Search papers, labs, and topics across Lattice.
AutoForest is an innovative end-to-end system designed to automate the generation of forest plots from biomedical studies, addressing the traditionally labor-intensive process of evidence synthesis. By leveraging large language models, it extracts key ICO (Intervention, Comparator, Outcome) elements, performs statistical synthesis, and produces publication-ready visualizations directly from unstructured clinical texts. A user study with clinicians demonstrated that AutoForest significantly accelerates the evidence synthesis process and reduces the expertise barrier for conducting meta-analyses.
AutoForest transforms the tedious process of generating forest plots into a seamless, automated workflow that can be executed with minimal expertise.
Systematic reviews rely on forest plots to synthesise quantitative evidence across biomedical studies, but generating them remains a fragmented and labour-intensive process. Researchers must interpret complex clinical texts, manually extract outcome data from trials, define appropriate interventions and comparators, harmonise inconsistent study designs, and carry out meta-analytic computations-typically using specialised software that demands structured inputs and domain expertise. While recent work has demonstrated that large language models can extract study-level data from unstructured text, no existing system automates the complete pipeline from raw documents to synthesised forest plots. To address this gap, we introduce AutoForest, the first end-to-end system that generates publication-ready forest plots directly from biomedical papers. Given one or more study papers, AutoForest automatically suggests ICO (Intervention, Comparator, Outcome) elements, extracts outcome data, performs statistical synthesis, and renders the final forest plot. We describe the system architecture, user interface and demonstrate its effectiveness on real-world examples through a user study involving clinicians, showing how AutoForest can accelerate evidence synthesis and substantially lower the barrier to conducting meta-analyses.
