Towards a Portable Model to Discriminate Activity Clusters from Accelerometer Data
Few methods for classifying physical activity from accelerometer data have been tested using an independent dataset for cross-validation, and even fewer using multiple independent datasets. The aim of this study was to evaluate whether unsupervised machine learning was a viable approach for the deve...
| Main Authors: | , , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2019-10-01
|
| Series: | Sensors |
| Subjects: | |
| Online Access: | https://www.mdpi.com/1424-8220/19/20/4504 |
| _version_ | 1828112582644334592 |
|---|---|
| author | Petra Jones Evgeny M. Mirkes Tom Yates Charlotte L. Edwardson Mike Catt Melanie J. Davies Kamlesh Khunti Alex V. Rowlands |
| author_facet | Petra Jones Evgeny M. Mirkes Tom Yates Charlotte L. Edwardson Mike Catt Melanie J. Davies Kamlesh Khunti Alex V. Rowlands |
| author_sort | Petra Jones |
| collection | DOAJ |
| description | Few methods for classifying physical activity from accelerometer data have been tested using an independent dataset for cross-validation, and even fewer using multiple independent datasets. The aim of this study was to evaluate whether unsupervised machine learning was a viable approach for the development of a reusable clustering model that was generalisable to independent datasets. We used two labelled adult laboratory datasets to generate a k-means clustering model. To assess its generalised application, we applied the stored clustering model to three independent labelled datasets: two laboratory and one free-living. Based on the development labelled data, the ten clusters were collapsed into four activity categories: sedentary, standing/mixed/slow ambulatory, brisk ambulatory, and running. The percentages of each activity type contained in these categories were 89%, 83%, 78%, and 96%, respectively. In the laboratory independent datasets, the consistency of activity types within the clusters dropped, but remained above 70% for the sedentary clusters, and 85% for the running and ambulatory clusters. Acceleration features were similar within each cluster across samples. The clusters created reflected activity types known to be associated with health and were reasonably robust when applied to diverse independent datasets. This suggests that an unsupervised approach is potentially useful for analysing free-living accelerometer data. |
| first_indexed | 2024-04-11T11:51:38Z |
| format | Article |
| id | doaj.art-15b64670b4b84bf3a054933b6f039e1b |
| institution | Directory Open Access Journal |
| issn | 1424-8220 |
| language | English |
| last_indexed | 2024-04-11T11:51:38Z |
| publishDate | 2019-10-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Sensors |
| spelling | doaj.art-15b64670b4b84bf3a054933b6f039e1b2022-12-22T04:25:17ZengMDPI AGSensors1424-82202019-10-011920450410.3390/s19204504s19204504Towards a Portable Model to Discriminate Activity Clusters from Accelerometer DataPetra Jones0Evgeny M. Mirkes1Tom Yates2Charlotte L. Edwardson3Mike Catt4Melanie J. Davies5Kamlesh Khunti6Alex V. Rowlands7Leicester Diabetes Centre, University Hospitals of Leicester, Leicester LE5 4PW, UKDepartment of Mathematics, ATT 912, Attenborough Building, University of Leicester, University Road, Leicester LE5 4PW, UKDiabetes Research Centre, University of Leicester, Leicester General Hospital, Gwendolen Road, Leicester LE5 4PW, UKDiabetes Research Centre, University of Leicester, Leicester General Hospital, Gwendolen Road, Leicester LE5 4PW, UKInstitute of Neuroscience, Henry Wellcome Building, Faculty of Medical Sciences, Newcastle University, Newcastle upon Tyne NE2 4HH, UKLeicester Diabetes Centre, University Hospitals of Leicester, Leicester LE5 4PW, UKLeicester Diabetes Centre, University Hospitals of Leicester, Leicester LE5 4PW, UKDiabetes Research Centre, University of Leicester, Leicester General Hospital, Gwendolen Road, Leicester LE5 4PW, UKFew methods for classifying physical activity from accelerometer data have been tested using an independent dataset for cross-validation, and even fewer using multiple independent datasets. The aim of this study was to evaluate whether unsupervised machine learning was a viable approach for the development of a reusable clustering model that was generalisable to independent datasets. We used two labelled adult laboratory datasets to generate a k-means clustering model. To assess its generalised application, we applied the stored clustering model to three independent labelled datasets: two laboratory and one free-living. Based on the development labelled data, the ten clusters were collapsed into four activity categories: sedentary, standing/mixed/slow ambulatory, brisk ambulatory, and running. The percentages of each activity type contained in these categories were 89%, 83%, 78%, and 96%, respectively. In the laboratory independent datasets, the consistency of activity types within the clusters dropped, but remained above 70% for the sedentary clusters, and 85% for the running and ambulatory clusters. Acceleration features were similar within each cluster across samples. The clusters created reflected activity types known to be associated with health and were reasonably robust when applied to diverse independent datasets. This suggests that an unsupervised approach is potentially useful for analysing free-living accelerometer data.https://www.mdpi.com/1424-8220/19/20/4504unsupervisedmachine learningphysical activityclusteringwrist-wornaccelerometerwalking |
| spellingShingle | Petra Jones Evgeny M. Mirkes Tom Yates Charlotte L. Edwardson Mike Catt Melanie J. Davies Kamlesh Khunti Alex V. Rowlands Towards a Portable Model to Discriminate Activity Clusters from Accelerometer Data Sensors unsupervised machine learning physical activity clustering wrist-worn accelerometer walking |
| title | Towards a Portable Model to Discriminate Activity Clusters from Accelerometer Data |
| title_full | Towards a Portable Model to Discriminate Activity Clusters from Accelerometer Data |
| title_fullStr | Towards a Portable Model to Discriminate Activity Clusters from Accelerometer Data |
| title_full_unstemmed | Towards a Portable Model to Discriminate Activity Clusters from Accelerometer Data |
| title_short | Towards a Portable Model to Discriminate Activity Clusters from Accelerometer Data |
| title_sort | towards a portable model to discriminate activity clusters from accelerometer data |
| topic | unsupervised machine learning physical activity clustering wrist-worn accelerometer walking |
| url | https://www.mdpi.com/1424-8220/19/20/4504 |
| work_keys_str_mv | AT petrajones towardsaportablemodeltodiscriminateactivityclustersfromaccelerometerdata AT evgenymmirkes towardsaportablemodeltodiscriminateactivityclustersfromaccelerometerdata AT tomyates towardsaportablemodeltodiscriminateactivityclustersfromaccelerometerdata AT charlotteledwardson towardsaportablemodeltodiscriminateactivityclustersfromaccelerometerdata AT mikecatt towardsaportablemodeltodiscriminateactivityclustersfromaccelerometerdata AT melaniejdavies towardsaportablemodeltodiscriminateactivityclustersfromaccelerometerdata AT kamleshkhunti towardsaportablemodeltodiscriminateactivityclustersfromaccelerometerdata AT alexvrowlands towardsaportablemodeltodiscriminateactivityclustersfromaccelerometerdata |