c o m p u t e r m e t h o d s a n d p r o g r a m s i n b i o m e d i c i n e 9 1 ( 2 0 0 8 ) 55–81
`
`j o u r n a l h o m e p a g e : w w w . i n t l . e l s e v i e r h e a l t h . c o m / j o u r n a l s / c m p b
`
`A review of smart homes—Present state
`and future challenges
`
`∗, Daniel Est`eve a,b, Christophe Escriba a,b, Eric Campo a,c
`Marie Chan a,b,
`a LAAS–CNRS, 7, avenue du Colonel Roche, F-31077 Toulouse, France
`b Universit´e de Toulouse; UPS, INSA, INP, ISAE, LAAS–CNRS, F-31077 Toulouse, France
`c Universit´e de Toulouse; UTM, LATTIS, 1 place Georges Brassens, F-31703 Blagnac, France
`
`a r t i c l e
`
`i n f o
`
`a b s t r a c t
`
`Article history:
`Received 25 October 2006
`Received in revised form
`30 December 2007
`Accepted 3 February 2008
`
`Keywords:
`Smart home
`Elderly people
`
`In the era of information technology, the elderly and disabled can be monitored with numer-
`ous intelligent devices. Sensors can be implanted into their home for continuous mobility
`assistance and non-obtrusive disease prevention. Modern sensor-embedded houses, or
`smart houses, cannot only assist people with reduced physical functions but help resolve
`the social isolation they face. They are capable of providing assistance without limiting
`or disturbing the resident’s daily routine, giving him or her greater comfort, pleasure, and
`well-being. This article presents an international selection of leading smart home projects,
`as well as the associated technologies of wearable/implantable monitoring systems and
`assistive robotics. The latter are often designed as components of the larger smart home
`environment. The paper will conclude by discussing future challenges of the domain.
`© 2008 Elsevier Ireland Ltd. All rights reserved.
`
`1.
`
`Introduction
`
`According to French INSEE figures, 16.4% of the French popu-
`lation is in the “over 65” age group and an additional 8% are
`over 75; furthermore, these proportions are likely to increase.
`As a result, the ratio of persons aged 16–65 to those aged 65
`and over will decrease from over 3:1 (its value in the 1990s) to
`about 2:1 by the year 2040 [1]. Similar demographic changes
`are taking place in most European countries, the U.S.A., and
`Japan. Overall, this trend suggests that by 2050 approximately
`20% of the world population [2] will be at least 60 years old.
`One way to avoid institutionalizing older persons (or at
`least to defer it as long as possible) and reduce spiraling
`medical costs is through technology. We wish to not only
`cure illness, but also promote wellness in all stages of life.
`In particular, technology can help persons age at home in
`safety and independence. For many years, home automation
`has been considered a highly promising field for developing
`
`electronic technologies. Even as early as the eighties, several
`applications were being considered to enhance personal com-
`fort and safety. Although some of these solutions were highly
`sophisticated, the home market has lagged the telecom-
`munications and automotive markets—two industries where
`the role of electronics has continually grown more impor-
`tant. Households spent a large fraction of their income on
`cars, computers, etc. and very little is left for home automa-
`tion devices. Still, it is worth mentioning progress in certain
`areas:
`
`• Remote management: data associated with energy (gas and
`power), water, and telecommunications expenses can now
`be transmitted to the utility company without anybody
`going on site. As for home comfort systems, heating, air
`conditioning, ventilation, lighting, and doors and windows
`can all be automated and manipulated by remote control.
`Various electrical appliances such as washing machines,
`
`∗ Corresponding author. Tel.: +33 561 336 951; fax: +33 561 336 208.
`E-mail address: chan@laas.fr (M. Chan).
`0169-2607/$ – see front matter © 2008 Elsevier Ireland Ltd. All rights reserved.
`doi:10.1016/j.cmpb.2008.02.001
`
`APPLE 1050
`
`1
`
`

`

`56
`
`c o m p u t e r m e t h o d s a n d p r o g r a m s i n b i o m e d i c i n e 9 1 ( 2 0 0 8 ) 55–81
`
`dishwashers, refrigerators, and cooking devices can be pro-
`grammed to carry out their tasks. Radios and TV sets, as
`well as other entertainment devices, can be connected to
`share programming channels.
`• Elderly assistance: ever since the eighties, the elderly have
`benefited from devices that signal assistance services.
