[NeuroCoder]

Autism can be reliably diagnosed at 2.5 years of age, so we probably aren't looking at a strong cultural bias in in how kids are presenting with autism. These models using data from countries where the healthcare systems provides support for appropriate screening and testing further minimizing bias due to culture. In other words, one would not expect these models to be overfit to phenotypically distinct representations of autism as a result of culture.

Variations in brain plasticity have been suspected of playing a role in autism for a long time. Brain plasticity may vary by region, throughout development, and at a molecular level. If population dependent variations in plasticity are indeed responsible for the lack of model generalizability, then the next step would be to do as I previously suggested. That is, identify where the many genome wide variations converge on biomarkers driving differences in plasticity.

[ruthmarx]

> Autism can be reliably diagnosed at 2.5 years of age, so we probably aren't looking at a strong cultural bias in in how kids are presenting with autism. These models using data from countries where the healthcare systems provides support for appropriate screening and testing further minimizing bias due to culture.

I'm not talking about bias due to culture but culture having a real physical effect on development. It wouldn't be noticeable in toddlers but would in older children, if present.

> Brain plasticity may vary by region,

That was my point, that local culture can play a large role in shaping how autism can manifest.

> then the next step would be to do as I previously suggested. That is, identify where the many genome wide variations converge on biomarkers driving differences in plasticity.

I think it would make sense to evaluate adults from different cultures with same or similar autism types and compare and contrast their neurology.