Poeiti Abi Saab standing behind large blue and yellow EUPeace letters with her arms outstretched, between two EUPeace banners at the 2025 EUPeace Conference in Pilsen.
Source: Université de Limoges
Description: Poeiti Abi Saab, PhD candidate at the University of Limoges, poses behind large EUPeace letters at the 2025 EUPeace Conference, hosted by the University of West Bohemia in Pilsen, Czech Republic. Her participation in the conference provided opportunities for interdisciplinary exchange on the ethical, social and legal dimensions of artificial intelligence in healthcare.

EUPeace Explains: Can AI help people recover after a stroke?

From personalised exercises at home to questions of fairness and human responsibility, Poeiti Abi Saab’s doctoral research explores how artificial intelligence could support stroke rehabilitation — and why technology must remain at the service of people.

Stroke is one of the leading causes of lasting disability in adults. After a stroke, everyday actions such as lifting an arm, remembering information or concentrating can become difficult, and recovery often depends on regular, repeated exercises over a long period. But what happens when patients return home and can no longer get as much rehabilitation as they need?

This is the question at the heart of Poeiti Abi Saab’s doctoral research at the University of Limoges, France. A state-certified psychomotor therapist and PhD candidate in rehabilitation sciences, she works at the HAVAE laboratory (Handicap, Activity, Ageing, Autonomy, Environment).

Her research asks whether artificial intelligence (AI) can help people continue their rehabilitation at home — while keeping sight of patients’ needs, healthcare professionals’ expertise and the risk of digital exclusion. Her work is part of AISN (Integrating AI in Stroke Neurorehabilitation), a Horizon Europe project investigating digital solutions for post-stroke rehabilitation.

When rehabilitation continues beyond the hospital

A stroke can affect a person’s ability to move, think, remember and manage everyday activities. Rehabilitation helps people regain these abilities through exercises tailored to their needs and repeated over time.

Access to that rehabilitation, though, is often limited — by how many professionals are available, by cost, and by distance. Once patients go home, keeping their recovery going can become hard, and families frequently step in as caregivers.

Could digital technology help bridge this gap?

Poeiti’s research looks at a system that turns repetitive rehabilitation exercises into game-like activities patients can do at home. Using AI, the system adapts the exercises to each individual, encouraging regular practice of movement and thinking skills alongside their usual care. The AISN ecosystem pairs two tools: the Rehabilitation Gaming System (RGS), used by patients, and the Medical Information Management System (MIMS), used by healthcare professionals.

But a promising technology is only part of the answer. It also has to be understandable, accessible and genuinely useful to the people meant to use it.

Contact

EUPeace @ University of Limoges
Laurent Bourdier
A patient performs arm movements using the RGSapp on a smartphone mounted on a rehabilitation device during a clinical assessment at a hospital.
Source: Université de Limoges
Description: A patient undergoes a clinical assessment at a hospital using the Rehabilitation Gaming System (RGS) app. By performing horizontal circular movements with her arm, the patient helps clinicians assess and measure her arm’s range of motion. The digital rehabilitation system is part of research into AI-assisted stroke rehabilitation at the University of Limoges.
Poeiti Abi Saab presenting her research on AI-assisted stroke rehabilitation at the 2025 EUPeace Conference in Pilsen, with a digital platform for clinicians displayed behind her.
Source: Université de Limoges
Description: Poeiti Abi Saab, PhD candidate at the University of Limoges, presents her research on AI-supported home-based stroke rehabilitation at the 2025 EUPeace Conference, hosted by the University of West Bohemia in Pilsen, Czech Republic. The presentation shows the Medical Information Management System (MIMS), a digital platform designed to support healthcare professionals in managing personalised rehabilitation exercises and monitoring patients’ progress.

Beyond technical performance: what makes a technology useful?

An AI-based rehabilitation tool can look promising in theory. In practice, its value depends on whether patients can use it with confidence and whether clinicians see it as a good fit for their work.

So Poeiti puts the experiences of patients, caregivers and clinicians at the centre of her evaluation — asking what they expect, what worries them, and what they need. Their answers reveal what makes a digital tool genuinely usable, and what might stop people reaching for it at all.

That raises a wider question: could a technology meant to improve access to care end up creating new inequalities? People differ in how confident they are with technology and in whether they have reliable devices or internet. Post-stroke older adults, people with limited digital skills and those in isolated areas can face extra barriers. If that is overlooked, a tool designed to widen access could quietly leave some people further behind. For Poeiti, judging usability is therefore also about making sure the benefits of innovation are shared fairly.

A question from Pilsen: Can we trust AI without depending on it?

In September 2025, Poeiti took part in the EUPeace Conference in Pilsen, Czech Republic, hosted by the University of West Bohemia. Supported by the EUPeace initiative, the research stay let her present her work on AI-assisted home rehabilitation and exchange ideas with researchers who approach artificial intelligence from ethical, social and legal angles.

