Background/Purpose: Duchenne muscular dystrophy (DMD) is a severe X-linked neuromuscular disorder with
progressive muscle degeneration. Although its genetic basis is well characterized,
the functional disease progression varies widely between individuals. This project
aims to analyze the relationship between genotype (e.g., mutation type, location,
and affected dystrophin isoforms) and functional decline. Based on these patterns,
an artificial intelligence (AI)-based model will be developed to predict individual
disease trajectories and support personalized care.
Methods: The study follows a two-phase approach. In phase one, pseudonymized clinical and
genetic data from pediatric DMD patients at the Children's Hospital of Eastern Switzerland
(St. Gallen) will be analyzed. Functional outcome measures include the North Star
Ambulatory assessment, 6-minute walk test, performance of the upper limb, and time
to rise. These will be linked to specific genotypic features. Machine learning models
will be developed using Python-based tools such as scikit-learn, XGBoost, and PyTorch.
Both supervised and unsupervised methods will be used to identify genotype-function
associations and to predict individual disease trajectories. In phase two, the model
will be validated using independent data from the Swiss Neuromuscular Disease Registry
(Muskelregister Schweiz).
Results: Data analysis is currently ongoing. The aim is to identify predictive genotype-function
associations and develop a prototype model to forecast functional progression based
on genetic profiles.
Conclusion: This study highlights the potential of combining clinical data and genetic information
to enhance personalized prognosis in DMD. It contributes to data-driven stratification
and exemplifies the clinical relevance of AI-supported modeling in paediatric neurology
and rare disease research.
