Testing and Iterative Improvement of the CEN ISO/TS 82304-2 Health App Quality Assessment: Pilot Interrater Reliability Study
Bibliographic record
Abstract
BACKGROUND: With the increasing use of health apps and ongoing concerns regarding their safety, effectiveness, and data privacy, numerous health app quality assessment frameworks have emerged. However, assessment initiatives experience difficulties scaling, and there is currently no comprehensive, consistent, internationally recognized assessment framework. Therefore, health apps often need to undergo several quality evaluations to enter different markets, leading to duplication of work. The CEN ISO/TS 82304‑2 health app assessment seeks to address this issue, aiming to provide an internationally accepted quality evaluation through a network of assessment organizations located in different countries. OBJECTIVE: This study aimed to develop and evolve the draft CEN ISO/TS 82304-2 assessment handbook and developer guidance by testing them across organizations in several countries. METHODS: Assessment organizations from 5 countries were engaged to evaluate 24 health apps using the evolving CEN ISO/TS 82304-2 assessment across 3 evaluation rounds. The information submitted by a given health app developer was evaluated by 2 assessment organizations, and interrater reliability was examined. In addition, app developers and assessors were asked to report how much time they spent on information collation or evaluation and to rate the clarity of the developer guidance or assessor handbook, respectively. The collected data were used to iteratively improve the handbook and guidance between rounds. RESULTS: The interrater reliability between assessment organizations improved from round 1 to round 2 and stayed relatively stable between rounds 2 and 3, with 80% (55/69) of assessment questions demonstrating moderate or better (Gwet AC1>0.41) agreement in round 3. The median time required by developers to prepare the assessment information was 8 hours and 59 minutes (IQR 5.7-27.1 hours) in round 3, whereas assessors reported a median evaluation time of 8 hours and 46 minutes (IQR 7.1-11.0 hours). The draft guidance and handbook were generally perceived as clear, with a median round-3 clarity rating of 1.73 (IQR 1.64-1.90) for developers and 1.78 (IQR 1.71-1.89) for assessors (0="very unclear", 1="somewhat unclear", and 2="completely clear"). CONCLUSIONS: To our knowledge, this is the first study to examine the consistency of health app evaluations across organizations located in different countries. Given that the CEN ISO/TS 82304-2 guidance and handbook are still under development, the interrater reliability findings observed at this early stage are promising, and this study provided valuable information for further refinement of the assessment. This study marks an important first step toward establishing the CEN ISO/TS 82304-2 assessment as a consistent, cross-national health app evaluation. It is envisioned that the assessment will ultimately help avoid duplication of work, prevent inequities by facilitating access to smaller markets for developers, and build trust among users, thereby increasing the adoption of high-quality health apps.
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 | 0.224 | 0.249 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.001 | 0.002 |
| Bibliometrics | 0.003 | 0.002 |
| Science and technology studies | 0.002 | 0.003 |
| Scholarly communication | 0.003 | 0.003 |
| Open science | 0.002 | 0.006 |
| Research integrity | 0.001 | 0.002 |
| Insufficient payload (model declined to judge) | 0.001 | 0.001 |
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; the direct Gemma label and the distilled Codex classifier agree on what is shown here.
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".