Extract Data examples by discipline

Ready-to-copy field sets for twelve research disciplines — from clinical medicine and public health to computer science, engineering, and economics.

Copy a set, then edit the wording to match your corpus. Type is the output structure: Answer, Yes/Maybe/No, or Specified with the choice list shown. For the full overview of what you can do with Extract Data — twelve field families, fifteen use cases, and four starter packs — see What you can do with Extract Data. For how to word a field, see Tips for creating better extraction fields.

1. Clinical medicine — intervention trials

Example topic: Effect of early mobilisation on ICU length of stay.

Field	Type	Instructions
Study design	Specified — RCT, Cluster RCT, Non-randomised trial, Prospective cohort, Retrospective cohort, Case-control, Before-after	Classify the design as the authors label it. Pick the single best fit.
Analysed sample	Answer	Number of participants in the primary analysis, not screened or randomised. Total first, per-arm in parentheses if reported. Format: "294 (148/146)".
Population	Answer	Who was studied: condition, mean or median age with its spread, and % female, exactly as printed. Format: "Mechanically ventilated ICU adults; mean age 62.4 (SD 14.1); 41% female".
Intervention	Answer	What the intervention group received: what was done, how often, for how long, and by whom. Format: "Progressive mobilisation, twice daily, from ICU day 1, physiotherapist-delivered".
Comparator	Answer	What the control group received, in the authors' words. If usual care, say what usual care included if described.
Primary outcome	Answer	The outcome the authors designate as primary, and how it was measured. Do not list secondary outcomes here.
Primary result	Answer	The effect on the primary outcome with its 95% CI and p-value exactly as printed, naming the metric. Format: "MD −1.9 days (95% CI −3.1 to −0.7), p=0.002".
Follow-up	Answer	Longest timepoint at which outcomes were measured, in the paper's unit. Format: "90 days".
Adverse events	Answer	Serious adverse events with denominators, per arm, exactly as printed. Do not convert counts to percentages. If none reported, answer "Not reported".
Author limitations	Answer	Limitations the authors state themselves, one short bullet each. Do not add your own.

2. Public health & epidemiology

Example topic: Air pollution exposure and childhood asthma incidence.

Field	Type	Instructions
Study design	Specified — Prospective cohort, Retrospective cohort, Case-control, Cross-sectional, Ecological, Time-series	Classify the design as the authors describe it.
Exposure measure	Answer	The pollutant and how exposure was assigned: monitoring station, land-use regression, satellite, personal monitor. Include averaging period. Format: "PM2.5, land-use regression, annual mean".
Outcome definition	Answer	How the health outcome was defined and ascertained: diagnosis code, registry, parent report, spirometry.
Population and setting	Answer	Age range, country or city, and cohort name if any. Format: "Children 0–12 y; Barcelona, Spain; INMA cohort".
Sample size	Answer	Number of individuals in the analysis, and person-years if reported. Format: "3,412 children; 28,700 person-years".
Confounders adjusted	Answer	Variables in the fully adjusted model, listed as printed. Format: comma-separated list.
Effect estimate	Answer	Effect for the main exposure–outcome pair, with metric, per-unit increment, and 95% CI exactly as printed. Format: "HR 1.13 (95% CI 1.04–1.22) per 5 µg/m³ PM2.5".
Study period	Answer	Calendar years of data collection. Format: "2008–2016".

3. Pharmacology & drug safety

Example topic: Adverse events reported for GLP-1 receptor agonists.

