10 selected Research Luxembourg results – April 2026

18 May 2026

This article was originally in French and German on science.lu 

University of Luxembourg (FSTM)

Extreme weather events are increasing across Europe. The University of Luxembourg, together with RSS-Hydro, has developed an innovative tool to better predict heavy rainfall and flood risks. The NWPLux project aims to create the first weather forecasting model specifically tailored to Luxembourg and the Greater Region.

Unlike systems relying on international data, NWPLux provides highly detailed local forecasts. Updated every six hours and continuously fed with satellite, radar, and local data, the model significantly improves rainfall detection. Tests on past floods showed an 8.3% increase in detection accuracy and a 13.7% reduction in false alarms.

Its main innovation is the coupling of weather forecasting with flood simulation, allowing detailed mapping of at-risk areas and helping authorities better anticipate and reduce damage.
Press Release

Luxembourg Centre for Systems Biomedicine (LCSB)

Two recent studies show how artificial intelligence can accelerate the discovery of treatments for Leigh syndrome, a rare and severe neurodegenerative disease mainly affecting children. The condition causes motor impairment, intellectual disability, and reduced life expectancy, with very few treatment options currently available.

Faced with a lack of reliable experimental models, an international team combined computational (in silico), experimental (in vitro), and animal (in vivo) approaches to identify new therapeutic leads. Researchers analyzed more than 5,500 drugs already used for other diseases and identified several as promising candidates for Leigh syndrome. Using computational modeling, they also clarified mechanisms of action in lab-grown organoids that mimic human tissues.

These findings highlight the potential of AI to speed up biomedical research while reducing both cost and development time.
Press Release

Wearable sensors to better monitor invisible symptoms in Parkinson’s disease

Luxembourg Centre for Systems Biomedicine (LCSB)

Parkinson’s disease is not limited to tremors: so-called “non-motor” symptoms, such as orthostatic hypotension, can significantly affect daily life. This drop in blood pressure when standing causes dizziness and fainting spells, which are difficult to detect during standard medical visits because they often occur at home.

To address this challenge, the PDHOME project—led in part by the Luxembourg Centre for Systems Biomedicine—is exploring the use of wearable sensors and artificial intelligence. Patients are equipped with devices that continuously measure movement and cardiovascular activity, both in clinical settings and at home, allowing symptoms to be tracked over several weeks.

In the long term, this approach could improve personalized patient monitoring, help adjust treatments more precisely, and enable earlier detection of complications, while also deepening understanding of these “invisible” symptoms.
Press Release

Thinking to communicate: AI makes brain–computer interfaces more efficient

University of Luxembourg (FSTM)

Can we communicate through thought? A new study from fesearchers at the University of Luxembourg are helping improve brain–computer interfaces, which translate brain activity into commands and could support people who cannot speak, including those with locked-in syndrome (a condition where a person is conscious but almost completely paralysed, typically able to communicate only through eye movements).

Traditional “silent speech” systems require long, exhausting training based on repeatedly imagining words. The researchers reduced this burden by training an AI model on brain signals recorded while participants spoke aloud, then applying it to imagined speech. Using EEG data and advanced AI, the system reaches about 86% accuracy for real speech and nearly 80% for imagined speech—an unprecedented result.

This method reduces training time and makes brain–computer communication more accessible, offering new possibilities for patients with severe paralysis.
Press Release

Discover more AI news in Luxembourg:

Public health: better measuring exposure to chemical substances to prevent diseases

Luxembourg Institute of Health (LIH)

Researchers at the Luxembourg Institute of Health are calling for a rethink in how exposure to chemical substances is studied. Despite its major impact—air and environmental pollution being linked to millions of premature deaths—most environmental chemicals are not systematically measured in humans.

The study highlights the concept of the chemical exposome, which includes all substances a person is exposed to and how they are transformed in the body. This remains poorly understood, especially the combined effects of multiple chemicals. To address this gap, researchers contribute to the Human Internal Chemical Exposome Atlas initiative, which uses advanced tools such as mass spectrometry and artificial intelligence to map human exposure. The project has already identified hundreds of chemical compounds in the human body.

This approach is crucial for improving risk assessment, strengthening public health policy, and better preventing environmentally related diseases.
Press Release

No reliable model can predict survival in elderly cancer patients

Luxembourg Institute of Health (LIH)

A large study led by the Luxembourg Institute of Health and experts from the International Society of Geriatric Oncology reveals a major gap in the care of older cancer patients.

The researchers analyzed 250 models designed to predict survival and mortality in people aged 65 and over. The results are concerning: none of these models are considered reliable for clinical use. More than half fail to include key factors such as comorbidities, nutritional status, or cognitive function—crucial elements in older patients. The study also highlights widespread methodological issues, including high risk of bias, poor variable selection, and insufficient validation.

The authors call for more robust models that incorporate geriatric data and adhere to stronger scientific standards.
Press Release

Discover more Luxembourg’s cancer research

Asylum in Europe: 20 countries apply accelerated procedures for applicants from countries deemed safe.

Université du Luxembourg (FHSE)

The European Migration Network has published an analysis on the concepts of “safe countries of origin” and “safe third countries” in preparation for the implementation of the Asylum Procedure Regulation (EU) 2024/1348, scheduled for June 2026. Based on contributions from 26 Member States (including Luxembourg) and Serbia, the report highlights a wide variety of national approaches.

Nineteen countries use national lists of safe countries of origin, while others, such as Finland and Portugal, assess cases individually. Twenty countries apply accelerated procedures for applicants from countries deemed safe, with shortened appeal deadlines. However, exceptions exist for unaccompanied minors and vulnerable individuals, and some states add extra criteria, particularly regarding human rights compliance.

The report also highlights good practices, such as training asylum officers, to ensure procedures are both fast and fair.
Press Release

Some cancer cells hijack healthy cells to feed themselves

Luxembourg Institute of Health (LIH)

A study by LIH reveals a previously unknown mechanism through which metastatic breast cancer cells exploit lung cells to promote their growth.

Researchers found that these cancer cells reprogram specific lung cells to increase the production of lipids (fats) essential for their survival. This hijacking of the tumour microenvironment supports tumour proliferation, and reducing lipid production could significantly slow disease progression.

The findings show that metastases depend not only on cancer cells themselves, but also on the healthy host cells they manipulate, suggesting new therapeutic strategies that target the tumour microenvironment rather than the tumour alone.
Press Release

University of Luxembourg (FSTM)

A recent study reveals a surprising phenomenon called “anti-dephasing,” which challenges the traditional view of “noise” in quantum computing. Here, noise refers to random disturbances—external influences or internal imperfections—that affect quantum systems such as qubits.

Usually seen as a major obstacle, this noise leads to quantum decoherence, the loss of quantum properties. However, researchers found that in some cases it can instead help preserve or even stabilize the system.

This discovery could improve understanding of how quantum states behave under disturbance and help design more robust qubits, accelerating the development of future quantum computers.
Press Release

Author: Diane Bertel
Editor: Lucie Zeches (FNR) ; Nicolas Stamets (Research Luxembourg)

Similar articles