OhioT1DM Dataset

Source

• Dataset metadata snapshot: source.md
• Metadata JSON: metadata.json
• Canonical external source: https://smarthealth.cs.ohio.edu/OhioT1DM-dataset.html

Action-Time-Series Notes

• Dataset type: type-1 diabetes longitudinal physiology.
• Temporal structure: continuous glucose monitoring and patient-event records over multiple weeks.
• Actions or interventions: basal insulin, bolus insulin, meals/carbohydrates, exercise, sleep, stress, and related self-management events.
• Suitability: strong small-scale physiology-control dataset; limited participants but clean action semantics.

Foundation TSFM Relevance

Agenda slot	Verdict	Evidence	Missing pieces
Causal structure, counterfactuals, and control	partially closes	Records glucose observations with insulin, meals, exercise, sleep, stress, and related self-management events as intervention-like inputs.	Small participant count and observational self-management logs limit causal identification.
Time representation and irregular event streams	adjacent	Combines continuous glucose monitoring with timestamped patient-event records over weeks.	Needs explicit irregular-event modeling and intervention-dose semantics in the wiki artifact.

Links Into The Wiki

• Action-Conditioned Time-Series Datasets
• Foundation Time-Series Model Research Agenda
• World Models