Embedded Files
90% of drugs fail in human trials, with 30% of them because of unforeseen side effects. This is an incredibly costly endeavor, since big pharma reports 1 failed drug costs $1 billion. We believe that through more accurate computational models of disease we can pick better drug targets by predicting their side effects early. This would then increase the success rate of all the following stages in the drug development pipeline.
Limitations of existing methods
Limitations of existing methods
Genome-scale metabolic models
Limited causal discovery since they’re based on knowledge bases of isolated interactions
Limited by the steady state assumption (stable flux of 0) which cannot capture disease progression from a healthy state since it involves flux changing across timesteps
Equations derived by hand
As Costello 2018 showed, they have very poor modeling accuracy
Neural Differential Equations for Metabolic Simulations
Neural Differential Equations for Metabolic Simulations
Toward the vision of accurate computational models for human biology, we developed NDEs for metabolic simulations (preprint), achieving 100x compute speedup and 90% accuracy improvement over existing methods.
In the plots above, we see our system's performance (orange) on an open-source timeseries dataset of metabolic function (blue), achieving dramatically better results that other available methods (green).
Page updated
Google Sites
Report abuse