Find out how the Medical Dashboard came about!

Business Data Challenger Hetty Wessel is developing a special Medical Dashboard (medical image processing dashboard) together with colleague Bassam Shoukri and medical specialists. A project that has already been awarded an innovation subsidy. The question for Hetty Wessel: “Did you have a special motive for conducting research specifically with medical images? How did you come up with this specific idea?” “Of all the information that we process around us in the professional world, it is estimated that around 80% qualitative information. This is a frugal estimate. We see that data science courses mainly focus (in general) on quantitative information. This mainly builds on what we have been doing for decades: quantities, sizes, numbers, ratios, etc. This approach is often ideally suited to unlocking information for business operations, for example, but there is also an increasing need to also provide images and to be able to interpret natural (unstructured) texts. We can 'see' more and more through the development of sensors and medical equipment, for example. But can we as people get enough out of this? How do we ensure that the extra information helps with interpretations and explanations for what happened and how can we include that in the prediction of what may happen? A data scientist is not a programmer who knows a few standard models and knows how to produce nice visuals. A Business Data Challengers data scientist understands the data that needs to be worked with, understands the application, the context, can develop models based on that context and take the correct methodological steps to achieve good input for the models. I see good training in the field of text mining at linguistics faculties. And unfortunately, this is often looked down upon by the more science-oriented courses. We therefore see that data science is not really getting off the ground yet. There are too many one-trick ponies without a good methodological and statistical background. I spoke to someone like that a while ago. He had applied a model without understanding that it was not possible in that context. “Yes, something came out of it, right?” That's the thing: something always comes out…. but then there is the connection with practice and that is what matters. As a data scientist, I wanted to focus mainly on qualitative information, because that is where the challenging questions lie. And of course that also has many quantitative components. Especially in the medical world, you see a huge increase in technologies to 'see' the human body (less invasively). Developments in MRI equipment, for example, have taken off enormously. It therefore makes sense to see how we can support the specialist in using the information from the images for research. There is still so much information left on the shelf because images are simply too large to process, especially when it concerns a large number of images in a scientific study. On the other hand, images are often not comparable at all unless you properly pre-process them. This is the substantive interest. We also see that healthcare is finding it increasingly difficult to continue to serve everyone. People are getting older and they sometimes continue to live through all kinds of tricks while this was not possible before. On the one hand, this is a very nice development, but on the other hand, we will no longer be able to accept that. So there is a serious social issue in healthcare that is increasing. If we contribute to solving this, I will be happy. I think that we are very well able to use our technologies (from preparation and analysis to prediction) to help fill the gaps in knowledge together with specialists. At BDC we always seek collaboration with specialists to strengthen each other. And finally, I also think it is important not to just process the images. Often many more sources are involved. Firstly, you already have different types of images, which together can provide additional information (for example an ECG with an fMRI - a special MRI technique for brain research with a three-dimensional image). In addition, we are also making progress in text mining, especially in the field of unstructured texts. Sometimes it is very useful to include such sources in an analysis. Image processing is of course also used for other purposes, such as in industry. These are partly the same technologies. But the social impact will probably be less significant. If there is an entrepreneur who wants to prove me wrong, I would be happy to recommend it.”