MPSD researchers win hackathon with AI system to gather chronic illness data
A team involving three young researchers from the Scientific Support Unit Computational Science at the MPSD has won a hackathon with an Artificial Intelligence (AI)-based solution for a data platform which helps patients manage their serious chronic illness.
The contest was organized by the European COMPAR-EU project, a scheme developing better ways for patients to manage serious health conditions like Type 2 diabetes, heart failure, obesity and chronic obstructive pulmonary disease. To improve the self-management of these illnesses, the project provides decision aids on an online platform for patients wanting to change their lifestyle, monitor symptoms or find support.
The platform pools vast amounts of evidence-based data to find the most effective and cost-efficient self-management approaches. Currently, researchers manually extract this data from scientific studies – a time-consuming process. The winning hackathon team came up with an automated AI-based solution to collect the essential data from the studies – such as the participants’ age, gender, disease and the chosen interventions. With this far more efficient method, the project can gather lots of data in less time. The evaluation of that data, in turn, can help medical staff to advise patients on the best ways to manage their health.
Furthermore, by systematically analyzing the literature and synthesizing research results COMPAR-EU aims to reduce the time lag between research evidence and clinical practice (see reference).
MPSD software engineer Ashwin Kumar was part of the team, together with visiting scientists Martin Lang and Kevin Yanes Garcia. Their team colleagues were Marilina Santero, Tatjana Scherer and Harsh Shah. “I really enjoyed working in an interdisciplinary team bringing together machine learning, programming and medical experience,” said Ashwin Kumar, “and the organizers did a fantastic job in creating a friendly and welcoming atmosphere.“
All three groups presented remarkable proposals, said hackathon organiser Oliver Gröne, but the winning team’s approach “convinced the jury because of the use of existing tools, such as NLP, BERT, robotreviewer and Grobid.”
The Scientific Support Unit for Computational Science, led by Hans Fangohr, supports ongoing research at the MPSD in all areas relevant to the use of computation for research. This includes the analysis of experimental data sets, design and use of computer simulation, software engineering for research software, data visualisation, high performance computing, reproducibility of results and re-usability of software.
Reference: Morris ZS, Wooding S, Grant J. The answer is 17 years, what is the question: Understanding time lags in translational research. J Royal Society Medicine 2011