`Miniature transmitters can be worn around the neck or
`wrist, or carried in a pocket, allowing an individual to sig-
`nal danger or request help simply by pressing a button. In
`effect, the device is an emergency telephone connected to
`a professional service center or a family member.
`
`In what ways has change taken place?
`
`• Personal computers (PCs) are increasingly available, allow-
`ing people to access the Internet. In 2004, according to
`French INSEE figures [1], 45% of French households had a
`microcomputer and 7% had an Internet connection.
`• There has been rapid development of miniature,
`autonomous, and wireless sensors. It is now much easier
`to capture the information needed for implementation of
`home care services.
`• Passive and active “tags” are being used more and more
`often. These are smart, wireless mini-chips that can sense
`and act by data transmission. They may be tied to a physical
`space, a machine, a device, a production line, or a human
`body.
`• Cellular phones can be fitted with a global positioning
`system (GPS), or in the future with the European Galileo
`tracking system. These can be used to establish a perma-
`nent link between one’s home and the outdoors, easing
`interactions and interconnections between various agents
`who are either monitoring or being monitored.
`
`Thus, ambitious smart homes can be examined from the
`perspectives of comfort, leisure, and safety. This article aims
`to identify and describe a selection of leading smart home
`projects throughout the world. Some independent but related
`technologies are considered, including wearable/implantable
`systems and assistive, interactive robots. The paper concludes
`by summarizing future challenges in the domain.
`
`2.
`
`Nomenclature
`
`Telemedicine is defined as “the use of audio, video, and other
`telecommunications and electronic information processing
`technologies for the transmission of information and data
`relevant to the diagnosis and treatment of medical con-
`ditions, or to provide health services or aid health care
`personnel at distant sites” [3]. Originally, the term described
`mainly consultation services delivered through interactive
`video. In the Internet and multimedia era, this domain has
`evolved a much broader scope ranging from health pro-
`motion to disease prevention. As well as being used for
`patient education, telemedicine networks now include clin-
`ical decision databases, electronic patient records, artificial
`intelligence, and administrative support. For this reason, the
`term telehealth is preferred: it describes “the full array of tech-
`nologies, networks and healthcare services provided through
`
`telecommunication, including delivery of educational pro-
`grams, collaborative research, patient consultation and other
`services provided with the purpose of improving health” [4].
`The term home telehealth refers to “the use of telecommunica-
`tions by a home care provider to link patients or customers to
`one or more out-of-home sources of care information, educa-
`tion, or service by means of telephones, computers, interactive
`television, or some combination of each” [5].
`A few years ago the term eHealth appeared, defined by
`Eysenbach as “An emerging field in the intersection of medical
`informatics, public health and business, referring to the health
`services and information delivered or enhanced through the
`Internet and related technologies. In a broader sense, the term
`characterizes not only a technical development, but also a
`state-of-mind, a way of thinking, an attitude, and a commit-
`ment to networked, global thinking, to improve health care
`locally, regionally, and worldwide by using information and
`communication technology” [6].
`Home healthcare in the U.S.A. refers to individual health-
`care and social services such as nursing, rehabilitation, social
`work and health assistance, when provided to patients in
`their place of residence or some other home-like setting.
`Telehomecare, a specific type of telemedicine, uses a mixture
`of telecommunication and videoconferencing technologies to
`enable communication between a healthcare provider at their
`clinic and a patient at their home. This interaction is called
`a “virtual visit”, as opposed to the term “actual visit” which is
`used to describe traditional “face to face” interactions. A vir-
`tual visit can now include physical assessment of the patient
`through heart, lung and bowel sounds, as well as vital signs
`such as blood pressure and pulse [7].
`Demiris introduces the concept of home-based eHealth,
`which includes both telehomecare and the smart home. In this
`context, the second term refers to unobtrusive disease preven-
`tion and monitoring of residents who may not receive other
`forms of home care, such as the disabled or elderly [7].
`Various assistive devices are available in all task domains [8].
`This term is used for systems that have been designed to fulfill
`a single function. A fully integrated system is a device with mul-
`tiple functions controlled through a single human-machine
`interface; household integrated systems have been described
`for at least several decades [9,10]. Cooper and Keating use the
`term integrated home systems. Thus, a number of alternative
`and equivalent names are used to describe the full integration
`of consumer electronics: home systems, integrated home sys-
`tems, smart houses, and intelligent homes. “All approaches in
`the field of integrated home systems enable communication
`between different consumer electronic devices in the home so
`that they can cooperate, and thus function as a system rather
`than as a collection of independent devices” [11].
