27.1 Introduction

Each year, the number of people living with self-limiting conditions, such as dementia, Parkinson's disease, and mental health problems, is increasing [1]. This is largely due to individuals living longer and improvements in diagnosis and treatments. The number of populace living with dementia worldwide [2] is currently estimated at 35.6 million, and this number is set to double by 2030 and more than triple by 2050. Additionally, one in four people currently experience some kind of mental health problem each year [3]. Supporting these sufferers places a considerable strain on organizations such as the National Health Service (NHS), local councils, frontline social services, and carers/relatives [4]. In monetary terms, dementia alone costs the NHS over £17 billion a year [5], exacerbated by the cost of depression patients, which is predicted to increase to 1.45 million in the United Kingdom, adding a further £2.96 billion cost to essential services by 2026.

This figure excludes other mental health conditions such as anxiety disorders, schizophrenic disorders, bipolar-related conditions, eating disorders, and personality disorders. Effective around-the-clock monitoring of these conditions can be a considerable challenge and often leads to patients having to reside in care homes and other accommodations. A safe independent living environment can be hard to achieve [6]. Currently, 20–30% of individuals with dementia are living alone, yet no technology exists that enables the automated monitoring. The same problems exist for patients with depression. Additionally, the number of people living alone has doubled over the last three decades, amounting to one in three people in the United Kingdom and United States. This is a growing concern as solitary living is proven to produce increased number of patients with depression [7].

In addition, the need to detect accurately sudden or worsening changes in a patient's condition is vital for early intervention. Community mental health groups, crisis and home resolution teams, assistive outreach teams, and early psychosis teams all play a key role in preventing costly inpatient admissions. If any changes are not dealt with early, the prognosis is often worse, and, as a result, costs for treatment will undoubtedly be higher [8]. An early intervention approach has been shown to reduce the severity of symptoms, improve relapse rates, and significantly decrease the use of inpatient care. Evidence suggests that a comprehensive implementation of early intervention practice (EIP) in England could save up to £40 million a year in psychosis services alone. Being able to detect deteriorating conditions in dementia patients earlier enables physicians to better diagnose and identify stage progression for the disease. This enables earlier intervention for the illness before cognitive deficits affect or worsen mental capacity, supporting the individual and their family in adapting to the illness simultaneously.

Analyzing a patient's electricity using smart meters ensures for accurate around-the-clock monitoring of patients by identifying certain and often subtle changes in behavior. This motoring can be utilized not only for safety but also for enabling the prediction of immediate, mid-, and long-term prognosis. In addition, this type of active monitoring enables the detection of certain patterns and trends to help facilitate early intervention. Monitoring patients using this method offers a vast improvement over existing assistive technologies as monitoring the physical and mental well-being of a patient becomes possible. By the end of 2020, the government aims to install smart meters in every household throughout the United Kingdom. This is also true for a large number of countries around the world such as Italy, United States, Netherlands, and Australia. The challenge, therefore, is how to interpret, analyze, and make use of the data collected by smart meters and the wider advanced metering infrastructure (AMI) for use in such monitoring applications.

Smart meters provide granular energy usage readings at 30 min intervals. However some countries such as Canada set their intervals as low as 15 min, which can be used to monitor and profile consumers. Additional devices, such as smart plugs, facilitate the identification of electrical appliances being used in the home, along with their duration of use and the amount of electricity being consumed. This information is extremely beneficial in determining abnormal user behavior, such as a device being left on for atypically long periods or devices not being used at all.

27.2.1 Advanced Metering Infrastructure

The AMI provides bidirectional communication between the consumer and the rest of the smart grid. A smart grid is a complex modern electricity system [9]. It uses sensors, monitoring, communications, and automation to improve the electricity system. Smart grids fundamentally change the way in which we generate, distribute, and monitor our electricity. This enhanced communication removes the traditional need for energy usage readings to be collected manually. Instead, a robust automatic reporting system with greater granularity of readings is offered [10].

There are many advantages of deploying the AMI, some of which include reduced costs for meter readings (possibility to access meters otherwise difficult to attend due to position or security reasons), support for real-time pricing, increased fraud detection, and reduced read-to-bill time, to name a few. As part of the larger smart grid, the AMI can be broken down into three specific areas, each with their own roles and functions, as shown in Figure 27.1.

Home Area Network (HAN): The HAN is housed inside the consumer premises and is made up of a collection of different devices. Firstly, the in-home display (IHD) unit, which is the most visible and accessible part of the AMI. It provides the consumer with up-to-date information, in real time, on electricity usage, as well as the units of energy that are being consumed. Secondly, the smart meter provides real-time energy usage to both the consumer and all of the stakeholders. Smart meters are able to store 13 months of data, keeping a record of total energy consumption. In addition to smart meters, smart plugs enable the identification of the individual devices that are responsible for the reported energy usage. Additionally they provide detailed information about their duration of use.

