DT4H uses cohorts as structured population context, not marketing segmentation. Cohorts define comparable baselines, expected variance, and inference priors.
Biomarkers · behavior · clinical context · self-report
Weighted membership across relevant population contexts
Expected ranges · variance · trajectories
Cohort-aware model starting state
Places an individual inside comparable biological, behavioral, and clinical contexts.
Represents cohort fit as weighted membership, not a single fixed segment.
Feeds reference-human expectations used for initialization and comparison.
Defines expected range rather than reducing populations to averages.
Allows cohort fit to change as the Twin accumulates evidence.
Provides context for evaluating drift, progress, and unexpected outcomes.
Cohort assignment is probabilistic context, not a fixed label.
Cohort fit does not independently determine clinical condition.
Reference humans are distributional priors, not generic avatars.
Cohort assumptions must update as evidence accumulates.
Cohort membership should be weighted and revisable, not a hard permanent label.
Cohorts provide modeling context; they should not be presented as fixed user identity.
As evidence accumulates, cohort fit should update and influence calibration.