MedSci Analysis & Figures

Plugin medsci-analysis from marketplace medsci-skills in Aperivue/medsci-skills. Commands read from the repo's manifest — 2 days ago.

Skills (62)

• academic-aio

  Use when a medical AI paper should be found and cited by AI search engines and RAG tools. Optimizes the title, abstract, summary box (Key Points, Research in Context), keywords, preprint, GitHub README/CITATION.cff and Hugging Face card, returning a visible pass/fail checklist.

• add-journal

  Use when a target journal has no profile yet. Reads the journal's author guidelines and writes a detailed /write-paper profile plus a compact /find-journal profile, public or user-local private, in the canonical format with quality gates.

• analyze-stats

  Use when data needs statistical analysis. Runs reproducible Python/R code for Table 1, diagnostic accuracy, agreement, regression, survival, propensity score, survey-weighted and repeated-measures models, with publication tables. Sample size is /calc-sample-size; pooling studies is /meta-analysis.

• architecture-zoo

  Renamed to /model-selection in v6 (removed in v7).

• author-strategy

  Use when analyzing a researcher's publication record from PubMed. Fetches an author's papers, classifies study types and author position, charts the patterns and writes a strategy report, with an optional trajectory-archetype classification. Works from PubMed metadata only.

• batch-cohort

  Use when one validated cohort analysis must be repeated across many exposure/outcome pairs. Generates one R/Python script per combination from a single methodology template, changing only the variables, and aggregates the results into a summary matrix.

• calc-sample-size

  Use when planning how many patients or cases a study needs before data collection (power analysis, IRB justification). Walks a decision tree to the right test and returns reproducible R/Python code and IRB-ready justification text. Analyzing collected data is /analyze-stats.

• check-reporting

  Use when auditing a manuscript item by item against a reporting guideline or risk-of-bias tool. Covers 49 reporting guidelines and risk-of-bias tools (STROBE, CONSORT, STARD, TRIPOD+AI, PRISMA, QUADAS and more), marking each item PRESENT, PARTIAL or MISSING. Not a reviewer critique (/self-review).

• clean-data

  Use when a clinical CSV/Excel dataset needs profiling and cleaning before analysis (missing values, outliers, duplicates, type mismatches). Profiles, flags and generates cleaning code in three stages, each gated on the researcher's approval. Never auto-cleans.

• contribute

  Use when sending local changes back to the project (an added journal profile, a fixed checklist item, an adapted skill) or reporting a false positive or bug. Finds what changed, scans it for patient data, shows every line and sends nothing until you confirm.

• cross-national

  Use when comparing an exposure-outcome association across countries with parallel national surveys (KNHANES, NHANES, CHNS). Harmonizes variables, runs parallel weighted analyses and builds comparison tables for 2-country (KR+US) or 3-country (KR+US+CN) designs.

• define-variables

  Use when exposure, outcome, covariate or eligibility definitions and cutoffs need a citable basis before the protocol. Reads the data dictionary first, then maps each variable to a guideline or published definition and the database columns in a citation-backed table.

• deidentify

  Use when clinical data may contain PHI and must be de-identified before any LLM-assisted analysis. A local Python script (no network or AI calls) detects identifiers with regex and heuristics in 12 country locale packs, with interactive terminal review.

• design-ai-benchmarking

  Use when designing a study that benchmarks AI systems against a human-expert panel, before data collection. Plans the arms, decoupled rubrics with anchors, planted calibration probes, reviewer panel, inter-rater reliability targets and LLM-as-judge versus human adjudication.

• design-study

  Use when checking a radiology or medical AI study design before drafting or submission. Reviews the analysis unit, cohort logic, leakage risks, comparator, validation strategy and reporting-guideline fit. AI-vs-expert benchmarks are /design-ai-benchmarking.

• explainability

  Renamed to /model-assessment in v6 (removed in v7).

• fill-icmje-coi

  Use when each author needs an ICMJE Conflict of Interest disclosure form (coi_disclosure.docx) for submission. Clones a pre-filled synthetic seed with every item marked None and replaces only date, name and manuscript title, so most authors just confirm or amend.

• fill-protocol

  Use when an institutional Word form (.doc/.docx IRB protocol, ethics application, grant template) must be filled without breaking its styles, tables, fonts or page layout. Renders content drafted by /write-protocol into the template; CJK-aware.

• find-cohort-gap

  Use when looking for research topics a longitudinal cohort database can answer (NHIS, UK Biobank, an institutional EMR or registry). Profiles the cohort, matches PI expertise, scans literature saturation and returns ranked topic proposals with gap evidence.

• find-journal

  Use when choosing where to submit a manuscript. Matches the abstract against curated journal profiles and returns ranked picks with scope fit, AI-disclosure policy, an acceptance-readiness pre-flight and a reject-fallback cascade. Impact and APC figures may be stale.