`The term rehabilitation integrated system refers to a group
`of rehabilitation assistance devices (usually electronic). When
`brought together, they can provide disabled individuals with
`better access and ergonomics [12]. Integrating assistive tech-
`nologies in this manner allows people with disabilities to more
`fully participate in society.
`Latin languages may use the word domotics, meaning
`automation of the house [13]. In English the concept is gener-
`ally referred to as the smart house [14]. In this paper, the terms
`“home”, “house”, “household”, and “housing” are considered
`
`2
`
`

`

`c o m p u t e r m e t h o d s a n d p r o g r a m s i n b i o m e d i c i n e 9 1 ( 2 0 0 8 ) 55–81
`
`57
`
`synonymous; “housing” has the additional sense of dwellings
`in general. Smart house is commonly used to refer to any living
`or working environment that has been carefully constructed
`to assist people in carrying out required activities.
`
`3.
`
`Review of projects
`
`The smart home concept is a promising and cost-effective way
`of improving access to home care for the elderly and disabled.
`Many research and development projects are ongoing, funded
`by international and governmental organizations. Note that
`some of the following smart homes have been designed to
`address a specific physical or mental disability.
`Smart homes can be classified according to the types
`of equipment and systems installed. The major targets are
`improving comfort, dealing with medical rehabilitation, mon-
`itoring mobility and physiological parameters, and delivering
`therapy. Technologies exist to help people deal with a reduc-
`tion or loss of mobility, vision, hearing, and cognitive ability;
`to continuously monitor vital parameters; to reduce accidents
`by anticipating risky situations; and to deliver therapy through
`wearable biomedical sensors. All these systems maintain a
`certain level of independence, thus providing a better qual-
`ity of life for the resident and his close relatives. However,
`the recipients of smart homes are not just those with severe
`pathologies or chronic illness; there are also those who simply
`want a better quality of life. For example, services enabling the
`“virtual visit” are particularly important in rural areas [15–17].
`A block diagram of a smart system is shown in Fig. 1. The
`functions that can be implemented in a smart home with ade-
`quate equipment, devices, or specific appliances are shown in
`Table 1. In the following section, some specific projects will be
`presented. The selected projects have been deemed among
`the most significant from an international perspective, but
`the list is not exhaustive. The smart homes discussed below,
`along with their equipment, key algorithms, and functions,
`are summarized in Table 2.
`
`3.1.
`
`Smart homes
`
`A number of smart homes have now been developed. Beyond
`issues of comfort and leisure, they are mainly intended to
`monitor elderly subjects with motor, visual, auditory or cogni-
`tive disabilities [18–21]. In each case the house and its various
`electrical appliances have been fitted with sensors, actuators,
`and/or biomedical monitors. The devices operate in a network,
`which is sometimes connected to a remote center for data
`collection and processing. The remote center diagnoses the
`ongoing situation and initiates assistance procedures.
`
`In the U.S.A.
`3.1.1.
`In Boulder, Colorado an “adaptive” house has been developed
`that uses neural networks to control temperature, heating,
`and lighting without previous programming by the residents.
`This system, called ACHE, attempts to economize energy
`resources while respecting the lifestyle and desires of its
`inhabitants. ACHE continuously monitors the environment
`and observes actions taken by the residents (using the lights,
`adjusting the thermostat). From these data, it infers patterns
`
`Fig. 1 – General organization of a smart system.
`
`in the home and uses reinforcement learning, a stochastic form
`of dynamic programming that samples trajectories in state
`space, to predict future behavior [22].