Field	Type	Instructions
Drug and dose	Answer	Drug studied with dose and schedule exactly as reported. Format: "Semaglutide 2.4 mg once weekly, subcutaneous".
Indication	Answer	The condition treated as stated by the authors.
Data source	Specified — RCT, Observational cohort, Spontaneous reports / pharmacovigilance, Claims database, Registry, Case report	Classify where the safety data came from.
Exposure period	Answer	Treatment duration and total exposure if reported. Format: "68 weeks; 1,961 patient-years".
Adverse events reported	Answer	All adverse events the paper reports with incidence, one bullet each, with denominators as printed.
Serious adverse events	Answer	Serious adverse events with denominators per arm, exactly as printed. If none, answer "Not reported".
Discontinuations	Answer	Number discontinuing due to adverse events, with denominator, per arm. Format: "70/1306 (5.4%) vs 15/652 (2.3%)".
Signal noted	Yes/Maybe/No	Do the authors identify a new or unexpected safety signal? yes only if they explicitly say so. no if no new or unexpected signal is described anywhere. maybe if a potential signal is discussed only tentatively.
Funding source	Specified — Industry, Public/government, Charity/foundation, Mixed, None stated	Classify the funding as stated in the funding or COI section.

4. HEOR & health economics

Example topic: Cost-effectiveness of biologics in rheumatoid arthritis.

Field	Type	Instructions
Model type	Specified — Markov, Decision tree, Discrete event simulation, Partitioned survival, Trial-based analysis, Other model	Classify the economic model structure as described.
Perspective	Specified — Healthcare payer, Societal, Provider, Patient, Multiple	Classify the stated analytic perspective.
Comparators	Answer	Interventions compared, one short bullet each, in the authors' wording.
Time horizon	Answer	The base-case time horizon exactly as stated. Format: "Lifetime (50 years)".
Discount rate	Answer	Discount rates for costs and outcomes as stated. Format: "3.5% costs, 3.5% outcomes".
Cost year and currency	Answer	Price year and currency of reported costs. Do not convert. Format: "2022 GBP".
Utility source	Answer	Where utility values came from: instrument, population, and citation as described. Format: "EQ-5D-3L, UK tariff, from trial data".
ICER	Answer	Base-case incremental cost-effectiveness ratio with currency, price year, and unit. Format: "£24,300 per QALY (2022 GBP)".
Sensitivity analysis	Answer	Which parameters drove the result in one-way or probabilistic sensitivity analysis, as the authors report.
Conclusion	Specified — Cost-effective at stated threshold, Not cost-effective, Dominant, Dominated, Mixed by subgroup	Classify the authors' base-case conclusion.

5. Medical devices, diagnostics & regulatory evidence

Example topic: Clinical performance of point-of-care troponin assays.

Field	Type	Instructions
Device or assay	Answer	Name and manufacturer of the device or assay evaluated, exactly as printed.
Indication and use	Answer	Clinical question and setting in which the device was used. Format: "Rule-out of MI in ED chest pain presentations".
Reference standard	Answer	The comparator standard against which performance was judged, as described.
Sample size	Answer	Number of patients or specimens in the performance analysis, with the number of positives if reported. Format: "1,282 patients; 145 MI-positive".
Sensitivity	Answer	Sensitivity with 95% CI exactly as printed, at the stated threshold. Format: "98.6% (95% CI 95.1–99.8) at 5 ng/L".
Specificity	Answer	Specificity with 95% CI exactly as printed, at the same threshold.
Adverse events / safety	Answer	Device-related adverse events or safety findings reported. If the study reports none, answer "Not reported".
Favourable outcomes	Answer	Findings the authors present as supporting device performance, one bullet each.
Unfavourable outcomes	Answer	Findings the authors present as limitations of device performance, one bullet each.
Funding / sponsor	Specified — Manufacturer-funded, Independent/public, Mixed, None stated	Classify funding as stated.

6. Psychology & behavioural science

Example topic: Efficacy of CBT for adolescent anxiety.

