For people affected by Parkinson's disease, balance problems are often part of a daily occurrence, and with that comes the challenge of not falling. There are studies that estimate that nearly two-thirds (60%) of people with Parkinson's sometimes fall. With dire consequences, up to and including hospitalizations. A study, conducted by the University of Oxford, looked at how wearable sensors, particularly the data collected from them, can help accurately predict fall risk in Parkinson's patients over a five-year period.
Accurate assessment of fall risk is critical to effective care planning for Parkinson's patients. However, current measurement and assessment methods are rather time-consuming and subject to a physician's subjective judgment. Thus, wearable sensors were used for the study. The results of the study were published in Digital Medicine.
The study
The team from the NeuroMetrology lab of the Nuffield Department of Clinical Neurosciences collected data from 104 people with Parkinson's disease who had not had previous cases. Six wearable sensors were used for this purpose. To collect the necessary data, patients were asked to perform tasks such as a two-minute walk and a 30-second posture task. In addition, the team used several commonly used questionnaires and clinical assessment charts to assess the severity of the disease and the patient's own perception of the decline in mobility.
After the initial test, follow-up tests were conducted at 24 and 60 months. The data were then all analyzed using machine learning methods. This allowed the researchers to identify key characteristics that distinguish people with Parkinson's with and without fall risk. The analysis revealed significant differences in characteristics related to walking and posture between those who did and did not fall.
Better understanding of fall risk
The study contributes to a better understanding of fall risk in people with Parkinson's and shows that wearable sensors can provide accurate predictions of fall risk. Moreover, thanks to the use of the sensors and AI technology, the study itself only needs to take three minutes. This, in turn, helps ease the workload of doctors and also reduces the burden on patients.
And, the researchers argue, when we are able to predict these falls, the logical next step is to provide a way to prevent them. If we are able to predict early on who is likely to fall, it could pave the way for more targeted and effective care programs and ultimately help prevent life-threatening falls.
“I am very pleased with the publication of this work. It is well documented that Parkinson's increases the risk of falls. This is work in progress from the past few years following patients from our OxQUIP cohort and it shows promise in accurately assessing falls and therefore allows us to think about effective care planning. This is a great opportunity to improve PD management and start developing realistic and effective prevention strategies,” said Professor Chrystalina Antoniades, lead author of the study.
Faster diagnosis
Sensors and machine learning are also being used to help diagnose Parkinson's earlier. To that end, American researchers this year developed a machine learning algorithm that analyzes data from wearable, motion-recording sensors. With this, parts of the testing process for diagnosing Parkinson's can be automated. Ultimately, the researchers say this could lead to more accurate and earlier diagnoses, which in turn could lead to earlier therapeutic interventions.
