Toggle Main Menu Toggle Search

Open Access padlockePrints

Gait event detection in cerebellar ataxia: A single vs. multiple sensor approach

Lookup NU author(s): Dr Javad SarvestanORCiD, Dr Silvia Del DinORCiD, Dr Lisa AlcockORCiD



This is the authors' accepted manuscript of a conference proceedings (inc. abstract) that has been published in its final definitive form by ISPGR, 2023.

For re-use rights please refer to the publisher's terms and conditions.


Gait event detection in cerebellar ataxia: A single vs. multiple sensor approachJavad Sarvestan1, Jens Seemann2, Silvia Del Din1, Matthis Synofzik2, Winfried Ilg2 & Lisa Alcock11- Translational and Clinical Research Institute, Faculty of Medical Sciences, Newcastle University, Newcastle Upon Tyne, UK.2- Departments of Cognitive Neurology, Hertie Institute for Clinical Brain Research, Tübingen, Germany. Background and AimMonitoring gait in the real-world using wearable sensors presents an exciting opportunity for improving access to healthcare and delivering timely interventions[1]. Sensor configuration (number and location) may influence the accuracy of gait event detection and calculation of subsequent gait outcomes. Using a wearable sensor on each foot is considered the most robust approach for detecting gait events, such as initial and final contacts, which are used to segment stance and swing phases. Using a single sensor on the lower back has advantages including; reducing data footprint (processing and storage), minimising patient burden by removing the need to move sensors between shoes thereby ensuring uninterrupted continuous data acquisition, as well as prolonging battery life[2]. Ataxic gait is characterised by uncoordinated movements with high movement variability[3]. This study evaluated the accuracy of a single sensor for detecting gait events compared to multiple sensors in a diverse group of patients with cerebellar ataxia and controls. Methods98 participants (control: n=43, 44% female, age=41.8±14.3y; pre-symptomatic: n=19, 74% female, age=38.1±12.4y; symptomatic: n=36, 36% female, age=49.1±11.9y) walked two 25m straight walks at their self-selected preferred pace in a laboratory setting. The Scale for the Assessment and Rating of ataxia (SARA) was used to define pre-symptomatic (SARA<3) and symptomatic (SARA>3) participants. A wearable sensor (APDM, Opal 128Hz) comprising accelerometer, gyroscope and magnetometer was affixed to the lower back and the dorsum of both feet (over the metatarsal bones). Two approaches were taken to identify gait events (initial contact-IC, and final contact-FC); a single sensor (lower back) using algorithms validated in older adults, people with Parkinson’s disease and ataxia patients[4,5] vs. multiple sensors (feet) using manufacturer algorithms. The positive predictive value (PPV) was calculated with a tolerance window of 0.5s and the median absolute error (MAE) for both ICs and FCs were calculated[6].ResultsA total of 9050 IC’s and 8660 FC’s were identified. PPV revealed minimal false positives for the detection of ICs for all participant groups with a median PPV of >90% (Fig.1A-C). False positives were higher for the detection of FCs with a median PPV of >80% for all participant groups. The MAE (Fig.1D) were similar for controls (IC=0.03s, FC=0.06s) and pre-symptomatic participants (IC=0.04s, FC=0.06s), and slightly higher for symptomatic participants (IC=0.06s, FC=0.07s).ConclusionsThere was a strong agreement between the single and multiple sensor approaches for gait event detection in controls and participants with cerebellar ataxia. Greater differences observed between the two approaches for the identification of FCs in symptomatic patients would influence the calculation of gait outcomes. Further efforts are required to determine whether the algorithm may be optimised for patients with cerebellar ataxia of moderate disease severity.References [1] Del Din 2019 Annals Neurology [2] Czech 2020 NPJ Digital Medicine [3] Ilg 2020 Neurology [4] Del Din 2015 IEEE [5] Hickey 2016 Phys Measurement [6] Bonci 2022 Front Bioeng & Biotec

Publication metadata

Author(s): Sarvestan J, Seemann J, Del Din S, Synofzik M, Ilg W, Alcock L

Publication type: Conference Proceedings (inc. Abstract)

Publication status: Published

Conference Name: (ISPGR 2023) International Society of Posture & Gait Research

Year of Conference: 2023

Online publication date: 13/07/2023

Acceptance date: 11/04/2023

Date deposited: 25/09/2023

ISSN: 2817-5042

Publisher: ISPGR


ePrints DOI: 10.57711/f05n-zq25