Datathon: Human activity recognition from wearables data

Saturday 11th June, 9:30-5:30, Reuben Common Room

This datathon will offer you an opportunity to begin acquiring skills, or expand existing capabilities, in machine learning, applied in this case to a healthcare dataset. It is intended that there should be two groups of students taking part in the datathon: those with little or no knowledge of coding, and those with previous experience (python, in this case) who want to apply their coding skills to an interesting dataset. We therefore expect students from all four of the University’s academic Divisions to be interested in taking part in the datathon, which will be run on Saturday 11th June in the Reuben Graduate Common Room (with lunch and dinner provided).

Millions of people worldwide are now regular users of smartwatches with activity trackers which use in-built accelerometers (e.g. Fitbit and Apple Watch). This enables analyses of human movement and physical activity levels at the population scale, which are of great value for epidemiological studies such as understanding the relationship between activity and sleep and how they may predict future health status.

Technical challenges remain in the processing and analyses of activity tracker data. Current approaches include the development of activity recognition using machine learning models that translate the accelerometer readings into activity labels (e.g. walking, running, sitting). Most of these models are developed using labelled data collected in a lab setting, which have a number of limitations such as very short measurements, a limited set of pre-specified activities, and the absence of hybrid movements.

In this one-day datathon, you will have the opportunity to develop a machine learning activity recognition model using unique accelerometer data collected in a free-living setting. The data were collected from ~150 subjects who wore an accelerometer for 24 hours along with a wearable camera during the daytime, making it the largest labelled accelerometer dataset collected in natural everyday environments. You will be encouraged to think about and comment on potential ethical issues associated with these data.


9:30am Introduction to wearables in health - Aiden Doherty
10am Machine learning of activity data - Shing Chan
10:30am Hands-on python tutorial on machine learning of human activity recognition
12pm Lunch & methods brainstorming
12:45pm Group work to innovate and implement (with support of tutors)
5pm Dinner and group presentations of results
5:30pm Close – Aiden Doherty


The datathon will be led by Professor Aiden Doherty, who is a world-leading expert in this field, and supported by graduate students who are familiar with the dataset.

Sign up for the datathon on Inkpath