AISysRev - LLM-based Tool for Title-abstract Screening

Описание: Systematic reviews are critical in fields like medicine, social sciences, and software engineering—but going through hundreds of titles and abstracts is slow and exhausting. AISysRev helps researchers cut down this effort.

With AISysRev, you upload a CSV file of titles and abstracts (for example exported from Scopus), set your inclusion and exclusion criteria, and run the screening with large language models through OpenRouter. The tool supports both zero-shot and few-shot screening, and combines AI suggestions with manual review so that researchers stay in control. For best results, we recommend using multiple LLMs through OpenRouter, since this increases reliability in the screening outcomes.

In our trial with 137 papers, we observed that studies often fell into three groups: Easy Includes, Easy Excludes, and more challenging Boundary Cases where human judgment is still essential. This suggests that AISysRev can speed up reviews without replacing the critical role of researchers.

You can run the tool in Docker, try multiple LLMs, and decide how much to rely on AI support.

🔗 Tool : https://github.com/EvoTestOps/AISysRev
🔗 Pre-print Paper: https://arxiv.org/abs/2510.06708
🔗 OpenRouter: https://openrouter.ai/
🔗 Scopus: https://www.scopus.com/