Digital phenotyping from wearables using AI characterizes psychiatric disorders and identifies genetic associations

Psychiatric disorders are influenced by genetic and environmental factors. However, their study is hindered by limitations on precisely characterizing human behavior. New technologies such as wearable sensors show promise in surmounting these limitations in that they measure heterogeneous behavior in a quantitative and unbiased fashion. Here, we analyze wearable and genetic data from the Adolescent Brain Cognitive Development (ABCD) study. Leveraging >250 wearable-derived features as digital phenotypes, we show that an interpretable AI framework can objectively classify adolescents with psychiatric disorders more accurately than previously possible. To relate digital phenotypes to the underlying genetics, we show how they can be employed in univariate and multivariate GWAS. Doing so, we identify 16 significant genetic loci and 37 psychiatric-associated genes, including ELFN1 and ADORA3, demonstrating that continuous, wearable-derived features give greater detection power than traditional, case-control GWAS. Overall, we show how wearable technology can help uncover new linkages between behavior and genetics.

The raw data is available from abcdstudy.org. Please contact [email protected] for additional questions or requests.

GWAS

Univariate GWAS

For univariate GWAS we employed plink2.

Binary trait

plink2 \
  --threads 6 \
  --out out/gwas_sumstats \
  --keep indivs2keep.txt \
  --pfile imputed/pgen.files/genotype \
  --pheno phenotypes.binary.tsv \
  --chr 1-22,X \
  --covar covariates.tsv \
  --covar-variance-standardize \
  --glm firth-fallback hide-covar omit-ref no-x-sex

Continuous trait

  --threads 6 \
  --out out/gwas_sumstats \
  --keep indivs2keep.txt \
  --pfile imputed/pgen.files/genotype \
  --pheno phenotypes.quantitative.tsv \
  --pheno-name cbcl_externalizing,cbcl_internalizing,liability_Xception_without_cbcl,liability_Xception_with_CBCL,liability_Xception_without_cbcl_v2,liability_Xception_with_CBCL_v2,XGB_without_cbcl,XGB_with_cbcl,XGB_without_cbcl_v2,XGB_with_cbcl_v2 \
  --chr 1-22,X \
  --covar covariates.tsv \
  --covar-variance-standardize \
  --glm hide-covar omit-ref no-x-sex

Multivariate GWAS

For multivariate GWAS we employed mvgwas-nf.

nextflow run mvgwas.nf --l 1000 --geno all.chr.vcf.gz --pheno phenotypes.tsv --cov covariates.tsv --out mvgwas.tsv -resume -with-singularity -with-trace -bg -with-mpi

For all other code requests please email [email protected]