`The MavHome project (University of Texas, Arlington) aims
`to create a home that acts as a rational agent, trying to
`maximize the comfort of its inhabitants while minimizing
`operation costs. The agent must be able to sense and pre-
`dict the occupants’ mobility habits and their use of electrical
`appliances. The goal is to construct a universal predictor (or
`estimator) of user mobility. The so-called LeZi method, a tech-
`nique of information theory, is used to create a probabilistic
`model predicting the inhabitant’s typical path segments, com-
`fort management scheme, and appliance use. Specifically, the
`Active LeZi (ALZ) algorithm calculates the probability of every
`possible action occurring in the currently observed sequence,
`
`3
`
`

`

`58
`
`c o m p u t e r m e t h o d s a n d p r o g r a m s i n b i o m e d i c i n e 9 1 ( 2 0 0 8 ) 55–81
`
`Table 1 – Functions that can be implemented in the smart houses and the equipments, objects or appliances used
`
`System description
`
`Method
`
`Author
`
`Mozer [22]
`Das et al. [23,24]
`
`Helal et al. [25,26]
`
`Lesser et al. [27]
`Kidd et al. [28]
`
`Krumm et al. [29] and Brumitt et
`al. [30]
`Intille [31]
`Tapia et al. [32]
`Rantz et al. [33]
`
`Sensors
`Sensors
`
`Ultrasonic location tracking sensors,
`smart floor
`Robot, home appliances
`Ultrasonic sensors, RF technology, video,
`floor sensors
`Multi-camera, badge
`
`Sensors
`Sensors
`Sensor network, robotics
`
`Elite Care [34]
`Adami et al. [35]
`
`Digital technologies
`Wrist actigraph
`
`Mihailidis et al. [36]
`Yamaguchi et al. [37] and Tamura
`et al. [38]
`Matsuoka et al. [40]
`Isoda et al. [41]
`
`Vision-based system
`Sensors, magnetic switches, health
`
`Sensors, video camera,
`Sensors, RFID-tagged objects, sensor floor
`
`Artificial neural networks
`Modeling resident activity based on their past
`movements; LeZi algorithm
`Accurate location calculation
`
`Intelligent agents
`Pattern recognition, artificial vision techniques; Hidden
`Markov models
`Image analysis, statistical representation
`
`Context decision computing
`Context decision computing
`Comparison between stored sensor data and predefined
`urgency Artificial intelligent entities (medication
`dispenser)
`Gathering, storing and transmitting health information
`Lifestyle monitoring, especially time spent in and out of
`bed
`Computer vision, artificial intelligence (AI)
`Monitoring motor behavior and bed temperature, data
`processing
`Statistical analysis of activity time series
`Spatio-temporal representation of user states and user
`decisions
`Detection of human behaviors and activities
`Vital signs (pulse, respiration, movement) analyzed
`through Fuzzy logic
`Simple signal analysis
`Assessment of physiological signs
`Summarization of daily action data
`Location-recognition algorithm
`Case-based reasoning
`Home devices
`Simple data analysis
`Statistical detection of abnormal inactivity or
`household appliance use
`Locating a person and determining their current activity
`Active safety alarm (button on pendant)
`
`Home appliances (cooker, night light)
`
`Assessment of physiological signs
`Statistical analysis of physiological signs
`
`Statistics, artificial neural networks
`Statistics, artificial neural networks
`Assessment of resident’s mobility and physiological
`signs
`Statistical model for anxiety
`Spatial recognition (chemotatic model)
`Statistics, pattern recognition, activity duration
`threshold
`
`Yamazaki et al. [42]
`Andoh et al. [43]
`
`Masuda et al. [44]
`Nishida et al. [45]
`Noguchi et al. [46]
`Ha et al. [47]
`Ma et al. [48]
`Orpwood et al. [49]
`Williams et al. [50]
`Barnes et al. [51]
`
`Perry et al. [52]
`Vermeulen et al. [53] and
`Harrington et al. [54]
`Hagen et al. [55] and Adlam et al.
`[56]
`Elger et al. [57] and Deafblind
`Interna-tional.org [58]
`Virone et al. [59] and Lebellego et
`al. [60]
`Guill ´en et al. [61]
`Tuomisto et al. [62] and Korhonen
`et al. [63]
`Chan et al. [64–70]
`Campo et al. [71]
`Celler et al. [72,73]
`
`Sensors, cameras, microphones, robots
`Sensors (pneumatic microphone, air
`cushion, pressure sensor)
`Sensor (air pressure)
`Sensors
`Sensors
`Sensors
`Sensors
`Sensors
`Sensors
`IR sensors, magnetic switches
`
`Sensors, magnetic switches
`Model house equipped with devices for
`home automation
`Sensors
`
`Sensors, TV
`Sensors
`
`IR sensors
`IR sensors
`Sensors
`
`Sensors, actuators
`
`Home automation
`
`IR sensors, magnetic switches
`
`Statistics, probability theory
`
`West et al. [74]
`Riedel et al. [75]
`Diegel et al. [76]
`
`Sensors, reed switches
`Video tracking system
`Sensors
`
`and predicts the action with the highest probability [23].
