Medical Literature Review Assistant

Problem to Solve

The hard part of a medical literature review is not reading individual papers, but keeping the search reproducible, the screening process complete, the quality appraisal consistent, and the conclusions aligned with evidence strength. This skill frames the task as evidence-based writing assistance rather than diagnosis or treatment advice.

Core Workflow

It follows a structured review path. First, define the question using PICO: population, intervention, comparison, and outcome. Then build the search strategy from MeSH terms, free-text terms, Boolean operators, and filters, while recording databases, search date, syntax, and limits. Screening proceeds through deduplication, title/abstract review, and full-text review, with reasons for exclusions preserved. Extraction focuses on study design, sample size, effect measures such as RR, OR, HR, MD + 95%CI, and risk of bias. Quality appraisal uses tools matched to study type, such as Cochrane RoB, NOS, QUADAS-2, or AMSTAR 2. Synthesis is labeled with GRADE, and the discussion covers publication bias, heterogeneity, and clinical applicability.

Scope and Limits

Use it for drafting reviews, structuring inclusion tables, documenting search strategies, and explaining evidence quality. It should not be used for patient diagnosis, treatment plans, medication recommendations, or single-case advice. By default it emphasizes recent literature and covers at least PubMed and Cochrane Library; extend the scope explicitly when older evidence, rare populations, or specialty databases are required.