Home-cage AI monitoring finds social and nest-building deficits in SPTAN1 R1098Q mice, with wild-type co-housing partially reducing hyperactivity and restoring rest patterns

The work introduces and runs a non-invasive, high-content home-cage phenotyping framework to examine social and exploratory dysfunction in SPTAN1 R1098Q mice. Relative to clinical and preclinical phenotyping dominated by motor and memory impairments, this framework brings social and exploratory behaviors into the standard phenotypic scope. The abstract states the pipeline integrates three open-source AI engines: SLEAP.ai for markerless multi-animal pose estimation, DeepOF for automated classification of autonomous and social behaviors, and SAM2 for segmentation and quantification of nest-building material usage; evidence comes from long-term monitoring, but the loaded text is an incomplete abstract without sample sizes, statistics, or effect sizes.

Long-term monitoring revealed that SPTAN1 R1098Q mutants show impaired spatial habituation in mutant-mutant pairs and display persistent, unreciprocated trailing (following) of cage mates. These social and exploratory alterations go beyond what conventional motor and memory phenotyping covers. Based on the abstract's wording 'Long-term monitoring revealed' and 'persistent, unreciprocated trailing (following) of cage mates'; the number of pairs, monitoring duration, and statistical tests are not given in the loaded text.

Mutants also showed a pronounced reduction in nest-building behavior, indicating that their behavioral abnormalities are not confined to motor or memory domains. Nest-building quantification is achieved through SAM2 segmentation and counting of nest-building material usage, bringing a behavior not covered by conventional phenotyping into automated measurement. The abstract states 'a pronounced reduction of nest-building behavior' but reports no magnitude of reduction, sample size, or control setup.

Co-housing with a wild-type littermate (MUT-WT) attenuated mutant hyperactivity and restored a more typical rest pattern, suggesting social context can partially buffer these behavioral abnormalities. This positions the social environment as a factor that can modulate behavioral phenotypes, rather than treating mutant behavior as a fixed intrinsic deficit. The abstract uses 'attenuated mutant hyperactivity and restored more typical rest pattern' and 'social context can partially buffer', which is suggestive rather than confirmatory; the loaded text provides no effect sizes or control details.

The framework targets behavioral phenotyping that requires long-term, non-invasive, simultaneous monitoring of multiple animals, and applies to social, exploratory, and nest-building dimensions in neurodegenerative and neurodevelopmental models; it is directly useful for laboratories seeking to reduce manual observation burden and advance 3R principles, and for researchers treating social environment as a phenotype-modulating variable. Results apply to the home-cage long-term monitoring setting rather than to short single-behavior tests.

The loaded text is only an abstract and contains no figures, sample sizes, statistical tests, or effect sizes, so the magnitude of mutant deficits, the robustness of MUT-WT versus MUT-MUT comparisons, and the boundary conditions of social buffering remain open questions; moreover, trailing is described as 'unreciprocated', and the criteria for that judgment and the reliability of automated classification would need confirmation in the full methods.