Challenging uncertainty in the pharmaceutical industry through machine learning
D.Predict overcomes the limitations of the traditional, rather heuristic, prediction methods through application of multiple machine learning algorithms.
Value of pharmaceutical companies are sensitive to expected value of their drug pipelines. Thus, precise drug valuation must come first, to reduce volatility and uncertainty in investment decisions.
Prediction of clinical trial success rate is a key variable in pipeline drug valuation. As can be imagined, these ‘predictions’ are complex and resource-consuming.
This is possibly why the process itself has remained in the ‘realm of art’ at best.
What is D.Predict?
D.Predict is a machine learning model that predicts success rate of clinical trials. It enables the user to simply enter clinical trial data on a web-based platform of the model and in return, find out a prediction of clinical trial results before actually conducting the trials.
D.Predict is for…
- Clients facing R&D investment for pipeline drugs
- Clients wishing to reevaluate the variables for clinical trial design
- Clients in need of rationale for R&D pipeline prioritization
How does D.Predict work?
- Multiple algorithms
D.Predict utilizes 6 different machine learning algorithms including the well-known 'deep learning' algorithm. Different algorithms would process the input data in order to pinpoint the most accurate results.
- Variable data
Comprehensive set of variables related to clinical trials are input into the model such as data on; drug indication, therapy area, molecule type and trial participants information.
- Continuous elaboration
The model is repeatedly trained in order to optimize output data accuracy. This is executed through constant reflection of results and adjusting of parameters.