Skip to Content

The Causal AI conference is back in San Francisco for 2024, bigger and better than ever.

Register Interest
  • Blog

causaLens & MayoClinic discover causal biomarkers of cancer

UK pioneer of next-generation Artificial Intelligence helps speed up early detection using causality

causaLens, the London deep tech company delivering the future of AI, has utilized its Causal AI technology in an important step towards adopting non-invasive methods for screening and understanding cancer. The research demonstrates that Causal AI enables reliable early detection of cancers based on easy-to-administer and cost-effective blood tests.

Using a strictly data-driven approach, causaLens, which this year received investment co-led by Dorilton Capital and Molten Ventures, discovered possible causal connections between the expression of genes and proteins and the manifestation and stage of colorectal cancer. 

Colorectal cancer is the fourth most common cancer – with 147,950 cases annually – and one of the most lethal cancers – with 53,000 deaths per year.

At present, the only screening method used for colorectal cancer is colonoscopy, to identify and remove precursor lesions. However, this procedure is expensive and invasive, and hence has limited reach for a type of cancer that is increasing by 2% a year among the under 50s, meaning that ever-shorter screening intervals are needed to prevent disease in higher-risk patients.

For these reasons, there is an urgent clinical need for techniques that can reliably detect early-stage colorectal cancers via non-invasive methods. The potential impact of causaLens’ research — undertaken in collaboration with academics from the Mayo Clinic, a prominent US medical research center — would be a significant improvement in the long-term prospects of patients with colorectal cancer.

“Causal AI is a fundamental scientific breakthrough and causaLens’ vision for Causal AI extends far beyond enterprise decision making. causaLens has the potential to disrupt a vast range of sectors and industries and has already demonstrated the value of its Causal AI technology in biological applications such as the discovery of cancer biomarkers”

-Nicholas Chia PhD, Mayo Clinic, Assistant Professor of Biophysics,

causaLens is the pioneer of Causal AI — the only AI technology that is capable of discovering and reasoning with cause-and-effect relationships. Causal AI can root out spurious correlations — mathematical relationships in which two or more variables are associated but not causally related, due to either coincidence or the presence of a third, unseen factor.

The company has enjoyed notable success in delivering transformational outcomes across a range of sectors, from improving investment portfolio returns to helping mitigate supply chain disruptions and navigating public health crises.

“We are excited to be leading the effort to apply Causal AI techniques to biological data, and specifically to cancer research. Our vision is to create a world in which humans can trust machines to meet the greatest technical and societal challenges, such as those posed by cancer data.”

Hana Chockler PhD, causaLens, Principal Investigator

causaLens and cancer research

causaLens obtained data from The Cancer Genome Atlas (TCGA) project, specifically the Gene Expression by RNAseq (IlluminaHiSeq) data and RPPA Protein Expression data from the Colon and Rectal Cancer (COADREAD) cohort. Preliminary findings supported the current understanding of the role of different RNA expressions, supporting causaLens’ hypothesis that a screening based on the blood test results is meaningful.  https://www.researchsquare.com/article/rs-967255/v1

The findings also highlighted challenges: the high number of candidate genes and proteins, compared to a much lower number of patient records; the extreme imbalance of the dataset, as it does not contain any healthy individuals; and the inability to intervene on the dataset. It was also deduced that the analysis of bimodal activations – a common practice in analysing the data – is more reliable when datasets are larger. Datasets may reduce in size when a sizeable subset of records is discarded (those where it cannot be deduced with any certainty whether the RNA expression is on or off).

Future research by causaLens will focus on overcoming these theoretical challenges, as well as the practical analysis of larger datasets.