2025

A foundation model for human-AI collaboration in medical literature mining

Wang, Zifeng, Cao, Lang, Jin, Qiao et al.

Understand

Systematic literature review is essential for evidence-based medicine, requiring comprehensive analysis of clinical trial publications.

However, the application of artificial intelligence (AI) models for medical literature mining has been limited by insufficient training and evaluation across broad therapeutic areas and diverse tasks.
Here, we present LEADS, an AI foundation model for study search, screening, and data extraction from medical literature.
The model is trained on 633,759 instruction data points in LEADSInstruct, curated from 21,335 systematic reviews, 453,625 clinical trial publications, and 27,015 clinical trial registries.

Built on

Transformers: State-of-the-art natural language processing
Original
Wolf, T. et al · 1910
Earlier work this paper cites.
Zero: Memory optimizations toward training trillion parameter models
Original
Rajbhandari, S., Rasley, J., Ruwase, O. & He, Y · 1910
Earlier work this paper cites.
Pytorch: An imperative style, high-performance deep learning library
Original
Paszke, A. et al · 1912
Earlier work this paper cites.
Literature mining for the biologist: from information retrieval to biological discovery
Jensen, L. J., Saric, J. & Bork, P · 2006
Earlier work this paper cites.
How to do a meta-analysis
Field, A. P. & Gillett, R · 2010
Earlier work this paper cites.
Literature mining for the discovery of hidden connections between drugs, genes and diseases
Frijters, R. et al · 2010
Earlier work this paper cites.
Exact: automatic extraction of clinical trial characteristics from journal publications
Kiritchenko, S., De Bruijn, B., Carini, S., Martin, J. & Sim, I · 2010
Earlier work this paper cites.
The mass production of redundant, misleading, and conflicted systematic reviews and meta-analyses
Ioannidis, J. P · 2016
Earlier work this paper cites.
Analysis of the time and workers needed to conduct systematic reviews of medical interventions using data from the prospero registry
Borah, R., Brown, A. W., Capers, P. L. & Kaiser, K. A · 2017
Earlier work this paper cites.
The significant cost of systematic reviews and meta-analyses: a call for greater involvement of machine learning to assess the promise of clinical trials
Michelson, M. & Reuter, K · 2019
Earlier work this paper cites.
Decoupled weight decay regularization
Original
Loshchilov, I. & Hutter, F · 2019
Earlier work this paper cites.
Language models are few-shot learners
Brown, T. et al · 2020
Earlier work this paper cites.
Retrieval-augmented generation for knowledge-intensive nlp tasks
Lewis, P. et al · 2020
Earlier work this paper cites.
How the fda regulates ai
Harvey, H. B. & Gowda, V · 2020
Earlier work this paper cites.
Decision makers need constantly updated evidence synthesis
Elliott, J. et al · 2021
Earlier work this paper cites.
Most cochrane reviews have not been updated for more than 5 years
Hoffmeyer, B. D., Andersen, M. Z., Fonnes, S. & Rosenberg, J · 2021
Earlier work this paper cites.
Generating (factual?) narrative summaries of RCTs: Experiments with neural multi-document summarization
Wallace, B. C., Saha, S., Soboczenski, F. & Marshall, I. J · 2021
Earlier work this paper cites.
The prisma 2020 statement: an updated guideline for reporting systematic reviews
Page, M. J. et al · 2021
Earlier work this paper cites.
Msˆ2: Multi-document summarization of medical studies
DeYoung, J., Beltagy, I., van Zuylen, M., Kuehl, B. & Wang, L. L · 2021
Earlier work this paper cites.
Chain-of-thought prompting elicits reasoning in large language models
Wei, J. et al · 2022
Earlier work this paper cites.
What ai in health care can learn from the long road to autonomous vehicles
Norden, J. G. & Shah, N. R · 2022
Earlier work this paper cites.

