top of page
Search

Meet the Team: Andreas Werle van der Merwe, machine learning architect and data scientist

Updated: May 26, 2023

“The ultimate challenge is designing an AI model that can explain itself clearly and concisely.”


Andreas Werle van der Merwe is at the epicentre of the Epcon engine. By working as a machine learning architect and data scientist, he is responsible for building Bayesian disease models - for TB or Covid-19, and related infectious illness - to better predict disease “hotspots”. It’s imperative for Andreas to ensure that the real-time data, the building blocks of an AI engine, is maintained and entered correctly.


EPCON's AI model

He plays a pivotal role in furthering EPCON’s development through incorporating new and existing tools. The goal is to achieve better spatial resolution of disease models and improved automated screening recommendations. Andreas certainly knows the ropes; he’s been with EPCON for three years, and has worked on various projects in Pakistan, South Africa and Belgium.


It’s no surprise that his background and skills set is impressive: Andreas studied Mechanical Engineering at Stellenbosch University in South Africa, where he developed skills in system design, numerical simulation, and control systems, among others. He went on to do a Mechanical/Mechatronic Engineering masters in the Biomedical Engineering Research Group, focusing on neuroevolutionary algorithms and the emergence of neural modularity in multilayer perceptron networks, followed by an internship in Cognitive Systems, where he worked on geospatial semantic image segmentation.


So what are the biggest challenges in this field? Andreas explains that success of an AI model is dependent on data quality. Poorer data or data that is incorrectly time stamped might lead to less valid results. Nevertheless, the future of using AI predictive models for infectious diseases is incredibly positive due to the increasing digitization and standardisation of health data. AI models stand ready to learn, but a lot will depend on the way in which medical and public health data is collected and protected.


For Andreas, the ultimate challenge is designing an AI model that can explain itself clearly and concisely. This is of the utmost importance as EPCON continues its groundbreaking work on integrating sophisticated disease models into the real world.


bottom of page