Field	Type	Instructions
Design	Specified — RCT, Quasi-experimental, Single-group pre-post, Cross-sectional, Longitudinal observational	Classify the design as the authors describe it.
Age range	Answer	Age range and mean age of participants exactly as printed. Format: "12–17 y; mean 14.3 (SD 1.6)".
Outcome instrument	Answer	The named scale used for the primary outcome, with version and scoring range if stated. Format: "SCARED, child version, 0–82".
Delivery format	Specified — Individual in-person, Group in-person, Therapist-guided digital, Self-guided digital, Blended	Classify how the intervention was delivered.
Dose	Answer	Number of sessions and session length as reported. Format: "12 sessions × 50 min over 14 weeks".
Control condition	Answer	What the comparison group received: waitlist, treatment as usual, active control, placebo.
Effect size	Answer	Between-group effect size for the primary outcome with 95% CI, naming the metric. Format: "Hedges' g −0.61 (95% CI −0.92 to −0.30)".
Attrition	Answer	Number and percentage lost to follow-up, per arm if reported. Format: "18/94 (19%) vs 11/91 (12%)".
Preregistered	Yes/Maybe/No	Does the paper state that the study or analysis plan was preregistered? yes only if a registry or ID is given. no if no registration statement or ID appears anywhere. maybe if registration is mentioned without an identifier.

7. Computer science & machine learning

Example topic: Retrieval-augmented generation methods on open-domain QA.

Field	Type	Instructions
Method name	Answer	The name the authors give their proposed method or model. Format: name only.
Architecture	Answer	Model architecture and size, and the base model it builds on. Format: "Decoder-only, 7B, Llama-2 base".
Task and dataset	Answer	Task addressed and the datasets evaluated on, as named by the authors. Format: comma-separated dataset names.
Training data	Answer	Data used for training or fine-tuning, with size, exactly as reported. If the model is used zero-shot, answer "Not applicable".
Primary metric	Specified — Accuracy, F1, Exact match, BLEU/ROUGE, AUC, Human evaluation, Other	Classify the headline metric the authors emphasise.
Best result	Answer	Highest score the authors report for their method on the primary benchmark, with the metric and dataset. Copy decimals exactly. Format: "54.2 EM (NQ)".
Baselines compared	Answer	Baselines the method is compared against, one short bullet each, with their reported scores if given.
Ablations run	Answer	Ablation studies reported and what each removed, one bullet each. If none, answer "Not reported".
Compute reported	Answer	Hardware and training or inference cost as reported. Format: "8× A100 80GB, 36 GPU-hours". If not stated, answer "Not reported".
Code available	Yes/Maybe/No	Does the paper state that code is publicly available? yes only if a repository link or explicit availability statement is given. no if no code-availability statement or link appears anywhere, including footnotes and appendix. maybe if code is available on request or promised for later release.

8. Engineering & materials science

Example topic: Mechanical properties of recycled PET composites.

Field	Type	Instructions
Material system	Answer	Matrix and reinforcement with weight or volume fractions exactly as printed. Format: "rPET + 20 wt% short glass fibre".
Processing method	Answer	How specimens were produced, with key parameters. Format: "Twin-screw extrusion, 265 °C, then injection moulding".
Test standard	Answer	The standard followed for the primary test, exactly as cited. Format: "ASTM D638 Type I".
Specimens tested	Answer	Number of specimens per condition, as reported. Format: "n=5 per condition".
Tensile strength	Answer	Tensile strength for the primary composition, with unit and standard deviation as printed. Do not convert units. Format: "48.2 ± 2.1 MPa".
Modulus	Answer	Young's modulus for the same composition, with unit and spread as printed.
Test conditions	Answer	Temperature, humidity, strain rate, and conditioning as reported.
Comparison to virgin material	Yes/Maybe/No	Does the paper compare the recycled material against a virgin-material control? yes if a virgin-material control is tested or reported alongside. no if no such comparison appears anywhere in the results. maybe if compared only against literature values.

9. Environmental science & ecology

Example topic: Effectiveness of marine protected areas on fish biomass.

Field	Type	Instructions
Location	Answer	Study location with country and named site or region. Format: "Medes Islands, Spain (NW Mediterranean)".
Protection level	Specified — No-take reserve, Partially protected, Multi-use, Unprotected control, Mixed	Classify the protection status studied.
Time since designation	Answer	Years between protection designation and data collection, as reported. Format: "18 years".
Sampling method	Answer	How the outcome was measured, with effort. Format: "Underwater visual census, 50 × 5 m transects, n=120".
Outcome measured	Answer	The ecological outcome and its unit exactly as reported. Format: "Total fish biomass, g/m²".
Effect reported	Answer	Difference between protected and control, with metric and uncertainty exactly as printed. Format: "+3.1× biomass inside reserve (95% CI 2.2–4.4)".
Confounders addressed	Answer	Habitat, depth, or fishing-pressure differences the authors control for, as described. If not addressed, answer "Not reported".

