The NUS Saw Swee Hock School of Public Health (SSHSPH) and Massachusetts Institute of Technology (MIT) co-organised the NUS-MIT Healthcare Analytics Symposium and Datathon 2017 on 30 June – 2 July 2017 with Merck Sharp & Dohme (MSD) Singapore as their industrial collaborator. Held for the first time in Singapore, the event gathered over 200 local and international clinicians, data scientists, software engineers and innovators to strengthen cross-disciplinary collaborations and explore how data analytics could be more effectively used in healthcare systems.
Dr Mengling ‘Mornin’ Feng, Organising Chairperson and Assistant Professor at SSHSPH, commenced the symposium on 30 June by highlighting that this event was a first of its kind to use the largest amount of real clinical data for the “datathon”, a term coined as a combination of data + hackathon to accentuate the application of the hackathon model to data analytics.
The datathon on 1–2 July 2017 saw 12 selected multi-disciplinary teams working together on analysing de-identified real-life data sets from the Medical Information Mart for Intensive Care (MIMIC) Database from Physionet, the Philips eICU Collaborative Research Database and the NUHS Department of Surgery EMR database. These three databases share similar data schemas which contain hourly physiologic readings from bedside monitors validated by ICU nurses. They also contain records of demographics, labs, nursing progress notes, discharge summaries, IV medications, fluid balance, and other clinical variables. Through these data sets, competing teams developed proposed analytics programmes that can derive useful insights which can be acted on. The teams were then judged on the insights and results derived and the innovativeness of their analytics.
The event ended with awarding the “The Flood” team led by Professor David Pilcher, Chair of the Australian and New Zealand Intensive Care Society Centre for Outcome and Resource Evaluation, as the winner of the datathon. The team developed a real time visualisation tool from the vast amount of continuous monitoring data to investigate the effectiveness of fluid bolus as a treatment to low blood pressure in ICU. Combining results from real-time visualisation and statistical models, they found that fluid bolus may be as effective as it was initially believed by many physicians. Individual patients factor greatly influence the treatment effects. This are definitely some interesting findings that are worth pursuing further.
Both the participating physicians and data scientists were very pleased with the outcomes of the event. Many have already asked for the next year’s event and even volunteered to help out in organising one. The participating physicians have been calling meetings to continue the work stated during the datathon and to formalise collaborations with the data scientists. This is exactly what we hoped to see out of the datathon.
See more photos here.