`MavHome combines several technologies: databases, multi-
`media computing, artificial intelligence, mobile computing,
`and robotics [24].
`In Florida, Helal et al. have developed a smart home project
`known as the “GatorTech Smart House”. It is based on a num-
`
`ber of individual smart devices: mailbox, entrance door, bed,
`bath, floor, etc. The bathroom mirror is used as reminder
`device. All these components are fitted with sensors and actu-
`ators and connected to an operational platform designed to
`optimize the comfort and safety of an older person [25]. The
`“GatorTech Smart House” also uses a high-precision ultrasonic
`
`4
`
`

`

`c o m p u t e r m e t h o d s a n d p r o g r a m s i n b i o m e d i c i n e 9 1 ( 2 0 0 8 ) 55–81
`
`59
`
`tracking system to locate occupants, evaluate their mobility
`habits, and better control the environment. An in-laboratory
`mockup of the house is described in reference [26]. It is inhab-
`ited by Matilda, a test dummy. Two beacons are placed on
`Matilda’s shoulders, and a smart phone is attached to her left
`hand. Triangulation provides the subject’s location.
`The University of Massachusetts at Amherst multi-agent
`systems laboratory has developed a distributed set of
`autonomous home control agents, and deployed them in a
`simulated intelligent home environment [27]. Their goal is to
`automate some of the tasks currently performed by humans,
`with an eye towards improving efficiency and quality of ser-
`vice. The simulated intelligent home consists of four rooms
`joined by a common hallway: a bedroom, a living room, a bath-
`room, and a kitchen. Various intelligent agents (WaterHeater,
`CoffeeMaker, AirConditioner, DishWasher, VacuumCleaner,
`etc.) control the home environment. Moreover, a robot is used
`to fetch items and to move goods from one location to another.
`The agents reason about their assigned task, and quantify the
`value of candidate actions based on the resident’s wishes and
`the availability of resources. An agent’s repertoire of primitive
`actions is defined by a set of discrete probability distributions
`in terms of duration, quality, and cost. The intelligent agents
`must interact and coordinate over shared resources (for exam-
`ple, the DishWasher agent uses electricity and hot water).
`The task modeling and allocation framework models and
`quantifies resources, agent interactions, task interactions, and
`the performance characteristics of primitive actions so that
`agents can reason about the trade-offs of different courses of
`action and adapt their behavior to the changing environment.
`The laboratory also built and designed the Multi-Agent Sur-
`vivability Simulator (MASS) and the Java Agent Framework as
`tools for evaluating the agents and their coordination.
`In the Aware Home Research Initiative at the Georgia Insti-
`tute of Technology, an interdisciplinary team of researchers
`built a three-story, 5040 ft2 home that functions as a living lab-
`oratory for the design, development, and evaluation of future
`domestic technologies [28]. The smart floor senses an indi-
`vidual’s footsteps, which allows the home to build a model
`based on the user’s habits and behavior. A number of math-
`ematical tools are used to create and evaluate the behavioral
`model: hidden Markov models, simple feature-vector averag-
`ing, and neural networks. Their main goal is to enable the
`elderly to remain in familiar surroundings as they age, not only
`to improve their quality of life but also to lengthen their life.
`Researchers on the Aware Home project have also developed
`a system of tracking and sensing technologies to help find
`frequently lost objects such as wallets, glasses, and remote
`controls. Each object is given a small radio-frequency (RF)
`tag. The user interacts with the system via LCD touch panels
`placed strategically throughout the house. The system guides
`the user to the lost object using audio cues.
`Microsoft’s EasyLiving project, based on “context aware
`computing”, uses tracking video to monitor residents. Images
`from the video feed are analyzed and processed using dis-
`tributed computing. The system identifies people-shaped
`clusters of blobs in real time, allowing the system to fol-
`low individuals through the house. Residents are recognized
`through an active badge system. The EasyLiving Geomet-
`ric Model provides sub-meter localization of entities in the
`
`Table 2 – Smart homes or systems with their
`equipments, key algorithms and functions used
`
`Function
`
`Equipment/Device/Object
`
`To support
`
`To monitor
`
`To deliver therapy
`
`Comfort
`
`Disabled users
`Rehabilitation robotics
`Companion robot
`Wheelchair
`Specialized interface
`Synthetic voice generation for
`control and command
`Visually impaired subjects
`Tactile screen
`Sensitive remote control
`Audible beacon
`Hearing-impaired subjects
`Visible alarm
`Teletype machine
`Electronic display screen
`Numerical documents
`
`Lifestyle
`Fixed systems
`Infrared sensors
`Wearable systems
`Active badge
`Accelerometer
`Physiological Signs (external or in
`vivo sensors)
`EEG (syncope, epileptic seizure,
`sleep disorder, etc.)
