Activity Monitoring

From Capsil Wiki

Jump to: navigation, search

Disclaimer: This Wiki is a work in progress and it has not yet been evaluated through a process of external review. No responsibility is accepted for errors or omissions or for any loss or damages which may result from the use of the information contained in the Wiki.

Contents

Introduction

Aging is undoubtedly one of the key global challenges that is exuberated by changing demographics and improving effectiveness in therapeutic interventions. Despite the successes of the reactive model of care the aging process threatens the elders’ quality of life. In contrast, continuous monitoring of elders has the potential of providing proactive models of care.

Although the process of aging is associated with a decline of many physical and cognitive functions, an elder can maintain independence and high quality of life provided that the individual is able to do the Activities of Daily Living (ADLs), challenging physical and mental exercises, and maintains their social activities and engagements. In fact, functional impairment is sometimes defined as difficulty performing, or requiring the assistance of another person to perform ADLs. Needless to say, early detection of changes in the elders’ activities is critical in providing appropriate level of care. In addition, changes in activities are likely to reflect important changes in their cognitive, sensory-motor (e.g., balance) or physical competence and as such may reflect symptoms of neurodegenerative diseases. One can easily think of scenarios in which the application of technological advances to the care of elders has the potential to restore his quality of care as well as quality of life.

Because of the importance of elders’ ADL competence, clinicians and caregivers adopted a formal, albeit subjective, approach to the assessment of activities. As such this judgment is subject to a variety of biases, including those due to contextual effects on sampling and denial. By the contextual effects of sampling, we mean biases that arise when an elder knows that he is being observed and that the outcome of this observation may lead to the loss of independence and possible move to the nursing home. Of course, infrequent sampling will distort the assessment interpretation of the data due to the variation in performance faster than sampling frequency – an effect referred to as aliasing in the information communication community. Sufficiently frequent or continuous assessment, therefore, not only reduces the biases, but also allows the assessment of variability.

Continuous assessment of elders’ activities will most likely be achievable using technology that is unobtrusively integrated within the normal living environments such as the elders’ houses or residential facilities. In order to achieve the ability to detect and classify the elders’ activities, the unobtrusive monitoring system must be capable of collecting data from a suite of distributed sensors combine them with appropriate models and process them with a set of inference algorithms, and utility-based decision-making processes.


Issues

There are numerous issues that need to be tackled before we have systems capable of automatic classification and assessment of human activities.

  • Complexity - Human activities are complex and sometimes difficult to classify even by human observers.
  • Individual Differences - Each elder may have idiosyncratic ways to execute even the simplest actions, so that any system developed for a populations must be adapted to the individual’s behaviors.
  • Reliability - Unobtrusive or minimally obtrusive sensing renders the inference even more difficult because the indirect observations are usually more noisy and unreliable.
  • Deviations from Normal - It is impossible to train on deviations and therefore these must be defined as incongruent events [1].
  • Maintenance - In order to assure economic feasibility for the activity monitoring system, it must be very robust and require minimal maintenance
  • Installation - The system must be easy to install by an informal or low-skill formal caregiver.
  • Power - The battery in any battery-powered devices must last at least six months. This may be achievable using intelligent power management or power harvesting techniques.
  • Ground Truth - In real-life situations it is very difficult, if not impossible to obtain the label for the actual activity performed by the observed elder
  • Stakeholder Differences- Each stakeholder may have different requirements for the monitoring and classification system.

Justification

The currently accepted approach to the assessment of the ADLs and IADLs is based on subjective judgment of the formal and informal, e.g., family, caregivers, even of the methodology is formalized [2]. As such this judgment is subject to a variety of biases, including those due to contextual effects on sampling and denial. By the contextual effects of sampling, we mean biases that arise when an elder knows that he is being observed and that the outcome of this observation may lead to the loss of independence and possible move to the nursing home. Of course, infrequent sampling will distort the assessment interpretation of the data due to the variability in performance – effect referred to as aliasing in the information communication community. On an intuitive level, aliasing is a misinterpretation of the observed data that occurs when the sampling rate is slower than necessary to capture the variability in the observed phenomenon. More frequent or continuous assessment, would therefore not only reduce the biases, but also allow assessment of variability.

The shortcomings of the current approaches to the assessment of ADLs and IADLs, therefore, offer an ideal opportunity to introduce technological solutions for frequent or continuous monitoring and assessments of ADLs and possibly a subset of IADLs.

