TY - GEN
T1 - Falls Prediction in Care Homes Using Mobile App Data Collection
AU - Dvir, Ofir
AU - Wolfson, Paul
AU - Lovat, Laurence
AU - Moskovitch, Robert
N1 - Publisher Copyright:
© 2020, Springer Nature Switzerland AG.
PY - 2020/1/1
Y1 - 2020/1/1
N2 - Falls are one of the leading causes of unintentional injury related deaths in older adults. Although, falls among elderly is a well documented phenomena; falls of care homes’ residents was under-researched, mainly due to the lack of documented data. In this study, we use data from over 1,769 care homes and 68,200 residents across the UK, which is based on carers who routinely documented the residents’ activities, using the Mobile Care Monitoring mobile app over three years. This study focuses on predicting the first fall of elderly living in care homes a week ahead. We intend to predict continuously based on a time window of the last weeks. Due to the intrinsic longitudinal nature of the data and its heterogeneity, we employ the use of Temporal Abstraction and Time Intervals Related Patterns discovery, which are used as features for classification. We had designed an experiment that reflects real-life conditions to evaluate the framework. Using four weeks of observation time window performed best.
AB - Falls are one of the leading causes of unintentional injury related deaths in older adults. Although, falls among elderly is a well documented phenomena; falls of care homes’ residents was under-researched, mainly due to the lack of documented data. In this study, we use data from over 1,769 care homes and 68,200 residents across the UK, which is based on carers who routinely documented the residents’ activities, using the Mobile Care Monitoring mobile app over three years. This study focuses on predicting the first fall of elderly living in care homes a week ahead. We intend to predict continuously based on a time window of the last weeks. Due to the intrinsic longitudinal nature of the data and its heterogeneity, we employ the use of Temporal Abstraction and Time Intervals Related Patterns discovery, which are used as features for classification. We had designed an experiment that reflects real-life conditions to evaluate the framework. Using four weeks of observation time window performed best.
KW - Falls prediction
KW - Outcomes prediction
KW - Temporal data mining
UR - http://www.scopus.com/inward/record.url?scp=85092239106&partnerID=8YFLogxK
U2 - 10.1007/978-3-030-59137-3_36
DO - 10.1007/978-3-030-59137-3_36
M3 - Conference contribution
AN - SCOPUS:85092239106
SN - 9783030591366
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 403
EP - 413
BT - Artificial Intelligence in Medicine - 18th International Conference on Artificial Intelligence in Medicine, AIME 2020, Proceedings
A2 - Michalowski, Martin
A2 - Moskovitch, Robert
PB - Springer Science and Business Media Deutschland GmbH
T2 - 18th International Conference on Artificial Intelligence in Medicine, AIME 2020
Y2 - 25 August 2020 through 28 August 2020
ER -