`EMG
`Heart rate
`Temperature
`Blood oxygen saturation
`Blood pressure
`Glucose
`
`Therapeutic devices
`Delivery of current to abort or
`forestall epileptic seizures
`Tremor suppression
`Drug delivery
`Hormone delivery (e.g., insulin)
`Active or orthotic boots (podiatry)
`Robotic devices for bimanual
`physical therapy
`
`Intelligent household devices
`Dishwasher, washing machine,
`refrigerator
`Stovetop, cooker
`Smart objects
`Mailbox
`Closet
`Mirror
`Intelligent house equipment
`Presence/motion sensors
`Video camera
`Magnetic switches
`Humidity, gas, and light sensors
`Smart leisure equipment
`TV, home cinema programs
`Interactive communication systems
`Communication with friends and
`family in case of emergency
`Intelligent environmental control
`equipment
`Windows and doors
`
`5
`
`

`

`60
`
`c o m p u t e r m e t h o d s a n d p r o g r a m s i n b i o m e d i c i n e 9 1 ( 2 0 0 8 ) 55–81
`
`Table 2 – (Continued )
`
`Function
`
`Equipment/Device/Object
`
`Heating
`Lighting
`Air conditioning
`Ventilation
`Physical activity
`Fitness devices
`
`environment. Measurements are used to define geometric
`relationships between the entities (people, devices, places and
`things) needed for a particular interaction. In an example sce-
`nario, the resident (Tom) wishes to start playing music. The
`smart home uses its geometric world knowledge to select
`those speakers and other components which are most suit-
`able for the task, based on Tom’s current location. Thus, Tom
`is able to focus only on the decisions that require his input:
`the music itself. Current development is focused on more fully
`integrating the various devices and providing a coherent user
`experience. Their research progresses on a variety of fronts,
`including middleware development (to facilitate distributed
`computing), geometric world modeling (to provide location-
`based context), computer perception (to gather data about
`world state), and better service description (to support decom-
`position of device control, internal logic, and user interface)
`[29,30].
`The House n group at MIT, or “the house of the future”,
`proposes a smart services delivery system that conducts qual-
`itative and quantitative studies on the relationships between
`environmental factors and the behavior of the subject. The
`system consists of three components: a set of state-change
`sensors used to collect data about the use of objects, a context-
`aware experience sampling tool (ESM) used by the subject
`to label his activities, and pattern recognition and classifica-
`tion algorithms for recognizing activities. The user model is
`based on a training data set. In practical tests, the sensors
`were installed on pieces of furniture, kitchen appliances, bath-
`room appliances, and the washing machine. The subjects were
`given a personal digital assistant (PDA) running the ESM soft-
`ware at the start of the study, and asked to collect their activity
`data. A na¨ıve Bayesian network approach is used to train the
`model and predict user activities. The authors of this study
`concluded that while the model’s accuracy for some activi-
`ties is better than chance, it is not as high as expected. Their
`main problems were the low quality and number of activity
`labels, and the small training set of 2 weeks. They expect that
`by collecting training data over a period of months, generat-
`ing higher quality activity labels by video observation or other
`methods, and improving the information collection boards
`and sensors, they will be able to greatly improve the accuracy
`of the model [31,32].
`The “Aging in Place” project, at the University of Missouri-
`Colombia, offers a long-term care model for seniors who want
`supportive health care services in a home environment of their
`choice. The “Aging in Place” project consists of two comple-
`mentary initiatives: Senior Care and TigerPlace. Senior Care
`was initially designed to provide community-based support
`and health services, including an environmental component,
`to the residents of TigerPlace. It now serves occupants of
`
`other private and public senior residences, and some seniors’
`private homes in Boone County, Missouri. The project is char-
`acterized by interdisciplinary research, innovative educational
`programs, and an ideal practice environment for health care
`providers; the overall goal is to implement better ways of
`caring for older people who wish to “age in place” [33]. The
`TigerPlace residence, designed in collaboration with the Amer-
`ican Corporation of Sikeston, Missouri, uses a network of
`wireless sensors connected to small computers. Some sensors
`measure proximity and motion, while others sense weight on
`a mat, hear calls, or assess a variety of vital signs. The system
`is designed to notice functional decline and call for an inter-
`vention in case something goes wrong. An important aspect
`of the project is training older participants to accept and use
`the technology. TigerPlace opened in 2004, and is located just
`a few miles from the MU campus.
