Aligning Federated Risk Identification with Equitable Capacity for Diabetes Prevention
Year: 2026
Type: preprint
Abstract: Federated diabetes prediction can expand risk identification without centralising records, but prevention depends on accessible services. This study develops a capacity-constrained model connecting identification, navigation, enrolment and programme completion. Published English programme counts provide empirical context; intervention effects and costs are hypothetical. Seven strategies were evaluated for 1,000 eligible adults over twelve months, with sensitivity analyses varying coverage, navigation benefit and resource requirements. Enhanced identification alone increased expected completions from 220.8 to 307.2 while widening the completion gap from 16.8 to 19.2 percentage points. With 180 navigator hours, throughput maximisation produced 343.2 completions and a 25.2-point gap. Preventing disparity from worsening retained 341.1 completions, a 0.6% reduction. Halving only the barrier-group navigation benefit increased this loss to 4.8%, demonstrating that equity costs depend on relative productivity. At least 53.3 navigator hours were needed to restore the baseline gap under base assumptions. Enhanced identification cost an illustrative £656 per additional completer against matched navigation. The contribution is a testable capacity-feasibility and comparator framework. Deployment decisions should connect validated prediction to funded support, subgroup participation and demonstrable incremental value.
Source: Preprints.org
Language: English
Field: Medicine
Domain: Health Sciences
Open Access status: green
OpenAlex ID: W7225331552