Clinical Research Tools
Use this skill for group-level human research work, not bedside care.
Typical triggers:
- choose between RCT, cohort, case-control, cross-sectional, diagnostic, or single-arm designs
- define primary and secondary endpoints, estimands, eligibility criteria, or subgroup analyses
- select the right reporting guideline such as CONSORT, STROBE, PRISMA, STARD, TRIPOD, SPIRIT, or CARE
- draft protocol, SAP, CSR, or evidence-summary outlines
- review bias, confounding, missing data, and sample-size assumptions
- prepare trial or real-world-evidence summaries for drug-discovery programs
Working Rules
- Keep the task at the study or cohort level.
- Separate confirmed study facts from proposed design choices.
- State assumptions behind endpoint, power, and statistical-model choices.
- Call out data leakage, immortal-time bias, selection bias, and confounding whenever relevant.
- Do not present DrugClaw as giving medical advice, treatment recommendations, or diagnostic decisions.
Study Design Map
Use this quick routing:
RCT: intervention efficacy, causal inference, registration-ready protocolsProspective cohort: prognosis, exposure-outcome tracking, real-world evidenceRetrospective cohort: registry or EHR analyses with explicit confounding controlCase-control: rare outcomes or exploratory risk-factor workCross-sectional: prevalence, survey snapshots, baseline association workDiagnostic accuracy: sensitivity, specificity, ROC, calibration, decision curvesPrediction model: risk scores, survival models, treatment-response models with external validation plans
Reporting Guideline Map
Choose and state the governing framework early:
CONSORT: randomized trialsSPIRIT: trial protocolsSTROBE: observational studiesPRISMA: systematic reviews and meta-analysisSTARD: diagnostic accuracy studiesTRIPOD: prediction modelsCARE: case reportsICH E3: clinical study reports
Protocol Workflow
For protocol or study-design requests:
- Define population, intervention or exposure, comparator, outcome, and timeframe.
- State inclusion and exclusion criteria.
- Define primary endpoint, key secondary endpoints, and censoring rules.
- Choose analysis populations: ITT, mITT, per-protocol, safety.
- Describe missing-data handling and sensitivity analyses.
- State sample-size assumptions clearly: alpha, power, effect size, event rate, dropout.
- Specify monitoring, ethics, registration, and data-governance requirements.
Statistical Planning Checklist
Always address:
- endpoint type: binary, continuous, count, time-to-event
- stratification variables and subgroup policy
- multiplicity control
- covariate adjustment policy
- temporal leakage and look-ahead bias
- external validation or temporal validation when building prediction models
- calibration, not only discrimination, for predictive work
Evidence Synthesis
For evidence summaries:
- identify study type and evidence level
- note patient population, line of therapy, biomarker context, and comparator
- distinguish efficacy, safety, and external-validity conclusions
- state what remains uncertain
- use cautious language for indirect or observational evidence
Outputs
Good outputs usually include:
- one-page design summary or protocol skeleton
- endpoint table
- statistical analysis outline
- bias and limitation section
- reporting-guideline checklist
Related Skills
For ClinicalTrials.gov, openFDA, or OpenAlex lookups, activate pharma-db-tools.
For cohort tables, biosignals, or DICOM datasets, activate medical-data-tools.
For citation cleanup, evidence matrices, or structured review drafting, activate literature-review-tools.
For hypothesis tests, regression, or effect-size reporting, activate stat-modeling-tools.
For Kaplan-Meier, log-rank, or Cox workflows, activate survival-analysis-tools.
For manuscript critique, hypothesis framing, or reproducibility checklists, activate scientific-workflow-tools.
For molecular, variant, pathway, or structure work, activate bio-tools, bio-db-tools, chem-tools, or docking-tools as appropriate.
