Lay summary by Dr Andrew Strange, reviewed by Dr Naomi Hartopp & a Parkinson’s disease lay panel
Background
Mobility issues affect individuals with a range of conditions, from neurological conditions such as Parkinson’s disease to physical injuries and heart disease. Mobility issues impact quality of life, leading to a higher risk of falls, loss of independence, and reduced physical fitness. Accurately measuring mobility in real-world settings (such as in the person’s home, not in a hospital) is crucial for understanding the effectiveness of interventions aimed at improving gait and overall mobility.
Monitoring these mobility issues has been challenging. Currently, there is a range of different computer programs (referred to as algorithms) which are run on wearable devices aimed at tracking a patient’s gait at home. These wearable devices help doctors and researchers better understand how an individual’s mobility may be affected, allowing the development of new treatments. Since the use of these devices are still new, it has been unclear which algorithm works best for patients in their own homes. This study aimed to assess several available algorithms to determine which is the best for detecting how patients walk at home.
Why is the study important?
This study is particularly important because it tests a range of algorithms in the patients’ home environment and real-world settings, not just within a hospital. Traditional clinical assessments often lack real-world validity as they are performed in controlled environments, where patients walk very differently. High quality monitoring devices, with good algorithms can provide continuous and accurate measurements of how a person walks in everyday settings, leading to better insights into mobility-affecting disorders.
What did the authors do and how did they do it?
The authors recruited 108 participants, including 20 healthy older adults and individuals with one of six conditions: Parkinson’s disease, multiple sclerosis, hip fractures, chronic obstructive pulmonary disease, and heart failure. This group represents a wide range of different mobility issues to test the devices. Participants used a wearable device to monitor their gait. Each algorithm was tested over 2.5 hours, during which participants were asked to go about their day as normal and perform specific tasks such as going up stairs. The collected data were compared against a previously established gold-standard reference known as Inertial Modules, Distance Sensors, and Pressure Insoles (INDIP). A total of 14 different algorithms were compared to INDIP, across measurements covering the different movements that make up a person’s walking gait. These measurements included stride length and speed, how good the algorithm was at detecting when walking started and stopped or when a patient initially placed their foot on the floor. How a patient performs the actions as part of walking is very informative about how their particular condition is progressing, so accurate monitoring of them is vitally important.
What are the results?
The research team identified that the best algorithm varied depending on the condition, but that the best algorithm for a condition was often still quite effective when used on participants with other conditions. Walking speed significantly affected algorithm effectiveness: walking speeds below 0.5 m/s, the boundary between “slow” and “medium” walking speeds, caused almost all of the tested algorithms to become less accurate. Since condition heavily affected walking speed, the best algorithm for each condition varied. The most challenging group was the fracture group, which tended to have a very slow, irregular gait.
What do the findings mean going forward for people with the disease?
Having assessed a number of different algorithms across various measurements, for each measurement one algorithm outperformed the others and whilst there was no one single perfect option for all conditions, there was a clear best choice for therapists and clinicians depending on the presented condition. This will allow much more accurate monitoring of patients, aiding the development of treatment plans and tracking disease progression.
This study can be found at
Journal of NeuroEngineering and Rehabilitation
Paper title
Assessing real-world gait with digital technology? Validation, insights, and recommendations from the Mobilise-D consortium
Lead Authors
M. Encarna Micó-Amigo, Silvia Del Din
Publication details including date of publication
Journal of NeuroEngineering and Rehabilitation, 14 June 2023.