Stratifying risk for onset of type 1 diabetes using islet autoantibody trajectory clustering
Authors' conclusion
This unsupervised machine learning approach provides a novel tool for stratifying risk for type 1 diabetes onset using multiple autoantibody characteristics. These findings suggest that age-dependent changes in IA-2A titres modulate risk for type 1 diabetes onset across 12 years of life. Overall, this work supports incorporation of islet autoantibody timing, type and titre in risk stratification models for aetiologic studies, prevention trials and disease screening.
Study context
- Population
- Children in The Environmental Determinants of Diabetes in the Young study with ≥1 positive islet autoantibody (GADA, IA-2A, IAA) measurement from years 1 to 12 of life
- Participants
- 1,415 children
- Exposure
- Islet autoantibody trajectory clusters
- Outcome
- Risk for T1D onset across autoantibody clusters
Methodological detailsunsupervised machine learning clustering, time-to-event analysis
- Unsupervised machine learning clustering
- Time-to-event analysis
Main finding
As reported in the abstract
T1D risk by autoantibody cluster
- Clusters identified in each year cohort, differing by autoantibody timing, titre and type
- 2-4
- Ages 1–3: clusters with high titres of all three autoantibodies, 1-year risk
- 20.87-56.25%
- Ages 1–3: clusters with high titres of all three autoantibodies, 5-year risk
- 67.73-69.19%
- Ages 4–8: high IA-2A titre clusters, 1-year risk
- 20.88-28.93%
- Ages 4–8: high IA-2A titre clusters, 5-year risk
- 62.73-78.78%
- Ages 4–8: high GADA titre clusters, 1-year risk
- 4.38-6.11%
- Ages 4–8: high GADA titre clusters, 5-year risk
- 25.06-31.44%
- Ages 9–12: clusters with high IA-2A titres alone, 1-year risk
- 14.82-30.93%
- Ages 9–12: clusters with high GADA and IA-2A titres, 1-year risk
- 8.27-25.00%
Observational data: associations, not effects.
Original publication
Stratifying risk for onset of type 1 diabetes using islet autoantibody trajectory clustering. Diabetologia. Published 30 November 2022. doi:10.
Composed by AI from the published abstract and checked against it. How Briefs are made →