With millions of factors going into drug development in the pharmaceutical industry there is an ongoing challenge to make use of relevant large-scale data sources. To support its development Captario is collaborating with Fraunhofer-Chalmers Centre for Industrial Mathematics (FCC). Analysts from Captario will work together with scientists from FCC to identify value-adding areas for application of machine learning and AI. Suitable methods and algorithms will be developed and implemented as features in Captario SUM, enabling swifter and more reliable decision support for users of SUM. This marks the first step in a long-term ambition of Captario to develop methods and software features in this area, in our continuous aspiration to be a game-changer in strategic decision support for the pharmaceutical industry.
Captario SUM and quantitative decision support
Captario develops the software SUM to enable quantitative modelling and support for strategic decision making in the pharmaceutical industry. The starting point of modelling is a graphical representation of the drug development project. The figure below illustrates how such a process model might look like for an oncology project. The model contains key activities, e.g. clinical trials, (rectangles) and key decision points (diamonds). The modelling approach is very flexible and allows rapid response to a wide variety of questions (for further details on the general modelling approach, see reference )
Manual information gathering vs automated AI support
To be useful, the model as described in the previous section needs to be populated with assumptions about parameters that could have a significant impact on project value. This is a general concern in the pharmaceutical industry, and with thousands of factors impacting drug development the industry is struggling with getting data into their processes. The gathering of these assumptions is currently an entirely manual process, where individuals pursue a tedious search in various data sources for relevant information. Alternatively, data might be provided based on gut feeling, personal experience or guesswork.
We believe that in a digitalized era this could, and should, be changed. It is our ambition to build features that enables the user to type keywords describing the situation at hand, e.g. by describing disease area, indication, current phase of development, type of treatment, etc, and then ask the software for automated guidance regarding the parameter(s) of interest. The figure below illustrates some areas in which decision support input may be required, and various types of data sources that may be utilized. Suitable machine learning and AI methodologies and algorithms should be developed to provide model input via a user-friendly interface.
The FCC collaboration
Captario will now partner with skilled and experienced scientists from the Fraunhofer Chalmers Research Centre for Industrial Mathematics, FCC, for our next step in leading the innovation in this area. The first phase of the collaboration is funded by a research grant from the Gothenburg area county council (Västra Götalands-regionen). Analysts from Captario will work together with scientists from FCC to identify the areas in which machine learning and AI would add most value, and to develop suitable methods and algorithms to for its application. We are convinced that this will prove to be a fruitful collaboration, as the highly skilled and experienced scientists from FCC will provide a perfect match with the internal competencies and pharmaceutical industry knowledge at Captario.
The goal and the future
By the end of spring next year, our goal is to have a first version of the AI augmented decision support available in the Captario SUM software. But this only marks the first step in a long-term ambition of Captario to develop methods and software features in this area. We are set out to be a game-changer in strategic decision support for the pharmaceutical industry. The currently ongoing initiative on machine learning and AI is indeed an exciting step in this journey.
 Wiklund SJ (2019) A modelling framework for improved design and decision-making in drug development. PLOS ONE 14(8): e0220812. https://doi.org/10.1371/journal.pone.0220812