Clinical

Clinical Research

Overview

Clinical study design, statistical analysis, and regulatory compliance for medical research.

Study Designs

| Design | Level of Evidence | Best For | |--------|------------------|----------| | RCT | I | Treatment efficacy | | Cohort (prospective) | II | Risk factors, prognosis | | Cohort (retrospective) | III | Exposure-outcome associations | | Case-control | III | Rare diseases, risk factors | | Cross-sectional | IV | Prevalence, correlations | | Case report/series | V | Novel observations |

Common Analyses

• Survival analysis: Kaplan-Meier curves, Log-rank test, Cox regression
• Diagnostic accuracy: Sensitivity, specificity, ROC curve, AUC
• Meta-analysis: Fixed/random effects, forest plot, heterogeneity (I-squared)
• Propensity score matching: For observational study confounding
• Nomogram: Predictive model visualization for clinical use

Sample Size Calculation

• Define: alpha (usually 0.05), power (usually 0.80), effect size, outcome type
• Tools: G*Power, R pwr package, Python statsmodels
• Report: formula used, assumptions, expected dropout rate

Regulatory Compliance

• IRB/Ethics committee: Required for all human subjects research
• Informed consent: Written, voluntary, comprehensive
• ClinicalTrials.gov: Registration before enrollment (ICMJE requirement)
• GDPR/HIPAA: Data privacy for patient information
• GCP (Good Clinical Practice): ICH E6 guidelines

Reporting Guidelines

| Guideline | Study Type | |-----------|-----------| | CONSORT | Randomized controlled trials | | STROBE | Observational studies | | PRISMA | Systematic reviews / meta-analyses | | STARD | Diagnostic accuracy | | TRIPOD | Prediction models | | SPIRIT | Study protocols |

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: Zaoqu-Liu
• Source: Zaoqu-Liu/ScienceClaw
• License: MIT

Install and usage instructions live in the source repository linked above.