Analysis of the Evolution of Participatory Health Technology in Type 1 Diabetes from the ".com" to LLM Era

The process of translating technological innovations from research to clinical practice in diabetes care remains slow. Analysing when technologies first appear in scientific literature versus clinical trials can reveal barriers to implementation.

We conducted a longitudinal text-mining analysis of type 1 diabetes-related technologies from peer-reviewed publications and the ClinicalTrials database (1990-2029). Technology-related keywords were extracted using a large language model and classified into thematic categories. Temporal enrichment in five-year intervals was assessed using Fisher's exact test and a hypergeometric test.

The literature revealed a greater number of significant enrichments than the clinical trials (33/98 vs. 23/98), with only 10 results overlapping between the two. Insulin administration systems were the most frequently enriched category in both datasets, while social media and communications, and hardware for healthcare professionals were rarely represented. Literature-based enrichments consistently preceded those in clinical trials, revealing a temporal lag between discovery and clinical testing.

These findings highlight the ongoing translational gap in digital diabetes research. Integrating implementation science into technology development and evaluation could speed up the adoption of innovations in clinical practice.