New Search Strategies Successfully Optimize Retrieval of Clinically Sound Treatment Studies in EMBASE. A review of: Wong, Sharon S‐L, Nancy L. Wilczynski, and R. Brian Haynes. “Developing Optimal Search Strategies for Detecting Clinically Sound Treatment Studies in EMBASE.” Journal of the Medical Library Association 94.1 (Jan. 2006): 41‐47. 14 May 2007 http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=1324770.
Bibliographic record
Abstract
Objective – To develop and test the sensitivity and specificity, precision andaccuracy of search strategies to retrieve clinically sound treatment studies in the EMBASE database. Design – Analytical study. Setting – Methodologically sound studies of treatment from 55 journals indexed in EMBASE for the year 2000. Subjects – EMBASE and hand searches performed at the Health Information Research Unit of McMaster University, Ontario, Canada. Methods – The authors compare the results of EMBASE searches using their search strategies with the “gold standard” of articles retrieved by hand search. Research assistants initially hand searched each issue of 55 selected journals published in 2000 to identify articles detailing studies on healthcare treatment of humans. Subject coverage of the journals was wide ranging and included obstetrics and gynaecology, psychiatry, oncology, neurology, surgery and general practice. Studies were then assessed to ensure they met the qualifying criteria: random allocation of participants to groups, outcome assessment of at least 80% of participants who began the study, and analysis consistent with study design. Initially, 3850 articles on treatment were identified, of which 1256 (32.6%) were methodologically sound. To construct a comprehensive set of search terms, input was sought from librarians and researchers in the US and Canada. This initially produced a list of 5385 terms, of which 4843 were unique and 3524 produced hits. Individual search terms with sensitivity greater then 25% and specificity greater then 75% were incorporated into search strategies for use within the OVID interface for the EMBASE database to retrieve articles meeting the same criteria. These strategies were developed using all 27,769 articles published in the 55 journals in 2000. This all inclusive approach was used to test the search strategies’ ability to identify high quality treatment articles from a larger pool of material. Main results – The single term which achieved best sensitivity was “random:mp,”with a sensitivity of 95.1%. This same term achieved a high specificity of 92.5%. The best‐performing single term for specificity was “randomized:tw” at 96.7%, but this did reduce sensitivity to 63.2%. The single term to achieve the best balance between the two was “clinical trial:mp,” with a sensitivity of 88.3% and specificity of 88.0%. Combining terms produced varied results, and Table 3 within the article details terms used to give the best combinations for sensitivity, specificity and optimisation of both. The best three‐term search strategies for sensitivity achieved a rate just shy of 99% with a specificity of 72.0%, while the optimum three‐term strategy for specificity achieved 96.7% but with a trade off of lowering the rate of sensitivity to 51.7%. The best‐performing combination of search terms to optimise sensitivity and specificity produced values exceeding 92% for both. Conclusion – The authors present search strategies which can successfully be used to retrieve methodologically sound studies on the prevention and treatment of disease and health complications indexed on the EMBASE database. A clear outline of the trade‐off between sensitivity and specificity of the strategies is included.
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 | 0.174 | 0.546 |
| Meta-epidemiology (narrow) | 0.004 | 0.003 |
| Meta-epidemiology (broad) | 0.010 | 0.008 |
| Bibliometrics | 0.068 | 0.050 |
| Science and technology studies | 0.002 | 0.002 |
| Scholarly communication | 0.010 | 0.015 |
| Open science | 0.004 | 0.006 |
| Research integrity | 0.003 | 0.003 |
| Insufficient payload (model declined to judge) | 0.021 | 0.005 |
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".