During a 1-week hackathon, researchers in data science paired with doctors to address medical issues from hospital management to novel materials.

Medicine is a source of great challenges to statistics, machine learning and data science in general. From hospital management to early diagnosis, from gene variants to the development of new biomaterials for prosthetic applications, in the years to come healthcare will benefit hugely by the growing knowledge coming from the analysis of large volumes of data.

To bridge the gap between the medical needs and the statistical skills that make this knowledge accessible, the Department of Mathematics has organized the 2019 Data Dive, a week-long hackathon from 8^th to 12^th July, that attracted 19 students and researchers from around College, to look at medical data.

The analysis problems were provided by medical researchers, and represent topical and pressing challenges. On the first day, they were presented to the participants, who then worked in teams on a problem of choice. Through the week, they had a chance to interact with the researchers, and ask for clarification and feedback from the doctors. On Friday, the results of a week of efforts were presented to the panel and the other participants, for a quick Q&A session.

The problems discussed came from the department of Epidemiology and Biostatistics (Abbas Dehghan, “Clustering of associations between genetic variants and omics data”) the National Heart and Lung Institute (Philip Molyneaux, “Fibrotic Lung Disease: Can diagnosis and prognosis be predicted?”) the Dementia Research Institute (Nathan Skene, “Mapping cell type annotations across single cell datasets”) and the department of Materials (Stuart Higgins, “Understanding how cells sense their environment, by analysing changes in cell shape and protein localization caused by different surface geometries”).

A special award was conferred to a team that worked on a problem (“Do boarders in hospital get worse care and worse outcomes?”) proposed by Tom Woodcock and Paul Sullivan at the CLAHRC. The team members Gabriel Burcea (CLAHRC), Yi Yang (Surgery and Cancer) Eric Shen (Surgery and Cancer), Adriaan Hilbers (Maths), Johannes Lutzeyer (Maths) and Alex Geringer-Sameth (Physics) developed a model that allowed them to get insights on the way that patients are assigned to different wards, and what impact this has on the level of care they receive based on a 5-year dataset.

Marina Evangelou, co-organizer of the hackathon, says “we are very pleased with the response we had from both the doctors and the data scientists. We aim to deploy novel methodologies to move toward effective medical solutions, and to initiate longer-term collaborations across departments”.