• fulltext-retrieval

  Use when you need full-text PDFs for a list of DOIs, such as a meta-analysis screening set. Batch-downloads open-access copies via Unpaywall, PMC, OpenAlex and Crossref, lists paywalled papers for manual access, and can convert PDFs to Markdown.

• generate-codebook

  Use when a tabular dataset (CSV, Excel, Parquet, Stata, SAS) needs a data dictionary. Profiles every variable (type, levels, range, missingness) into codebook.md and codebook.json and flags coded values of unknown meaning as [NEEDS DICTIONARY] instead of guessing.

• grant-builder

  Use when drafting a grant or challenge proposal for a radiology or medical AI project, including a Korean government industry-academia plan. Structures significance, innovation, approach, aims, milestones and consortium roles, keeping claims evidence-based and executable.

• humanize

  Use when a manuscript or response-to-reviewers letter reads as AI-written. Scans for 27 AI writing patterns and rewrites flagged passages, preserving technical accuracy and bounding how much text changes. Not general copy-editing; that is /polish-language.

• imaging-data

  Use when preparing a medical-imaging dataset (DICOM/NIfTI) for modelling. Profiles spacing, orientation, intensity, label integrity, foreground fraction and target volume, gates them against the plan, then plans and audits preprocessing and augmentation for leakage.

• intake-project

  Use when starting or inheriting a radiology research project and need to know what it is. Classifies the project type, summarizes its current state, lists missing inputs, recommends next steps and scaffolds lightweight project memory files.

• lit-sync

  Use when references in a .bib file (often from /search-lit) should land in Zotero and Obsidian. Syncs them to the Zotero library, writes Obsidian literature notes and extracts cross-cutting concept notes once enough accumulate. A folder of PDFs is /obsidian-paper-vault.

• ma-scout

  Use when looking for a meta-analysis topic before any protocol exists. Starts from a professor's publication profile or from a clinical question, finds gaps, assesses feasibility and returns a ranked topic list. Running the review itself is /meta-analysis.

• make-figures

  Use when a paper needs publication-ready figures or a visual abstract. Makes ROC, forest, calibration, Kaplan-Meier and Bland-Altman plots, CONSORT/STARD/PRISMA flow diagrams, confusion matrices, pipeline diagrams and journal visual abstracts, checking the journal's AI-image policy first.

• manage-project

  Use when managing a manuscript project over time. Scaffolds the project structure, tracks writing progress across phases, maintains project memory files, and generates submission checklists and backwards timelines (commands init, status, sync-memory, checklist, timeline).

• manage-refs

  Use when references must be written, rendered or converted. Checks [@key] citation keys, renders the reference list with a journal CSL via pandoc, converts [N] markers, injects Zotero Word field codes and runs manuscript-DOCX cross-reference QC. Read-only auditing is /verify-refs.

• meta-analysis

  Use when running a systematic review and meta-analysis, DTA or intervention. Covers PROSPERO protocol, search, screening, extraction, risk of bias (QUADAS-3, RoB 2, ROBINS-I), bivariate/HSROC or random-effects pooling, forest plots, heterogeneity and PRISMA reporting. Topic scouting is /ma-scout.

• mllm-eval

  Use when designing or auditing how an LLM or multimodal LLM is evaluated on a clinical task (report generation, VQA, text extraction). Covers reference standard, clinical-efficacy metrics beyond BLEU/ROUGE, hallucination, contamination and prompt sensitivity. Imaging models are /model-assessment.

• model-assessment

  Use when validating or evaluating a trained medical-imaging model. Audits split leakage and validation design, computes task-correct held-out metrics (Dice + HD95, AUROC + AUPRC, FROC, calibration), and covers uncertainty/OOD and Grad-CAM explainability, each with a gate.

• model-card

  Use when a trained medical-imaging model needs its documentation. Fills a Model Card and a Datasheet for its dataset from facts you supply, adds a METRIC-informed data-quality pass and gates that no required section is empty. Never invents numbers, provenance or licence.

• model-evaluation

  Renamed to /model-assessment in v6 (removed in v7).

• model-scaffold

  Use when you need a runnable PyTorch training repo for a medical-imaging task (segmentation, classification, detection, synthesis, self-supervised, or fine-tuning a pretrained backbone). Emits a patient-level seed-locked split, train/evaluate scripts, config and a Methods stub.

• model-selection

  Use when choosing the model for a medical-imaging study. Picks a paper-grounded architecture family (CNN/ViT, U-Net/nnU-Net, detection, SAM/foundation, GNN), then vets the concrete repo or checkpoint for licence, version pin, weight provenance and benchmark overlap.

• model-sourcing

  Renamed to /model-selection in v6 (removed in v7).

• model-validation

  Renamed to /model-assessment in v6 (removed in v7).