Similar

Finetuned language models are zero-shot learners
Wei, J. et al · 2022
Cited alongside, same era.
Flashattention: Fast and memory-efficient exact attention with io-awareness
Original
Dao, T., Fu, D. Y., Ermon, S., Rudra, A. & Ré, C · 2022
Cited alongside, same era.
The problems with systematic reviews: a living systematic review
Uttley, L. et al · 2023
Cited alongside, same era.
Can chatgpt write a good boolean query for systematic review literature search?
Wang, S., Scells, H., Koopman, B. & Zuccon, G · 2023
Cited alongside, same era.
Assessing the ability of chatgpt to screen articles for systematic reviews
Original
Syriani, E., David, I. & Kumar, G · 2023
Cited alongside, same era.
Medcpt: Contrastive pre-trained transformers with large-scale pubmed search logs for zero-shot biomedical information retrieval
Jin, Q. et al · 2023
Cited alongside, same era.
Jointly extracting interventions, outcomes, and findings from rct reports with llms
Wadhwa, S., DeYoung, J., Nye, B., Amir, S. & Wallace, B. C · 2023
Cited alongside, same era.
Summarizing, simplifying, and synthesizing medical evidence using gpt-3 (with varying success)
Shaib, C. et al · 2023
Cited alongside, same era.
Ai-generated text may have a role in evidence-based medicine
Peng, Y., Rousseau, J. F., Shortliffe, E. H. & Weng, C · 2023
Cited alongside, same era.
Foundation models for generalist medical artificial intelligence
Moor, M. et al · 2023
Cited alongside, same era.
Appraising the potential uses and harms of llms for medical systematic reviews
Yun, H., Marshall, I., Trikalinos, T. & Wallace, B. C · 2023
Cited alongside, same era.
Mistral 7b (2023)
Original
Jiang, A. Q. et al · 2023
Cited alongside, same era.

Then

Closing the gap between open source and commercial large language models for medical evidence summarization
Zhang, G. et al · 2024
Later among the works it cites.
Accelerating clinical evidence synthesis with large language models
Original
Wang, Z. et al · 2024
Later among the works it cites.
ClinicalBench: Can llms beat traditional ml models in clinical prediction?
Original
Chen, C. et al · 2024
Later among the works it cites.
High-performance automated abstract screening with large language model ensembles
Original
Sanghera, R. et al · 2024
Later among the works it cites.
Performance of a large language model in screening citations
Oami, T., Okada, Y. & Nakada, T.-a · 2024
Later among the works it cites.
CSMed: Bridging the dataset gap in automated citation screening for systematic literature reviews
Kusa, W., E Mendoza, O., Samwald, M., Knoth, P. & Hanbury, A · 2024
Later among the works it cites.
https://openai.com/index/hello-gpt-4o/
Hello gpt-4o · 2024
Later among the works it cites.
https://openai.com/index/chatgpt/
Introducing chatgpt · 2024
Later among the works it cites.
https://www.anthropic.com/news/claude-3-haiku
Claude 3 haiku: our fastest model yet · 2024
Later among the works it cites.
The llama 3 herd of models (2024)
Original
Dubey, A. et al · 2024
Later among the works it cites.
Biomistral: A collection of open-source pretrained large language models for medical domains (2024)
Original
Labrak, Y. et al · 2024
Later among the works it cites.
https://openai.com/index/new-embedding-models-and-api-updates/
New embedding models and api updates · 2024
Later among the works it cites.
Data extraction for evidence synthesis using a large language model: A proof-of-concept study
Gartlehner, G. et al · 2024
Later among the works it cites.
Can large language models replace humans in systematic reviews? evaluating gpt-4’s efficacy in screening and extracting data from peer-reviewed and grey literature in multiple languages
Khraisha, Q., Put, S., Kappenberg, J., Warraitch, A. & Hadfield, K · 2024
Later among the works it cites.
https://www.ncbi.nlm.nih.gov/home/develop/api/
Ncbi provides several public apis that allow programmatic access to many databases and tools · 2024
Later among the works it cites.
https://clinicaltrials.gov/data-api/api
Clinicaltrials.gov api · 2024
Later among the works it cites.
Panacea: A foundation model for clinical trial search, summarization, design, and recruitment
Lin, J., Xu, H., Wang, Z., Wang, S. & Sun, J · 2024
Later among the works it cites.
A generalist vision–language foundation model for diverse biomedical tasks
Zhang, K. et al · 2024
Later among the works it cites.
A foundation model for joint segmentation, detection and recognition of biomedical objects across nine modalities
Zhao, T. et al · 2024
Later among the works it cites.
Llava-Med: Training a large language-and-vision assistant for biomedicine in one day
Li, C. et al · 2024
Later among the works it cites.

Beyond the bibliography

alphaXiv searches the wider corpus for related work and actual follow-ups.

Open on alphaXiv

alphaXiv is searching for related work…

Reading the bibliography…