clinical-decision-support_skill

Overview

This skill generates professional clinical decision support (CDS) documents tailored for pharmaceutical research, clinical guideline development, and regulatory submissions. It produces publication-ready LaTeX/PDF reports that combine biomarker-stratified cohort analyses and evidence-based treatment recommendation reports. Outputs are formatted for compact, data-dense presentation and optimized for drug development and evidence synthesis workflows.

How this skill works

Provide the target disease state, cohort data or literature sources, and required deliverables (cohort analysis or treatment recommendations). The skill performs biomarker integration, statistical analyses (survival curves, hazard ratios, Cox regression, p-values), and GRADE evidence grading, then assembles a LaTeX document with mandatory executive summary, figures, tables, and formatted recommendations. Every document includes at least one AI-generated schematic (decision flowchart, cohort flow diagram, or stratification tree) and is rendered as a publication-ready PDF with 0.5in margins.

When to use it

• Produce biomarker-stratified cohort analyses for phase 2/3 trials or real-world evidence studies
• Create evidence-based treatment recommendation reports with GRADE grading and decision algorithms
• Prepare regulatory-facing documents (IND/NDA appendices, subgroup efficacy summaries)
• Develop clinical practice guidelines or consensus recommendations for specialty societies
• Generate medical affairs materials: KOL briefs, advisory board summaries, or publication-ready analyses

Best practices

• Provide clean, de-identified cohort datasets and a clear analysis plan to ensure reproducible results
• Specify biomarkers, endpoints (OS, PFS, ORR), and comparison groups up front to guide statistical testing
• Request required figure types (Kaplan-Meier, waterfall, flow diagram) to satisfy the schematic mandate
• Use standardized vocabularies (SNOMED-CT, LOINC) and follow ICH-GCP formatting for regulatory submissions
• Include source references and trial identifiers to support GRADE assessments and evidence tables

Example use cases

• Biomarker-stratified efficacy report for a novel oncology agent with Kaplan-Meier plots and hazard ratios
• Guideline-style treatment recommendations for second-line therapy with GRADE 1A/2B grading and TikZ flowchart
• Subgroup analysis report for a Phase 3 study showing forest plots across molecular subtypes
• Real-world evidence cohort study summarizing outcomes by genomic alteration with waterfall and swimmer plots
• Companion diagnostic strategy document linking biomarker thresholds to line-of-therapy recommendations