Project funding and team
The University of Manchester awarded a £10,000 Data Science Institute grant in January 2018 for “Developing novel data science methodology for analysing gait patterns for the early detection of cognitive decline.” The collaborative project team included Omar Costilla Reyes, Patricia Scully, Iracema Leroi, and Krikor B. Ozanyan. Dr. Costilla Reyes, then in the final months of his PhD, announced in December 2017 that the project had been funded through the institute’s 2017 call for small projects. Scully and Ozanyan supervised his doctoral research, and Leroi contributed clinical expertise on the healthcare theme.
The grant followed earlier presentations of the work. In 2017 he had presented “Analysis of multimodal human-generated sensor data: a deep learning approach for early detection of dementia” as a poster at the Advances in Data Science conference in Manchester, and in July 2018 he presented “Early detection of dementia with ambient sensing and deep machine learning” at the EuroScience Open Forum in Toulouse with the institute’s support.
Why gait is studied in cognitive decline
Gait is the pattern of a person’s walking. Walking looks automatic, but it draws on attention and executive function, the mental skills used to plan, switch between tasks, and ignore distractions. Studies therefore ask people to walk while performing a second, mentally demanding task, because changes in walking under that “dual task” are studied as a possible marker of executive function.
Research focus and contribution
The project applied data-science methods to gait patterns as a potential source of information for research into cognitive decline, building on Dr. Costilla Reyes’s floor-sensor studies of walking under cognitive load. Floor sensors record the footsteps of people walking across them without any device attached to the body, and the project asked how machine-learning methods could draw useful information from that raw sensor data. The same four collaborators later coauthored the 2020 IEEE Sensors Journal paper on age-related differences in walking patterns under a cognitive task, which analyzed recordings from healthy adults. The project advanced research methods for studying gait, bringing engineering, data science, and clinical expertise together around the question of how early changes in thinking show up in the way people walk.