During my PhD I have been working on interpretable supervised machine learning approaches to deal with real world medical data, with the final goal of having computer aided clinical decision support system for medical professionals , and knowledge extraction from provided datasets.
Here you can find my academic journey till now (my CV).
I am a IEEE graduate student member
This marks the start of my PhD journey in October 2016, under the joint supervision of Dr. Kerstin Bunte (daily supervisor, University of Groningen), Prof. Dr. Michael Biehl (PhD Promoter and 2nd supervisor, University of Groningen), Prof. Dr. Peter Tino (External supervisor, University of Birmingham) and Prof. Dr. Wiebke Arlt (External supervisor, University of Birmingham).
In 2018 there was a major change in the dataset I had started off with. The dataset became more interesting and challenging. This work is ongoing but here is a brief summary of the changes and how our classifiers performed on this updated dataset. This is the work I had presented as a poster at an event in January 2020.
Next, the classifiers which my team and I developed, were tested on other datasets. Following this a new variant of it was developed to investigate multi-class problems with non-linear decision boundaries. In this contribution to IJCNN 2020 (we have published a preprint of it at Arxiv.org) my co-authors and I also explained why we need XAI as increasing number of anthropocentric applications are becoming dependent on Machine learning.
Interdisciplinary aspect of my research
For the major part of my PhD research, my supervisors in Machine Learning (Dr. Kerstin Bunte, Prof. Dr. Michael Biehl and Prof. Dr. Peter Tino) and I have collaborated with Dr. Elizabeth Baranowski and Prof. Dr. Wiebke Arlt from IMSR, University of Birmingham. Together we use interpretable machine learning methods such as variants of Learning vector quantization (LVQ) to investigate rare steroidogenic disorders in babies and neomates. In the simplest terms this is what my research with the IMSR group looks like
My other medical collaborators are from the GRIAC group of the University Medical Center Groningen. With this group my supervisor and I use our developed algorithms to learn more about different stages of Asthma. I’ll make an equally ridiculous figure to describe my asthma research in the coming weeks.