10. Education

Example topic: Impact of flipped classrooms on student outcomes.

Field	Type	Instructions
Education level	Specified — Primary, Secondary, Undergraduate, Postgraduate, Professional/CPD, Mixed	Classify the level of the learners studied.
Subject area	Answer	The subject or course in which the intervention was delivered, as named.
Design	Specified — RCT, Quasi-experimental with control, Pre-post single group, Cross-sectional survey, Qualitative	Classify the study design.
Sample	Answer	Number of students analysed, and number of classes or institutions if reported. Format: "412 students, 6 sections, 1 university".
Implementation length	Answer	How long the intervention ran, in the paper's unit. Format: "One 15-week semester".
Outcome measure	Answer	How learning was measured: exam score, standardised test, grade, self-report scale. Name the instrument.
Effect on outcome	Answer	Difference between conditions with effect size and CI or p-value exactly as printed.
Control condition	Answer	What the comparison group experienced, in the authors' words.
Fidelity reported	Yes/Maybe/No	Does the paper report whether the intervention was implemented as designed? yes if fidelity or adherence data is reported. no if implementation fidelity is not mentioned anywhere. maybe if fidelity is discussed without data.

11. Qualitative research & implementation science

Example topic: Barriers to telehealth uptake among older adults.

Field	Type	Instructions
Method	Specified — Semi-structured interviews, Focus groups, Ethnography/observation, Document analysis, Mixed qualitative	Classify the primary data-collection method.
Theoretical framework	Answer	The named theory, model, or framework guiding the study. If none is named, answer "Not reported".
Participants	Answer	Who took part, how many, and how they were recruited. Format: "24 adults aged 68–91, purposive sampling from 3 primary care practices".
Setting	Answer	Where the study took place: country, service, and context.
Analysis approach	Answer	The analytic method as named, with any coding procedure described. Format: "Reflexive thematic analysis, two coders, NVivo".
Themes identified	Answer	The themes the authors report, one short bullet each, in the authors' own labels.
Barriers reported	Answer	Barriers the authors identify, one short bullet each. Include only barriers this study found.
Facilitators reported	Answer	Facilitators the authors identify, one short bullet each. If none, answer "Not reported".
Saturation reported	Yes/Maybe/No	Do the authors state that data saturation was reached? yes only if explicitly stated. no if saturation is not mentioned anywhere. maybe if implied (e.g. "no new themes emerged") without the term.
Ethics approval	Answer	The ethics committee and approval reference exactly as printed. If absent, answer "Not reported".

12. Economics, business & policy

Example topic: Minimum wage effects on employment.

Field	Type	Instructions
Jurisdiction and period	Answer	Geography and years covered by the analysis. Format: "United States, 1990–2019, state-level".
Data source	Answer	The dataset used, named exactly, with its unit of observation. Format: "QCEW, county-quarter".
Identification strategy	Specified — Difference-in-differences, Synthetic control, Instrumental variables, Regression discontinuity, Event study, OLS/descriptive	Classify the main identification approach.
Treatment definition	Answer	How the policy or treatment is defined and measured in the analysis.
Outcome variable	Answer	The dependent variable and its unit exactly as defined. Format: "Log teen employment rate".
Main estimate	Answer	The headline coefficient with its standard error or CI exactly as printed, naming the specification. Format: "−0.021 (SE 0.014), preferred DiD specification".
Robustness checks	Answer	Robustness or placebo tests reported, one short bullet each.
Data/code available	Yes/Maybe/No	Does the paper state that replication data or code is available? yes if an availability statement or link is given. no if no such statement appears anywhere. maybe if available on request.