How is artificial intelligence being used in rare
neuromuscular diseases? A webinar organised by the
Horizon Europe-funded DREAMS project
examined applications in genetic diagnosis, drug
discovery and real-world healthcare data, with examples
from the specialists developing these approaches.
Artificial Intelligence for Rare Neuromuscular
Diseases: From Early Diagnosis to Drug Discovery
and Clinical Translation.
-
Dr Gohun Seo
Chief Medical Officer
3billion
-
Rubal Ravinder
Machine Learning Engineer
Kantify
-
Christopher Rudolf
CEO
Volv Global
Moderated by François Lamy,
Vice-President of AFM-Téléthon.
01
Genetic diagnosis / 3billion
When a negative genetic test is not the end of the story
Dr Gohun Seo began with a clinical case: a 46-year-old
woman who had experienced progressive muscle weakness
since adolescence. A targeted muscular dystrophy gene
panel had returned a negative result. Further
investigation using exome sequencing identified a
genetic alteration associated with spinal muscular
atrophy (SMA). The relevant gene had not been included
in the original panel.
The case illustrated a limitation of targeted testing:
the genes selected for analysis may not include the
one responsible for a patient’s symptoms.
Seo also presented findings from EasyNMD,
a screening initiative developed by 3billion in
collaboration with Biogen. Of 232 patients evaluated,
51 received a genetic diagnosis. Only two of those
diagnoses were SMA, the condition the programme had
initially set out to identify.
Another part of Seo’s presentation addressed
automated reanalysis: revisiting
unresolved genetic cases as new gene–disease
associations and scientific evidence become available.
Earlier inconclusive results may then become
interpretable, without necessarily requiring new
sequencing.
02
Drug discovery / Kantify
Which compounds are worth taking into the laboratory?
Rubal Ravinder explained how Kantify uses machine
learning to investigate interactions between proteins
and small molecules. These predictions can help
researchers identify potential biological targets
and decide which compounds merit experimental
investigation.
She described an earlier collaboration with I-Stem
on limb-girdle muscular dystrophy R2, combining
laboratory screening and AI-based analysis to
investigate the proteins that might explain observed
cellular responses. This was a separate research
example, not a result from DREAMS.
-
STEP 01
AI-supported compound prioritisation
-
STEP 02
Literature review and clinical input
-
STEP 03
Experimental screening
Ravinder also shared an
early-stage example from DREAMS.
An AI-supported assessment of approximately 20,000
compounds was followed by literature review and
clinical input to select seven for experimental
screening. Four were reported as primary hits in
an initial assay.
The example connects computational prioritisation
with experimental biology: AI helps identify compounds
to investigate, while laboratory work tests whether
the predicted activity is observed.
03
Real-world data / Volv Global
Finding patients within existing healthcare data
Christopher Rudolf brought a third perspective:
the use of population-scale real-world healthcare
data and machine learning to investigate gaps in
rare disease diagnosis and treatment pathways.
This approach examines patterns in existing healthcare
information to help identify people who may warrant
further clinical assessment. Its scope also includes
diagnosed patients who may be eligible for treatments
they are not currently receiving.
The question is different from the one facing a drug
discovery team: where might information already
collected during healthcare point to an overlooked
diagnosis or treatment opportunity?
The research behind DREAMS
Investigating shared mechanisms across rare
neuromuscular disorders
DREAMS combines patient-derived induced pluripotent
stem cell models, experimental drug screening and
AI-based methods to investigate shared biological
mechanisms across five rare neuromuscular disorders:
Duchenne muscular dystrophy, DNM2-related centronuclear
myopathy, LMNA-related Emery-Dreifuss muscular dystrophy,
Pompe disease and Danon disease.
The project aims to generate evidence for future
therapeutic development, including the investigation
of existing medicines for potential new uses.
Explore the DREAMS research approach
→
DREAMS is funded by the European Union under Horizon
Europe. Grant Agreement No. 101080229.
