Dataset Information

Undersampling and cumulative class re-decision methods to improve detection of agitation in people with dementia.

ABSTRACT: Agitation is one of the most prevalent symptoms in people with dementia (PwD) that can place themselves and the caregiver's safety at risk. Developing objective agitation detection approaches is important to support health and safety of PwD living in a residential setting. In a previous study, we collected multimodal wearable sensor data from 17 participants for 600 days and developed machine learning models for detecting agitation in 1-min windows. However, there are significant limitations in the dataset, such as imbalance problem and potential imprecise labels as the occurrence of agitation is much rarer in comparison to the normal behaviours. In this paper, we first implemented different undersampling methods to eliminate the imbalance problem, and came to the conclusion that only 20% of normal behaviour data were adequate to train a competitive agitation detection model. Then, we designed a weighted undersampling method to evaluate the manual labeling mechanism given the ambiguous time interval assumption. After that, the postprocessing method of cumulative class re-decision (CCR) was proposed based on the historical sequential information and continuity characteristic of agitation, improving the decision-making performance for the potential application of agitation detection system. The results showed that a combination of undersampling and CCR improved F1-score and other metrics to varying degrees with less training time and data.

Supplementary information

The online version contains supplementary material available at 10.1007/s13534-023-00313-8.

SUBMITTER: Meng Z

PROVIDER: S-EPMC10769992 | biostudies-literature | 2024 Jan

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Undersampling and cumulative class re-decision methods to improve detection of agitation in people with dementia.

Meng Zhidong Z Iaboni Andrea A Ye Bing B Newman Kristine K Mihailidis Alex A Deng Zhihong Z Khan Shehroz S SS

Biomedical engineering letters 20230828 1

Agitation is one of the most prevalent symptoms in people with dementia (PwD) that can place themselves and the caregiver's safety at risk. Developing objective agitation detection approaches is important to support health and safety of PwD living in a residential setting. In a previous study, we collected multimodal wearable sensor data from 17 participants for 600 days and developed machine learning models for detecting agitation in 1-min windows. However, there are significant limitations in ...[more]

PMID: 38186943

Similar Datasets

Project description:BackgroundMany people with dementia (PwD) live and die with undiagnosed and untreated pain and are no longer able to report their suffering. Several pain assessment tools have been developed, tested, and implemented in clinical practice, but nursing home patients are reported to be still in pain. Clinicians and research groups worldwide are seeking novel approaches to encode the prediction, prevalence, and associations to pain in PwD.ParticipantsThe data in this analysis are acquired from the COSMOS study, a cluster-randomized controlled trial (2014 to 2015), aimed to improve the quality of life in nursing home patients (N = 723) through the implementation of a multicomponent intervention. We utilize baseline data of PwD (N = 219) with complete datasets of pain and agitation.MethodSystems analysis explores the relationship between pain and agitation using the Mobilization-Observation-Behavior-Intensity-Dementia (MOBID-2) Pain Scale, Cohen-Mansfield Agitation Inventory (CMAI), and Neuropsychiatric Inventory-Nursing Home version (NPI-NH). For each patient, the individualized continuous time trajectory, and rates of change of pain and agitation are estimated. We determine the relationship between these rates by analyzing them across the entire group.ResultsWe found that the new analysis method can generate individualized estimations for pain and agitation evolution for PwD, as well as their relationship. For 189 of 219 PwD, results show that whenever pain increases or decreases, agitation does too, with the same rate. The method also identifies PwD for whom pain or agitation remains constant while the other varies over time, and patients for whom agitation and pain do not change together. The algorithm is scalable to other variables and compatible with wearable devices and digital sensors.ConclusionWe presented a new approach to clinical data analysis using systems concepts and algorithms. We found that it is possible to quantify and visualize relationships between variables with a precision only dependent on the precision of measurements. This method should be further validated, but incipient results show great potential, especially for wearable-generated continuous data.

Project description:IntroductionGlobally, the number of people living with dementia (PLwD) is projected to increase substantially. Dentists will encounter an increasing number of PLwD retaining natural teeth who need dental care. Dental treatment planning can be complex for PLwD as both oral diseases and dental treatment can be detrimental to patients' oral and social function and comfort. In addition, patients may lack the capacity to make decisions about their treatment, posing further challenges for dentists.ObjectiveThis study aimed to explore dentists' approach to treatment decision-making with or for PLwD.MethodsSemistructured one-to-one interviews were completed with 22 dentists between April 2020 and March 2021. Data generation and analysis followed the principles of constructivist grounded theory with data being collected and analyzed simultaneously using a maximum variation sample. Interviews were recorded, transcribed verbatim, and then analyzed, leading to a theoretical understanding of how dentists approach treatment decisions for PLwD.ResultsFour data categories describe dentists' approaches to treatment decision-making for PLwD. Dentists sought to provide individualized care for PLwD. However, they described planning care based on risk-benefit analyses that primarily considered biomedical factors and generic assumptions. There was an underemphasis on the psychological or social implications of dental care or its delivery. Furthermore, while some dentists attempted to involve patients in treatment decisions, they reported fewer attempts to identify the views and preferences of PLwD identified as lacking decisional capacity. In this scenario, dentists reported leading the decision-making process with little regard for patients' known or identified preferences and minimal involvement of PLwD's family.ConclusionsDentists acknowledge the complexity in treatment decision-making for PLwD yet focused heavily on biomedical considerations, with an underemphasis on individuals' preferences and autonomy. Considering psychosocial aspects of care alongside biomedical factors is essential to support holistic person-centered care for this growing patient cohort.Knowledge transfer statementThis study highlighted that dentists may only be considering biomedical aspects of care when considering what treatment is appropriate for people living with dementia. When making or supporting treatment decisions, dentists should explore patients' preferences and actively consider relevant psychosocial factors. These can be actively identified through considered discussions with patients as well as their family members. Gathering this information should support more person-centered and value-concordant decision-making for people living with dementia.

Dataset Information

Undersampling and cumulative class re-decision methods to improve detection of agitation in people with dementia.

Supplementary information

Publications

Undersampling and cumulative class re-decision methods to improve detection of agitation in people with dementia.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets