Introducing Synthesa’s Research Series: Evidence Synthesis Powered by AI
At Synthesa, Inc., we are building the next generation of evidence synthesis tools, powered by large language models (LLMs). Our work is grounded in research that explores how LLMs, such as ChatGPT, can support and accelerate the systematic review process.
Does the long-term administration of proton pump inhibitors increase the risk of adverse cardiovascular outcomes? A ChatGPT powered umbrella review.
In our latest featured publication, published in Acta Cardiologica, our co-founder Lefteris Teperikidis and colleagues conducted a ChatGPT-powered umbrella review to investigate whether long-term proton pump inhibitor (PPI) use increases the risk of major adverse cardiovascular events (MACE).
The study not only evaluated a clinically important question, but also tested the ability of ChatGPT to assist with:
– Generating search strategies
– Screening studies based on titles and abstracts
– Extracting data using PICO criteria
– Drafting sections of the review manuscript
– Synthesizing and summarizing the results
The findings showed that ChatGPT was able to support many stages of the evidence synthesis process using the first LLM available by OpenAI (ChatGPT V.3.5). At the same time, certain tasks, such as risk of bias assessment, required human oversight — highlighting the value of combining artificial intelligence with expert clinical judgment.
This research is part of a growing body of work that demonstrates how AI can transform evidence synthesis from a time-intensive process into a faster, more adaptive, and scalable approach to generating evidence for clinical and regulatory decision-making.
The full study is available here: https://pubmed.ncbi.nlm.nih.gov/37431972
This post is the first in a series highlighting the research that informs our work at Synthesa. Stay tuned for more publications that demonstrate how we are advancing evidence synthesis through the combined power of AI and clinical expertise.
