The therapeutic drug development process is long and risky. Before a new drug can be used in humans it has to go through a series of development steps to ensure that it is safe and effective. This process begins with exploratory studies that typically involve conducting experiments using cell lines cultured in vitro (out of the body) and in vivo (in the body) in model organisms (often mice and rats) to determine the drug’s biological mode of action.
If successful in these early exploratory phases the drug then enters a clinical phase, in which it is determined if the biology learnt from the pre-clinical stages of research is applicable to humans via carefully regulated clinical trials.
Hypothesis: Machine learning can be used to determine how information passes from one phase of the biomedical research pipeline to the next – and when it does not – and thereby significantly improve the discovery process with substantial economic and healthcare benefits.
At this early stage we will focus on the problem of mapping biology from mouse to man in order to determine the circumstances under which experiments in mice are likely to provide information that is clinically relevant. This is a key bottleneck in the translational process and constitutes a problem that we can immediately address within the scope of this pilot project. Specifically, we will:
The first two of these aims are central to our understanding of complex life, and therefore of significant interest in their own right, aside from their translational significance. The third aim may be considered as a practical corollary of the first two, with direct relevance to the drug discovery process. This case study can be viewed as the first step toward a general research programme that seeks to develop ML tools specifically designed to effectively pass information through the biomedical research process which we will develop in the coming years.
Principal Investigator: Dr Ben MacArthur