Authors
Mathis Batanga Loko, Nicolas Panayotis, Jean-Christophe Roux, Frédéric Alexandre, Sandrine Parrot
Published in
Journal of neuroscience methods. Pages 110949. Nov 10, 2026. Epub Nov 10, 2026.
Abstract
Rett syndrome is a genetic neurodevelopmental disorder caused by mutations of the MECP2 gene. It is characterized by severe neurological, cognitive, autonomic, and motor dysfunctions. Central catecholaminergic deficits, together with alterations in GABAergic and glutamatergic systems, have previously been identified in the brain of Mecp2 knockout mice using conventional data analyses. Here, we reanalyzed neurochemical data from this well-established mouse model using machine-learning approaches to identify new neurochemical information.
We leveraged a neurochemical database featuring tissue concentrations of dopamine, noradrenaline, serotonin, GABA and glutamate from micro-dissected brain areas in both Mecp2-knockout and wild-type mice. We then extracted qualitative and quantitative data at the early-onset and fully symptomatic stages of the disease. Five machine learning models were trained to predict genotype, and their predictive performance was evaluated.
Despite the small number of individuals per group, the best model was obtained with XGBoost fitting. SHAP-derived variables of importance analysis identified neurotransmitters and brain regions that were consistent with neurochemical alterations previously implicated in the pathophysiology of Rett syndrome.
Our results indicate that machine learning, which remains largely underused in the field of neurochemistry, can effectively predict group differences from brain neurochemical profiles in a well-established mouse model for Rett syndrome.
The satisfying performances highlight the potential of machine learning and modern computational tools to extract additional information from existing neurochemical datasets, complement conventional analyses, and for the redesign of future neurochemical studies.
PMID:
43008899
Bibliographic data and abstract were imported from PubMed on 11 Nov 2026.
Read full publication at:
Please sign in
to see all details.
Advertisement
Stats
- Recommendations n/a n/a positive of 0 vote(s)
- Views 16
- Comments 0