• obsidian-paper-vault

  Use when turning a folder of research PDFs into Obsidian notes, even if Obsidian is not named. Writes one templated literature note per paper and extracts cross-linked atomic concept notes, never overwriting existing ones. A .bib or Zotero library is /lit-sync.

• orchestrate

  Use when the user describes a research goal without naming a skill, or the task spans several skills. Classifies the request, plans the order and routes to the right medsci-skills skill(s) instead of producing their output itself.

• peer-review

  Use when reviewing someone else's manuscript for a journal, such as after a review invitation or for a revised R1/R2 version. Drafts a structured, constructive review in the journal's format. Never for your own manuscript; that is /self-review.

• polish-language

  Use when a manuscript needs a copy-edit for consistency and non-native English clarity. Flags abbreviation, US/UK spelling, en-dash range, P/p, hyphenation, number-style and unit-spacing issues, then polishes style only. AI-tell removal is /humanize.

• preprocess-imaging

  Renamed to /imaging-data in v6 (removed in v7).

• present-paper

  Use when preparing an academic talk such as a journal club, grand rounds, seminar, conference presentation, or lecture/teaching deck. Analyzes the source, drafts audience-adapted speaker scripts, builds or augments the PPTX with speaker notes and prepares Q&A.

• profile-imaging

  Renamed to /imaging-data in v6 (removed in v7).

• publish-skill

  Use when turning a personal agent skill into an open-source, distributable one. Runs a PII audit, generalizes personal details, checks licence compatibility, reviews cross-platform adapters and walks through packaging.

• radiomics-ml

  Use when building or auditing a radiomics or tabular clinical-ML prediction model with a classical learner (LASSO, SVM, random forest, XGBoost and similar). Enforces nested CV, dimensionality control, in-fold feature selection, feature stability, calibration and external validation.

• render-pdf-doc

  Use when rendering a Markdown document (English or Korean) such as a proposal, IRB cover letter, handout or reference table to PDF via pandoc and xelatex, with auto-fitted table widths and CJK fonts. Not for manuscripts with a bibliography (/manage-refs) or Word forms.

• replicate-study

  Use when applying a published cohort study's methodology to a different database. Extracts the design from the source paper, maps variables to the target database with a harmonization table, generates the analysis code and reports every forced deviation.

• review-paper

  Use when writing a literature review article (narrative, scoping PRISMA-ScR or systematic). Scaffolds a 7-part skeleton with a scope and non-overlap statement, summary-table stubs and reporting-guideline wiring, without inventing citations. Original research is /write-paper.

• revise

  Use when a manuscript comes back with reviewer or editor comments. Numbers every comment, classifies it MAJOR/MINOR/REBUTTAL, drafts a point-by-point response with tracked manuscript changes, routes new analyses to /analyze-stats and writes the cover letter.

• search-lit

  Use when finding papers or building a reference list. Searches PubMed, Semantic Scholar and bioRxiv/medRxiv, includes only references verified through an API, and generates BibTeX. Auditing an existing reference list is /verify-refs.

• self-review

  Use when checking your own manuscript before submission from a reviewer's perspective. Returns anticipated Major/Minor comments with fixes, including numerical, citation and leakage checks, with an optional multi-reviewer panel. Someone else's paper is /peer-review.

• setup-medsci

  Use when a skill fails for a missing tool or the environment needs checking. Diagnoses Python, R, Node, Claude Code, Git, Zotero and configured MCP servers and prints a pass/fail table with links to the setup docs. Read-only; installs nothing.

• sync-submission

  Use when building, auditing or freezing a journal submission package from the canonical manuscript. Detects drift between the source and the per-journal submission copy, builds byte-preserving packages with manifests, and records current, stale or frozen status.

• uncertainty-imaging

  Renamed to /model-assessment in v6 (removed in v7).

• verify-refs

  Use when checking whether a manuscript's references are real. Audits each citation against PubMed and CrossRef, flags fabricated or mismatched entries and writes qc/reference_audit.json. Audit-only; never edits references or refs.bib. Citation-key checks are /manage-refs.

• version-dataset

  Use when you must prove an analysis ran on the intended data or lock a dataset version. Builds a deterministic content-hash manifest (file SHA-256, schema, per-column value hashes), verifies later copies against it for drift, and diffs two manifests.

• write-paper

  Use when drafting a medical research manuscript or any IMRAD section. Runs an 8-phase pipeline from outline to submission-ready draft for original articles, AI validation studies, case reports, meta-analyses, technical notes and more. Checking a draft is /self-review.

• write-protocol

  Use when drafting an IRB or ethics research protocol. Writes Background, Study Design, Sample Size and Statistical Plan in full prose and leaves institution-specific sections as TODO skeletons. Filling an institutional Word form is /fill-protocol.

Similar plugins