Big data approaches to decomposing heterogeneity across the autism spectrum

摘要: Autism is a diagnostic label based on behavior. While the criteria attempts to maximize clinical consensus, it also masks wide degree of heterogeneity between and within individuals at multiple levels analysis. Understanding this multi-level high translational importance. Here we present organizing principles frame work examining in autism. Theoretical concepts such as 9spectrum9 or 9autisms9 reflect non-mutually exclusive explanations regarding continuous/dimensional categorical/qualitative variation individuals. However, common practices small sample size studies case-control models are suboptimal for tackling heterogeneity. Big data an important ingredient furthering our understanding In addition being 9feature-rich9, big should be both 9broad9 (i.e. large size) 9deep9 collected same individuals). These characteristics help ensure results from population more generalizable facilitate evaluation utility different A model9s can shown by its ability explain clinically mechanistically phenomena, but explaining how variability manifests across The directionality bottom-up top-down, include importance development characterizing change progress made with 9supervised9 built upon priori theoretically predicted distinctions dimensions importance, will become increasingly complement unsupervised data-driven discoveries that leverage unknown multivariate data. Without better model autistic people, towards goal precision medicine may limited.