Related Elderly Aspects in CAPSIL

Enabling Technologies

  • Technical Approaches to ADL Assessment - This section provides a brief overview of a subset of approaches previously used for inferencing and classification of ADLs based on a variety of technological and algorithmic approaches. One of the firsts studies of ADLs using objective monitoring techniques were performed by Togawa and his colleagues [Yamaguchi, 1998], who monitored several daily activities of the subject (sleeping hours, toileting, meals) as well as a number of physiological parameters. Since then, there have been multiple attempts to develop techniques for inference and classification of ADLs based on different technologies and yielding varying degree of success.
  • Sensors and Hardware - The technological approaches range from unobtrusive passive infrared sensors to more complex passive or active radio-frequency identification (RFID) systems based on unique tags attached to most of the relevant objects.
  • Algorithms and Software Systems - The algorithms range from aggregation and visual representation – visualization – to sophisticated probabilistic methods. As it turns out even simple depiction of the sensor activities can be very effective for exploratory analysis but the desired inferences do require statistical data processing, pattern recognition and classification.
  • Evaluation - Although there are ever increasing efforts to develop techniques for monitoring of activities, there is surprising small number of studies of these systems in elders’ homes. There are several empirical studies that attempt to evaluate the sensor systems in laboratory environments, [Abowd and Mynatt, 2004; Doctor, Hagras, Callaghan, 2005; Helal, Mann, et al, 2005]

Although the ultimate evaluation requires monitoring for a substantial length of time, there are results from short term studies, e.g., 13 day installation, Glascock and Kutzik, 2000. The system was used to assess wakeup time and medication tracking using simple statistical analysis. An alternative approach has been attempted by the ILSA project of Honeywell.

  • Ethical and Social Aspects - Although the advantages of monitoring systems applied to care for elders is quite apparent there are numerous ethical issues associated with such systems. These issues range from security and privacy to the social and psychological implications of machine-based care delivery. Although there is increasing effort to address these issues, preliminary results of these studies suggest that the actual decisions to use such systems by the elders will depend on the case-specific cost-benefit tradeoffs.

The following is a preliminary collection of a sample of academic and industrial organizations involved in research, development and implementation of activity monitoring systems.

Academic Programs of Note

Industry Programs of Note

  • Health Anywhere (Formerly VaaSah) - Independent Living at Home
  • HomeFree - A system for localization, mobility assessment and safety assurance (wandering mitigation) based on active radio frequency identification (RFID).
  • Vigil - A system based on passive RFID
  • Aipermon GmbH & Co. KG - AiperMotion is a three-dimensional activity sensor (accelerator) for measuring, recording and motivating everyday activities. Data are transferred through the AiperCoach home system via Bluetooth which is later retrieved by the program supervisor. Aipermon’s evaluations are then used to generate written responses by mail or email to the user for successful weight and activity management.
  • ADT/GE: Quietcare System - Fast-acting alert system: Detects potential health risks , Prevents falls and hospitalization and Protects privacy and dignity. See GE QuietCare system for further information.
  • Continua Health Alliance - The Continua Health Alliance is an organization dedicated to enabling interoperable health care products and solutions worldwide, Member companies include: Roche, Novartis, Intel, Samsung, google etc. See Standards for further information.
  • Honeywell HomMed - Genesis DM - Genesis DM is seamlessly integrated into the innovative new Honeywell HomMed LifeStream™ telehealth platform, providing web-enabled, on-demand access to disease-specific symptom management (DSSM), customizable by diagnosis and symptoms. This tele-health device measures heart rate, blood pressure, and weight, and provides customizable subjective disease-related queries for a more complete picture of an individual’s health. Automated set up and automatic patient engagement with a friendly voice and easy-to-use interface guide the patient at every step.
  • Carematix - Wellness system. The CWS provides easy monitoring of the basic wellness parameters via a wireless connection between the monitoring device and a hub (transceiver) located in the home. The hub transmits the information to the Carematix internet server where the data is added to the patient's chart.
  • Using a web-browser, the caregiver can track the patient's data, graph the results, monitor trends, annotate variances, set alert criteria, and send reminders and receive alerts via e-mail or pager
  • Zeo personal sleep coach - Body worn sleep sensor, personal display unit and coaching website designed to promote better sleeping patterns and quality.


References

  1. ↑ M.B. Patterson and J.L. Mack. The Cleveland Scale for Activities of Daily Living (CSADL): Its reliability and validity. Journal of Clinical Gerontology, 7, 15-28, 2001
  2. ↑ M.B. Patterson and J.L. Mack. The Cleveland Scale for Activities of Daily Living (CSADL): Its reliability and validity. Journal of Clinical Gerontology, 7, 15-28, 2001

Navigation

Retrieved from "http://capsil.org/capsilwiki/index.php/Activity_Monitoring"
Personal tools