On 13 December 2022, Hein van der Wall successfully defended his PhD thesis titled ‘Exploring machine learning techniques in the context of early-stage clinical research’ at Leiden University.
In drug development and clinical trials, biomarkers may be used to help identify populations for a study, monitor therapeutic response, and identify side effects. Although an increasing number of analyses is performed, it is not always clear 1) how to handle the collected data, and 2) whether all these analyses are useful to study. Hein hypothesized that a good way to use these data is to employ a machine learning algorithm. The models built based on the data could then serve as new biomarkers to recognize the intended and unintended effects of (new) drugs. Also, after using a model, its predictions could be explained. It may lead to a better understanding of how the various features used in the algorithm influence the cause and effect of a specific drug. With the help of machine learning, the data collected at an early stage of clinical drug discovery is optimally exploited.
In his PhD thesis, Hein applied machine learning techniques for data analysis on a variety of large data sets obtained in early-stage clinical research projects. This includes classical data consisting of electrical signals from the ECG of healthy subjects, innovative data originating from measurements in a driving simulator, and emerging data derived from DNA analysis of the microorganisms living on the skin of patients with skin disease. He showed that machine learning can be applied on these types of data to detect and evaluate the effect of drugs and other interventions. Read Hein’s thesis here.
Hein was supervised by promotors Prof. dr. J. Burggraaf and Prof. dr. G.J.P. van Westen and co-promotor Dr. R.J. Doll. His research was performed at CHDR and the division Drug Discovery and Safety of the Leiden Academic Centre for Drug Research (LACDR), Leiden University.