`Elite CARE (creating an autonomy-risk equilibrium) is an
`assisted living facility in Portland, Oregon using smart home
`technologies. It is inhabited by retirees, some of whom suffer
`from dementia or Alzheimer’s disease [34]. The aim is to pro-
`long independence and help the staff identify health problems
`early. Health information is processed in real time using dig-
`ital technologies, including the Internet. The system detects
`behavioral cues indicating change in an individual’s physical
`or cognitive condition, enhances social networks via elec-
`tronic mail, and regulates ambient conditions. Researchers at
`Oregon Health and Science University associated with the Elite
`CARE project have developed a method of unobtrusively mon-
`itoring the residents’ sleep characteristics. The data permit
`estimation of each resident’s bedtime and wake-up time, as
`well as their position shifts during sleep [35].
`In Toronto, Canada, Mihailidis et al. have developed a
`vision-based system capable of tracking the gross and fine
`motor movements of older adults. The vision system consists
`of three agents: sensing, planning, and prompting. The sens-
`ing agent was developed using a Sony video camera and a
`Matrox Meteor II frame grabber installed on a 2.4-GHz personal
`computer. Both statistics-based and physics-based methods
`of segmenting skin color in digital images are used for face and
`hand tracking in real time. The main goal of this technology
`is not just to recognize and track the hand positions associ-
`ated with each activity of daily living (ADL) step, but also to do
`so discreetly—the better to support a policy of aging-in-place
`[36].
`
`3.1.2.
`In Asia
`In Japan, about 15 smart houses are being developed. They
`usually aim to maximize the use of assistive technologies,
`enabling older people to live at home by creating a smart and
`comfortable environment. The Japanese Ministry of Interna-
`tional Trade and Industry built 13 examples called “Welfare
`Techno-Houses (WTH)”. The objective of these care houses is
`to improve the quality of life of both elderly people and their
`caregivers. The WTHs have been used as a test bed for new
`diagnostic technologies as well as the evaluation of residents’
`living conditions. The researchers collect data on residents’
`health and physiological signs by equipping the bathroom
`with fully automated medical devices. Their physical activity
`is monitored by equipping the rooms with infrared (IR) sen-
`sors and the doors with magnetic switches. Many weeks of
`
`6
`
`

`

`c o m p u t e r m e t h o d s a n d p r o g r a m s i n b i o m e d i c i n e 9 1 ( 2 0 0 8 ) 55–81
`
`61
`
`raw data on an experimental subject’s localization, ECG, and
`bed temperature are now available [37,38].
`The smart house of Dr. Matsuoka, based in Osaka, automat-
`ically detects unusual events that may be caused by disease
`or an accident through its 167 sensors. Seventeen electrical
`appliances are also fitted with sensors (refrigerator, TV set,
`rice cooker, air conditioning, etc.). Each sensor is associated
`with one or more activities: getting up, going to bed, prepar-
`ing meals, having a wash, working in an office, and so on.
`Matsuoka uses a two-step method to translate the raw sensor
`signals into behavioral data. Each time-segment of sensor data
`is to be associated with one of a limited number of living states,
`and the similarity of series in the same category is evaluated.
`The method used to construct the states is principal compo-
`nent analysis, which reduces high-dimensional data sets to a
`manageable number of independent linear combinations [39].
`Thus, the first step extracts principal components from vec-
`tors of sensor data obtained within a specified time window.
`The second step identifies statistical clusters in the complete
`sensor data based on the principal component decomposi-
`tion. There are only two free parameters in this analysis: the
`number of principal components sought, and the distance
`permitted between clusters. The method was verified using
`a 1-year observation of a four-person family interacting with
`the system. Unusual states were detected a total of 73 times
`during this period. Based on the family’s testimony, 19 of these
`events coincided with a real change in their habitual behavior.
`The cases of agreement included staying up late and going out
`at night, for example [40].
`One multimedia laboratory, NTT DoCoMo, has developed a
`system for modeling and recognizing personal behavior based
`on