTs and with natalizumab. Results: Of 5,648 patients, 922 (16.6%) with suspected and 4,646 (83.4%) with confirmed COVID-19 were included. Male sex, older age, progressive MS, and higher disability were associated with more severe COVID-19. Compared with glatiramer acetate, ocrelizumab and rituximab were associated with higher probabilities of hospitalization (4% [95% CI 1–7] and 7% [95% CI 4–11]), ICU/artificial ventilation (2% [95% CI 0–4] and 4% [95% CI 2–6]), and death (1% [95% CI 0–2] and 2% [95% CI 1–4]) (predicted marginal effects). Untreated patients had 5% (95% CI 2–8), 3% (95% CI 1–5), and 1% (95% CI 0–3) higher probabilities of the 3 respective levels of COVID-19 severity than glatiramer acetate. Compared with pooled other DMTs and with natalizumab, the associations of ocrelizumab and rituximab with COVID-19 severity were also more pronounced. All associations persisted/enhanced on restriction to confirmed COVID-19. Discussion: Analyzing the largest international real-world data set of people with MS with suspected/confirmed COVID-19 confirms that the use of anti-CD20 medication (both ocrelizumab and rituximab), as well as male sex, older age, progressive MS, and higher disability are associated with more severe course of COVID-19.
GDSI

Associations of disease-modifying therapies with COVID-19 severity in multiple sclerosis

Steve Simpson-Yap, Edward De Brouwer, Tomas Kalincik, and 9 more authors

Neurology, 2021

Abs DOI

Background and Objectives: People with multiple sclerosis (MS) are a vulnerable group for severe coronavirus disease 2019 (COVID-19), particularly those taking immunosuppressive disease-modifying therapies (DMTs). We examined the characteristics of COVID-19 severity in an international sample of people with MS. Methods: Data from 12 data sources in 28 countries were aggregated (sources could include patients from 1–12 countries). Demographic (age, sex), clinical (MS phenotype, disability), and DMT (untreated, alemtuzumab, cladribine, dimethyl fumarate, glatiramer acetate, interferon, natalizumab, ocrelizumab, rituximab, siponimod, other DMTs) covariates were queried, along with COVID-19 severity outcomes, hospitalization, intensive care unit (ICU) admission, need for artificial ventilation, and death. Characteristics of outcomes were assessed in patients with suspected/confirmed COVID-19 using multilevel mixed-effects logistic regression adjusted for age, sex, MS phenotype, and Expanded Disability Status Scale (EDSS) score. Results: Six hundred fifty-seven (28.1%) with suspected and 1,683 (61.9%) with confirmed COVID-19 were analyzed. Among suspected plus confirmed and confirmed-only COVID-19, 20.9% and 26.9% were hospitalized, 5.4% and 7.2% were admitted to ICU, 4.1% and 5.4% required artificial ventilation, and 3.2% and 3.9% died. Older age, progressive MS phenotype, and higher disability were associated with worse COVID-19 outcomes. Compared to dimethyl fumarate, ocrelizumab and rituximab were associated with hospitalization (adjusted odds ratio [aOR] 1.56, 95% confidence interval [CI] 1.01–2.41; aOR 2.43, 95% CI 1.48–4.02) and ICU admission (aOR 2.30, 95% CI 0.98–5.39; aOR 3.93, 95% CI 1.56–9.89), although only rituximab was associated with higher risk of artificial ventilation (aOR 4.00, 95% CI 1.54–10.39). Compared to pooled other DMTs, ocrelizumab and rituximab were associated with hospitalization (aOR 1.75, 95% CI 1.29–2.38; aOR 2.76, 95% CI 1.87–4.07) and ICU admission (aOR 2.55, 95% CI 1.49–4.36; aOR 4.32, 95% CI 2.27–8.23), but only rituximab was associated with artificial ventilation (aOR 6.15, 95% CI 3.09–12.27). Compared to natalizumab, ocrelizumab and rituximab were associated with hospitalization (aOR 1.86, 95% CI 1.13–3.07; aOR 2.88, 95% CI 1.68–4.92) and ICU admission (aOR 2.13, 95% CI 0.85–5.35; aOR 3.23, 95% CI 1.17–8.91), but only rituximab was associated with ventilation (aOR 5.52, 95% CI 1.71–17.84). Associations persisted on restriction to confirmed COVID-19 cases. No associations were observed between DMTs and death. Stratification by age, MS phenotype, and EDSS score found no indications that DMT associations with COVID-19 severity reflected differential DMT allocation by underlying COVID-19 severity. Discussion: Using the largest cohort of people with MS and COVID-19 available, we demonstrated consistent associations of rituximab with increased risk of hospitalization, ICU admission, and need for artificial ventilation and of ocrelizumab with hospitalization and ICU admission. Despite the cross-sectional design of the study, the internal and external consistency of these results with prior studies suggests that rituximab/ocrelizumab use may be a risk factor for more severe COVID-19.
npj Digit. Med.

Personalized federated learning for predicting disability progression in multiple sclerosis using real-world routine clinical data

Ashkan Pirmani, Edward De Brouwer, Ádám Arany, and 70 more authors

npj Digital Medicine, 2025

Abs DOI

Early prediction of disability progression in multiple sclerosis (MS) remains challenging despite its critical importance for therapeutic decision-making. We present the first systematic evaluation of personalized federated learning (PFL) for 2-year MS disability progression prediction, leveraging multi-center real-world data from over 26,000 patients. While conventional federated learning (FL) enables privacy-aware collaborative modeling, it remains vulnerable to institutional data heterogeneity. PFL overcomes this challenge by adapting shared models to local data distributions without compromising privacy. We evaluated two personalization strategies: a novel AdaptiveDualBranchNet architecture with selective parameter sharing, and personalized fine-tuning of global models, benchmarked against centralized and client-specific approaches. Baseline FL underperformed relative to personalized methods, whereas personalization significantly improved performance, with personalized FedProx and FedAVG achieving ROC-AUC scores of 0.8398 ± 0.0019 and 0.8384 ± 0.0014, respectively. These findings establish personalization as critical for scalable, privacy-aware clinical prediction models and highlight its potential to inform earlier intervention strategies in MS and beyond.