Resistance to data loss from the Freestyle Libre: impact on glucose variability indices and recommendations for data analysis
Bibliographic record
Abstract
Like many wearables, flash glucose monitoring relies on user compliance and is subject to missing data. As recent research is beginning to utilise glucose technologies as behaviour change tools, it is important to understand whether missing data are tolerable. Complete Freestyle Libre data files were amputed to remove 1–6 h of data both at random and over mealtimes (breakfast, lunch, and dinner). Absolute percent errors (MAPE) and intraclass correlation coefficients (ICC) were calculated to evaluate agreement and reliability. Thirty-two (91%) participants provided at least 1 complete day (24 h) of data (age: 44.8 ± 8.6 years, female: 18 (56%); mean fasting glucose: 5.0 ± 0.6 mmol/L). Mean and continuous overall net glycaemic action (CONGA) (60 min) were robust to data loss (MAPE ≤3%). Larger errors were calculated for standard deviation, coefficient of variation (CV) and mean amplitude of glycaemic excursions (MAGE) at increasing missingness (MAPE: 2%–10%, 2%–9%, and 4%–18%, respectively). ICC decreased as missing data increased, with most indicating excellent reliability (>0.9) apart from certain MAGE ICCs, which indicated good reliability (0.84–0.9). Researchers and clinicians should be aware of the potential for larger errors when reporting standard deviation, CV, and MAGE at higher rates of data loss in nondiabetic populations. But where mean and CONGA are of interest, data loss is less of a concern. Novelty: As research now utilises flash glucose monitoring as behavioural change tools in nondiabetic populations, it is important to consider the influence of missing data. Glycaemic variability indices of mean and CONGA are robust to data loss, but standard deviation, CV, and MAGE are influenced at higher rates of missingness.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.235 | 0.581 |
| Meta-epidemiology (narrow) | 0.002 | 0.001 |
| Meta-epidemiology (broad) | 0.003 | 0.002 |
| Bibliometrics | 0.004 | 0.007 |
| Science and technology studies | 0.001 | 0.002 |
| Scholarly communication | 0.008 | 0.005 |
| Open science | 0.006 | 0.003 |
| Research integrity | 0.004 | 0.005 |
| Insufficient payload (model declined to judge) | 0.003 | 0.003 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".