90% of clinical trials fail - after consuming 1 Billion USD and 10 years of work. They fail because a) the wrong patients are selected or b) the drug target turns out to be not central to the disease mechanism.
Despite the availability of data from previous trials, the industry struggles to remedy the issue via biomarkers that ensure the right patient are selected, or identifying drug targets genetically linked to the core of the disease mechanism.
Why does pharma struggle? Biomedical data has a problematic structure - few patients and millions of features / gene-variants. Diseases are also not determined by single gene variants, but interactions of many genes - we end up with billions of features.
Current Machine Learning approaches don't adequately deal with such 'wide' data and fail to identify the desired polygenic biomarkers and genetically-linked drug targets.
Our Machine Learning techniques are specifically designed for high-dimensionality biomedical data and proven to find the complex interactive patterns hidden in such 'wide' data.
Using data from previous trials, we identify superior biomarkers for patient stratification, as well as drug targets genetically linked to the disease. We make sure that the addressed drug target is genetically verified and thus central to the disease mechanism and that the right patients are selected.
Our polygenic biomarkers predict disease status at a much higher accuracy than conventional methods, but do not cost more when integrated into a diagsnostics kit. Diagnostics companies that use our biomakers have the edge over competition.
Pharma use our biomarkers for patient stratification in clinical studies to make sure that their cases have the indication that their drug targets. Because of the high predictive accuracy of our biomarkers, they can be further developed into clinically-verified drug targets to boost success rates of drugs, or discover new therapies.
Our biomarkers can accurately predict what treatment works best for each individual patient. Collaborating doctors and clincs are improving their patients well-being and saving health insurance companies millions of dollars.
Oxford, UT Austin
During his graduate studies at Oxford and Texas, Jörn investigated human cognition in order to optimize artificial intelligence. Insights from hands-on research in an indigenous community, neural-networks and agent models led to an outside-the-box machine learning understanding. After rounding off his skills with commercial experience at a big data start-up, Jörn brought his data skills to biotx.ai.
FU Berlin, Cambridge
Marco is a biochemist with expertise in chemical synthesis and action of drug molecules. While deeply immersed in several experiments during his Marie Curie Fellowship at the University of Cambridge, he realized the great potential of using novel machine learning approaches in drug combinations. This experience led to the development of a companion diagnostic kit and the foundation of biotx.ai.
We'll expand our team by the end of the year. We're looking for the best, in terms of coding, business acumen, data skills and independent problem solving. High rewards. If you're interested, get in touch.