Utilising Artificial Intelligence (AI) to analyse multidimensional data is creating a stronger profile for disease diagnosis and a more accurate understanding about disease progression.

In essence, it’s changing the way researchers conduct their studies and the trajectory of research, leading to a deeper understanding of certain diseases for earlier diagnosis, better management strategies, and possibly new discoveries leading to cures.

Dr Robin Vlieger, a neuroscientist at the School of Medicine and Psychology has incorporated machine and deep learning methods in his research to better understand neurological disorders such as Parkinson’s Disease and Multiple Sclerosis (MS).

So, what does machine learning (a branch of AI) involve?

Dr Vlieger explained, “Initially, we gather data from patients. The range of data we can plot and look at is vast. It could include blood work, data related to a person’s ability to balance, and EEG (Electroencephalogram aka brain activity) data, to name just a few.”

“In general, it is challenging to see the relationship between different variables beyond the 3-dimensional space.  Machine Learning eliminates these difficulties and allows mathematical statistical models to be learned and added to, making quantifying these relationships much easier.”

“For example, in the case of an MS patient, they may deal with fatigue. But there are multiple types of fatigue – physical, cognitive and mental.  If we split the data sets for the different types of fatigue and then add in a person’s blood values and their postural sway (balance) data, we can start to create an understanding about the individual and the cohort overall.”

“This leads to very rich information that allows us to find the relationships between variables and how they relate to the individual.”

For deep machine learning to be accurate, there needs to be comparison between the models used to train the computer for the algorithms initially used, and the final models used to come up with validated findings.  Between testing and validating, the researcher must train the computer to tweak or tune the algorithms.

Dr Vlieger works in this way, a relatively new approach, to ensure his research has direct comparison methods and can be validated.

“We train and test the algorithm to find the best model. If some individuals are misclassified during testing, we revisit the data to understand why they're not classified accurately.”

“It was during my PhD studies that I learned how important the combination of understanding machine learning and also understanding the patient and their data is for creating an algorithm that produces accurate, impactful research.”

“In its simplest iteration, splitting the data of males and females results in very different information than if all data is run through the algorithm together, because men and woman are different.”

“If you think about this more broadly (eg. a person’s blood type, their weight or height) there are going to be many variables that can impact a diseases progression, so testing against controls and validating is exceptionally important for trying to create a population level understanding of a disease verses personalised diagnosis.”

Dr Vlieger acknowledges that this type of research, which ultimately hopes to cure diseases, requires significant time and funding.

He currently works with a broad team of researchers including clinicians, biologists, software engineers and app designers, to create deeper insights into neurological conditions.

His significant contribution to this area of research resulted in Dr Vlieger being named a finalist in the Outstanding New Researcher category for the 2025 CHARM Rising Star Awards.