Bibliographic record
Abstract
In an ongoing effort of this Journal to develop and further the theories, models, and best practices around eHealth research, this paper argues for the need for a "science of attrition", that is, a need to develop models for discontinuation of eHealth applications and the related phenomenon of participants dropping out of eHealth trials. What I call "law of attrition" here is the observation that in any eHealth trial a substantial proportion of users drop out before completion or stop using the application. This feature of eHealth trials is a distinct characteristic compared to, for example, drug trials. The traditional clinical trial and evidence-based medicine paradigm stipulates that high dropout rates make trials less believable. Consequently eHealth researchers tend to gloss over high dropout rates, or not to publish their study results at all, as they see their studies as failures. However, for many eHealth trials, in particular those conducted on the Internet and in particular with self-help applications, high dropout rates may be a natural and typical feature. Usage metrics and determinants of attrition should be highlighted, measured, analyzed, and discussed. This also includes analyzing and reporting the characteristics of the subpopulation for which the application eventually "works", ie, those who stay in the trial and use it. For the question of what works and what does not, such attrition measures are as important to report as pure efficacy measures from intention-to-treat (ITT) analyses. In cases of high dropout rates efficacy measures underestimate the impact of an application on a population which continues to use it. Methods of analyzing attrition curves can be drawn from survival analysis methods, eg, the Kaplan-Meier analysis and proportional hazards regression analysis (Cox model). Measures to be reported include the relative risk of dropping out or of stopping the use of an application, as well as a "usage half-life", and prediction models reporting demographic usage discontinuation in a population. Differential dropout or usage rates between two interventions could be a standard metric for the "usability efficacy" of a system. A "run-in and withdrawal" trial design is suggested as a methodological innovation for Internet-based trials with a high number of initial dropouts/nonusers and a stable group of hardcore users.
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 | 0.296 | 0.591 |
| Meta-epidemiology (narrow) | 0.002 | 0.001 |
| Meta-epidemiology (broad) | 0.005 | 0.004 |
| Bibliometrics | 0.004 | 0.005 |
| Science and technology studies | 0.005 | 0.025 |
| Scholarly communication | 0.012 | 0.023 |
| Open science | 0.007 | 0.012 |
| Research integrity | 0.014 | 0.031 |
| Insufficient payload (model declined to judge) | 0.018 | 0.007 |
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".