MetaScreener

MetaScreener is a web tool that uses large language models to screen literature for systematic reviews and meta analyses. A researcher defines inclusion criteria, loads a set of references, and the tool applies those criteria across the abstracts or full texts, producing include or exclude decisions that the researcher can review and override. It accepts RIS files exported from sources such as PubMed, Scopus and Web of Science, and PDF documents for full text screening, with OCR for scanned files. Users choose their own model provider from OpenAI, Anthropic, Google or DeepSeek and supply their own API key. The site states that processing happens in the browser with direct calls to the provider and that keys stay stored locally. Batch processing handles hundreds of documents with progress tracking. Performance analytics report confusion matrices, sensitivity, specificity and Cohen's Kappa, and results export as RIS, CSV, Excel or JSON. The tool is built for academic and clinical researchers running evidence reviews who want AI to do the first pass while a person makes the final call. No programming is required. MetaScreener itself is free, with no fees, commissions or revenue sharing. Users pay their chosen model provider directly. The site gives typical costs of about $0.50 to $2 for reviews up to 500 papers, $2 to $8 for 500 to 2,000 papers and $8 to $25 for larger reviews, depending on provider.