One conversation stayed with her: could an AI recommendation system make healthcare professionals so reliant on its suggestions that they stop thinking critically about their patients?

For Poeiti, the answer is to treat AI as a complement to human expertise, not a replacement. Digital tools may help with real gaps in today’s rehabilitation — such as the difficulty of keeping up intensive practice at home — but clinical decisions must stay with the professionals.

“AI should never make the final decision.”

If a system’s recommendation conflicts with a clinician’s judgement, it is the clinician who remains responsible for the patient’s care. That view echoes what clinicians in her own research told her: they saw AI as support for their expertise, not a substitute for it.

What the research shows so far

So far, Poeiti has asked stroke survivors, their carers and clinicians what they expect from such a system and how easy they find it to use. She has looked closely at how that ease of use can even be measured reliably — work she has published, as first author, in international journals. And she has run a first trial of the technology in people’s homes: the early signs are encouraging, with small improvements in arm movement and in thinking skills — though she is careful to stress that these results are preliminary, and not yet proof that the system works.

The next steps are to publish the full trial results and to keep exploring how digital rehabilitation can fit into everyday care. Poeiti will defend her doctorate in December 2026, and hopes in the longer term to develop a “hybrid” model that combines human support with digital tools. Her work has already been recognised with several awards and invitations to present abroad.

A patient performs arm movements during a clinical assessment using the RGSclinic application, with a rehabilitation game displayed on a screen in a hospital treatment room.
Source: Université de Limoges
Description: A patient undergoes a clinical assessment at a hospital using the Rehabilitation Gaming System (RGSclinic). By performing horizontal circular movements with their arm, the patient helps clinicians observe and measure the arm joint’s range of motion. The assessment forms part of research into AI-assisted stroke rehabilitation at the University of Limoges.

Rehabilitation as a question of social justice

For Poeiti, what is at stake reaches well beyond a single new healthcare technology.

“For me, rehabilitation is fundamentally a question of social justice.”

The chance to recover after a stroke, she argues, should not depend on where someone lives, how much money they have, or how comfortable they feel with technology. AI-assisted rehabilitation could help bring exercises into more people’s homes — but only if these tools are designed and tested with the people who will actually use them.

That is where her research meets bigger questions of inclusion and societal change: technology can help people regain their independence, but only when it strengthens human care and takes account of the inequalities that shape people’s lives. Through her doctoral work and her international collaborations, Poeiti Abi Saab is helping to shape a conversation that goes beyond AI itself — not just about building smarter technology, but about making sure progress in rehabilitation reaches people fairly.

Similar news

Three smiling women stand side by side in front of a EUPeace roll-up banner in a wood-panelled hall. The banner shows the EUPeace and European Union logos, a photograph of a building with flags, the website www.eupeace.eu and the logos of several partner universities. Two of the women wear EUPeace lanyards with name badges.
Source: Hannes Kaufmann
Description: Three women pose for a photograph beside a EUPeace roll-up banner in an indoor space with wooden wall panels. The banner carries the EUPeace logo, the text 'Co-funded by the European Union', an image of a building with flags, the address www.eupeace.eu and the logos of partner universities, including Universität Marburg, Justus-Liebig-Universität Giessen, Università della Calabria and Universidad Pontificia Comillas. The women wear EUPeace lanyards with name badges, and the occasion appears to be an EUPeace event.
A crowded climate demonstration with protesters holding hand-painted placards displaying messages in different languages, including “DON’T BURN OUR FUTURE” and “UNSERE ZUKUNFT”.
Source: Dominic Wunderlich via pixabay.com
Description: A climate demonstration with protesters holding handmade placards carrying messages about climate change in different languages, including English and German. The image provides a visual context for the article’s focus on climate communication and the ways in which language can shape public discussions about climate change, health and social equity.
A human hand in a grey sleeve shakes a glowing, turquoise wireframe hand that extends from the screen of an open laptop. The laptop stands on a wooden desk next to a blank sheet of paper, with a blurred blackboard in the background.
Source: Kiquebg via pixabay.com
Description: A handshake between a human hand and a digital hand emerging from a laptop screen illustrates how algorithms and digital systems are becoming part of migrant integration in Europe. The image accompanies the first part of a two-part series on digital migration, based on a contribution by Claudio Di Maio from the University of Calabria.
Open silver combination padlock standing on a white computer keyboard.
Source: Sasun Bughdaryan via unsplash.com
Description: Close-up photo of a silver combination padlock standing upright on a white computer keyboard. The shackle is open and the dials show 8-8-8, with a small set of keys lying next to the lock. The background fades from dark brown to warm amber light. The image suggests cybersecurity, data protection or an unsecured system.

Area

Group

Initiative

University

Activity