1---2name: clinical-research-tools3description: Clinical research workflow guide for protocol design, endpoint selection, evidence grading, reporting-guideline selection, statistical planning, and clinical-trial evidence synthesis. Use when the user asks to design or review human-subjects research, trial analyses, observational studies, study protocols, CSRs, or clinical evidence summaries without requesting patient-specific diagnosis or treatment decisions.4---56# Clinical Research Tools78Use this skill for group-level human research work, not bedside care.910Typical triggers:11- choose between RCT, cohort, case-control, cross-sectional, diagnostic, or single-arm designs12- define primary and secondary endpoints, estimands, eligibility criteria, or subgroup analyses13- select the right reporting guideline such as CONSORT, STROBE, PRISMA, STARD, TRIPOD, SPIRIT, or CARE14- draft protocol, SAP, CSR, or evidence-summary outlines15- review bias, confounding, missing data, and sample-size assumptions16- prepare trial or real-world-evidence summaries for drug-discovery programs1718## Working Rules19201. Keep the task at the study or cohort level.212. Separate confirmed study facts from proposed design choices.223. State assumptions behind endpoint, power, and statistical-model choices.234. Call out data leakage, immortal-time bias, selection bias, and confounding whenever relevant.245. Do not present DrugClaw as giving medical advice, treatment recommendations, or diagnostic decisions.2526## Study Design Map2728Use this quick routing:29- `RCT`: intervention efficacy, causal inference, registration-ready protocols30- `Prospective cohort`: prognosis, exposure-outcome tracking, real-world evidence31- `Retrospective cohort`: registry or EHR analyses with explicit confounding control32- `Case-control`: rare outcomes or exploratory risk-factor work33- `Cross-sectional`: prevalence, survey snapshots, baseline association work34- `Diagnostic accuracy`: sensitivity, specificity, ROC, calibration, decision curves35- `Prediction model`: risk scores, survival models, treatment-response models with external validation plans3637## Reporting Guideline Map3839Choose and state the governing framework early:40- `CONSORT`: randomized trials41- `SPIRIT`: trial protocols42- `STROBE`: observational studies43- `PRISMA`: systematic reviews and meta-analysis44- `STARD`: diagnostic accuracy studies45- `TRIPOD`: prediction models46- `CARE`: case reports47- `ICH E3`: clinical study reports4849## Protocol Workflow5051For protocol or study-design requests:521. Define population, intervention or exposure, comparator, outcome, and timeframe.532. State inclusion and exclusion criteria.543. Define primary endpoint, key secondary endpoints, and censoring rules.554. Choose analysis populations: ITT, mITT, per-protocol, safety.565. Describe missing-data handling and sensitivity analyses.576. State sample-size assumptions clearly: alpha, power, effect size, event rate, dropout.587. Specify monitoring, ethics, registration, and data-governance requirements.5960## Statistical Planning Checklist6162Always address:63- endpoint type: binary, continuous, count, time-to-event64- stratification variables and subgroup policy65- multiplicity control66- covariate adjustment policy67- temporal leakage and look-ahead bias68- external validation or temporal validation when building prediction models69- calibration, not only discrimination, for predictive work7071## Evidence Synthesis7273For evidence summaries:74- identify study type and evidence level75- note patient population, line of therapy, biomarker context, and comparator76- distinguish efficacy, safety, and external-validity conclusions77- state what remains uncertain78- use cautious language for indirect or observational evidence7980## Outputs8182Good outputs usually include:83- one-page design summary or protocol skeleton84- endpoint table85- statistical analysis outline86- bias and limitation section87- reporting-guideline checklist8889## Related Skills9091For ClinicalTrials.gov, openFDA, or OpenAlex lookups, activate `pharma-db-tools`.92For cohort tables, biosignals, or DICOM datasets, activate `medical-data-tools`.93For citation cleanup, evidence matrices, or structured review drafting, activate `literature-review-tools`.94For hypothesis tests, regression, or effect-size reporting, activate `stat-modeling-tools`.95For Kaplan-Meier, log-rank, or Cox workflows, activate `survival-analysis-tools`.96For manuscript critique, hypothesis framing, or reproducibility checklists, activate `scientific-workflow-tools`.97For molecular, variant, pathway, or structure work, activate `bio-tools`, `bio-db-tools`, `chem-tools`, or `docking-tools` as appropriate.
drugclaw