Challenging the Status Quo: A Biomarker Use Case
Traditional keyword searching is highly inefficient, subject to bias and is not always comprehensive, providing limited potential for knowledge discovery and hypothesis generation. This selective approach introduces a bias towards familiar areas of expertise, which can lead to missed opportunities for novel insights and innovations. This is where AI comes in.
Reviewing The Literature: Selection Bias
With a 9% annual surge in scientific publications,¹ researchers are faced with the daunting task of sifting through thousands of papers to get answers. Due to project timeline demands, researchers often resort to reading just titles and abstracts to identify relevant studies.² This selective approach introduces a bias towards familiar areas of expertise, which can lead to missed opportunities for novel insights and innovations.³ This is where AI comes in.
Remove Bias with Causaly Cloud
AI excels in its ability to quickly analyze large datasets, going beyond just titles and abstracts and far exceeding human capacity. This allows researchers to ask questions and get to answers quicker. Powered by best-in-class, proprietary biomedical relationship extraction, Causaly is at the forefront of this innovation. Causaly’s human-centric AI machine-reads all available literature, rapidly extracting relationships to uncover hidden connections which might otherwise be missed. This allows all evidence to be considered by the user without applying selection bias or causing reading fatigue. Here, we demonstrate how Causaly’s differentiated AI can be used to accelerate biomarker discovery.
Biomarker Discovery: Causaly vs. PubMed
As an example, we have compared the utility of Causaly Cloud vs. PubMed for biomarker discovery in the context of ovarian cancer, as illustrated in Figure 1.
All Biomarkers: Using Causaly Cloud, evidence for 6,900+ biomarkers of ovarian cancer (supported by almost 29,000 documents) were extracted. Performing the same search in PubMed (keywords “biomarkers” “ovarian cancer”) returned a list of 22,000+ papers rather than insights.
Disease Progression: With Causaly, 3,600+ biomarkers of ovarian cancer progression were uncovered. Remarkably, PubMed returned 71% fewer results compared to Causaly, demonstrating how keyword searching is not comprehensive.
Biomarkers in 2023-2024: 800+ biomarkers of disease progression reported in 2023-2024 publications were identified from ~700 supporting publications. Performing the same query in PubMed returned just 90 publications, underscoring the risk of missed opportunities.
Traditional keyword searching is highly inefficient, subject to bias and is not always comprehensive, providing limited potential for knowledge discovery and hypothesis generation. With 1 million new research publications added to PubMed each year,⁴ extracting meaningful molecular signatures is extremely challenging.
With human-centric AI, scientists can cut through the noise, extracting relevant biological insights from comprehensive data sources within seconds, increasing productivity by up to 90%. By machine-reading millions of publications, Causaly can accelerate biomarker discovery, enabling early drug discovery teams to transition discovery projects into clinical programs with confidence.
Want to find out more?
More on Biomarkers