„AI (Artificial Intelligence) is the new electricity. Just as electricity transformed almost everything 100 years ago, today I actually have a hard time thinking of an industry that I don’t think AI will transform in the next several years.” – says Andrew Ng, professor at Stanford University, one of the most prominent AI professionals in the world. Mr. Ng is most probably right. In fact, AI may change our life quicker than electricity did before and there is no need for decades to enter into the age of AI. Machine Learning, the most used implementation of AI already takes part of almost every field of scientific or business research and development. Research and development is often based on enormous amount of data which is to be analysed, explored and understood. Machine Learning aided solutions can heavily support this work by providing a wide-spread toolkit to cope with the challenge of the data heavy scientific and business areas.
Pharmacology, the study of different drugs and medication actions, is also based on gathering, analysing and understanding data, coming from experiments or other use cases. Effect of a certain agent is hidden in complex, often non-linear relationships and could be found by looking for involute patterns among hundreds of features and millions of values. Machine Learning is a very promising tool to find and represent those relationships and patterns.
Gedeon Richter Plc has aimed at state-of-the art data-driven operations and decision-making processes for several years. The next milestone of this endeavour is that the company has initiated the application of AI in pharmacological research. As a first step, the analysis of data from the IntelliCage system used for animal experimentation based cognitive tests was analysed with involvement of Neuron Solutions Ltd. IntelliCage is an intelligent residential cage in which animal behaviour can be observed continuously without disturbing or in other way intervening in day-to-day behaviour of subjects (animals). The appliance collects research data 24 hours a day – seven day a week. As a result, we have tens of thousands of data records at our disposal as opposed to traditional behavioural experiments.
The purpose of the project was not only to further improve the evaluation of the given cognitive tests by machine learning, but also to transfer knowledge to the data science experts and pharmacology analysts of Richter.
In the first part of the project, data gathered during IntelliCage experiments were explored with various data science techniques which resulted in a cleaned, prepared and compact database for machine learning modelling.
In the next, i.e. modelling phase, several different machine learning techniques were investigated and tried. We have found that the modelling approach of neural networks based representation (self-supervised) learning was the most useful out of these techniques and lead us to the conclusion that we should continue to deal with these machine learning models for further evaluation of cognitive tests.
It is important to point out that the use of machine learning always implies experimenting and iteration, since we can hardly predict with certainty – based on merely theoretical knowledge and „desktop research” – which of the available machine learning techniques will produce the results we need.
Neural models helped us primarily characterize the effect of the tested substances and to rank these substances relative to each other, but lead us to create other important research results.
Last, but not least as mentioned above, we provided constant knowledge transfer to Richter’s project team consisting of researchers, analysts and data scientists to support them in further use of machine learning solutions in research or other enterprise operation areas.
We are proud to have completed already the second project with the excellent experts of Gedeon Richter Plc in this important field, which is also exciting from a professional point of view, and contributed to the increasingly wide dissemination and application of Artificial Intelligence.
Applying Artificial Intelligence in Richter’s Pharmacological Research
