Validation of The Umbrella Collaboration for Tertiary Evidence Synthesis in Geriatrics: Mixed Methods Study
Bibliographic record
Abstract
BACKGROUND: The synthesis of evidence in healthcare is essential for informed decision-making and policy development. This study aims to validate The Umbrella Collaboration® (TU®), an innovative, semi-automatic tertiary evidence synthesis methodology, by comparing it with Traditional Umbrella Reviews (TUR), which are currently the gold standard. OBJECTIVE: The primary objective of this study is to evaluate whether TU®, an AI-assisted, software-driven system for tertiary evidence synthesis, can achieve comparable effectiveness to TURs, while offering a more timely, efficient, and comprehensive approach. METHODS: This comparative study evaluated TU® against TURs across eight matched projects in geriatrics. For each selected TUR, a parallel TU® project was conducted using the same research question. Outcomes of interest (OoIs), effect sizes, certainty ratings, and execution times were systematically compared. Effect sizes were assessed both quantitatively, by transforming TUR metrics to Cohen's d and correlating them with TU®'s RTU metric, and qualitatively, through categorical classifications (trivial, small, moderate, large). Certainty levels were compared by mapping GRADE ratings and TU®'s sentiment analysis scores onto a common 0-1 scale. Execution time was measured precisely in TU®, while TUR durations were estimated from literature benchmarks. Statistical analyses included chi-squared tests and Spearman correlations. RESULTS: Eight TURs in geriatrics were matched with parallel projects using TU®. TU® replicated 84.9% (73/86) of the OoIs identified by TURs and reported an additional 337 OoIs, representing a 4.77-fold increase in outcome identification. In the comparison of effect size classifications, full concordance was observed in 50.0% of cases and consistent concordance (full plus one-level deviation) in 93.8%, with a moderate strength of association (Cramér's V = 0.339). The correlation of transformed certainty values between TU® and GRADE yielded a statistically significant Spearman coefficient (ρ = 0.446; P = .025). The average execution time per TU® project was 4 hours and 46 minutes, compared to estimated durations of 6-12 months for TURs. CONCLUSIONS: The Umbrella Collaboration® demonstrated high concordance with TURs, replicating 84.9% of the outcomes identified by TURs and identifying nearly five times as many additional outcomes. The experimental effect size metric (RTU) showed moderate agreement with conventional measures, and the certainty ratings derived from sentiment analysis correlated acceptably with GRADE-based assessments. While further validation is needed, TU® appears to be a valid and efficient approach for tertiary evidence synthesis, offering a scalable and time-efficient alternative when rapid results are required. INTERNATIONAL REGISTERED REPORT: RR2-10.2196/67248.
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 | 0.525 | 0.716 |
| Meta-epidemiology (narrow) | 0.002 | 0.002 |
| Meta-epidemiology (broad) | 0.004 | 0.008 |
| Bibliometrics | 0.011 | 0.011 |
| Science and technology studies | 0.003 | 0.002 |
| Scholarly communication | 0.007 | 0.004 |
| Open science | 0.004 | 0.008 |
| Research integrity | 0.002 | 0.002 |
| Insufficient payload (model declined to judge) | 0.005 | 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".