Dr Aidan Kubeyev is a Computational Scientist at Concr, responsible for developing and applying machine learning modelling to cancer research.
“When will there be a cure for cancer?” is a question that I’m frequently asked when I say that I work at an oncology techbio company.
As a computational scientist who never worked in oncology prior to joining Concr, I was surprised by how much cancer research is trial-and-error. Matching the right patient to the right treatment (precision oncology) is severely limited by the process of drug development, where novel drugs fail to demonstrate sufficient response rates in trials. For example, IQVIA estimated drug success rates in oncology to be only 5%.
Simply put, biology is so complex that we don’t fully understand it. By and large, cancer scientists perform experiments on isolated cells rather than ecosystems of tissues. Machine learning (ML) using diverse data ranging from genetics to patient clinical data is not only logical but a necessary approach to improving clinical drug development and patient treatment outcomes by guiding the selection of correct therapy.
This is our primary focus at Concr. Instead of a one‐drug‐fits‐all model, we adopt an approach that is tailored to specific patient features like age, hormone status, cancer stage etc. Using our unique ML modelling adopted from cosmology, we integrate complex biological and clinical data into a single unified model where individual data parcels are interdependent with one another, as they are in a human body.
In a way, we’re mimicking complex pathophysiology using relevant information: genomic, transcriptomic, digital pathology, drug response and clinical data. Our partners can subsequently ask this artificial system relevant biological questions, which could be answered at a cohort or individual level, making it suitable for predicting patient drug response. For example, “what will be my drug’s therapeutic efficacy in publically-available cell lines?” or “what will be the most predictive biomarker signature for patient responders?”.
We’re always looking to make the most of existing cancer research data, so discover more about how our modelling capabilities can assist you with your research questions.
Back
LinkTree