Wide Area Network (WAN): This section of the AMI handles the communication between the HAN and the utility companies. The WAN is responsible for sending all polled meter data to the utility companies and other grid stakeholders, using a robust backhaul network, such as Carrier Ethernet, GSM, CDMA, or 3G [11]. The geographical location of the smart meter dictates what WAN technologies need to be implemented due to the constraints of certain communication technologies. The data aggregator unit (DAU) is the communication device that is used to collect the energy usage data from the home gateway or the smart meter. The acquired data is transmitted using one of the communication technologies mentioned above to the control center. The meter data management system (MDMS) is the central control center, which provides the storage and data processing capabilities for the obtained smart meter data. The MDMS also collects information regarding the status for the generation, transmission, and distribution of the wider smart grid.

Service users are a number of organizations and utility companies that have access to the data for analysis purposes. Energy suppliers communicate remotely with the smart metering equipment in order to perform a number of tasks such as taking meter readings, updating price information on the IHD, and identifying readings on a change of supplier or change of tenancy. Energy network companies access data to evaluate the loads on their network, at the local level, and to respond appropriately to loss of supply. Consumers can allow other organizations to have access to the data from their smart meter. For example, energy switching sites could use the accurate information generated from smart meters to establish the amount of energy used by the consumer and advise on the best tariff based on their individual energy requirements.

27.2.2 Smart Meters

Smart meters are seen as one of the most important components of the AMI and smart grid [12]. They are the foundation of any future smart electricity grid and provide consumers with highly reliable, flexible, and accurate metering services. Smart meters provide real-time energy usage readings at granular intervals to both the consumers and all of the smart grid stakeholders [13]. They obtain information from the end users' load devices and measure the energy consumption. Added information such as any home-generated electricity is provided to the utility company and/or system operator for better monitoring and billing process. This is achieved by monitoring the performance and the energy usage characteristics of the load on the grid.

Smart meters are able to store 13 months of historical energy usage, which allows for the creation of detailed energy usage profiles [14]. Currently readings are taken at 30-min intervals; however as these meters become more sophisticated, they are able to measure household power consumption at ever finer timescales [15]. Smart meters are able to report energy usage as low as 1 min intervals, even though this is not currently widely deployed due to the vast amount of data it would generate [16]. This is important when trying to identify individual devices and their duration of use. Figure 27.2 highlights the additional information obtained from increasing the frequency of reading. This can have a significant impact on the ability of identifying individual devices.

Smart meters are able to perform a wide variety of roles and bring many benefits over the traditional electricity meter. Some of these roles and benefits include the following:

• • Accurately record and store information for defined time periods (to a minimum of 30 min). This enables remote, accurate meter readings with no need for estimates [14]. As these meters become more sophisticated, they are able to measure household power consumption at ever finer timescales.
• • Offer two-way communications to and from the meter so that, for example, suppliers can read meters and update tariffs remotely [17].
• • Enable customers to collect and use consumption data by creating a HAN to which they can securely connect data access devices [18].
• • Communicate with microgeneration, home appliances, and equipment within the property [19]. Smart meters will be able to control smart home appliances and communicate with other smart meters within reach. This enables devices to be switched on when grid demand is low and turned off when demand is high.
• • Allow customers to collect and use consumption data by creating a HAN to which they can securely connect data access devices.
• • Enable other devices to be linked to the HAN, permitting customers to improve their control of energy consumption [20].
• • Support time-of-use tariffs, under which the price varies depending on the time of day at which electricity is used [21]. Energy prices are more expensive during peak times. Billing consumers by time, as well as usage, encourages them to change their consumption habits.
• • Support future management of energy supply to help distribution companies manage supply and demand across their networks [22]. This is achieved automatically through previously agreed demand response actions.

Figure 27.3 shows the different metering infrastructures; it highlights the key differences between the conventional energy meter and the smart meter. The newer smart meter removes the need for manual data collection. Instead energy usage data is collected automatically, providing instant readings and automatic billing.

Smart meters collect and upload a wide variety of data from consumer usage to power generation. This presents an opportunity to accommodate independent living while monitoring for safety. As previously stated smart meters will be installed in every UK property by the year 2020. Within this frame, the smart meter becomes a potential tool for the implementation of policies relevant to the community as generator and storage of information and sensitive data. Being able to interface with these devices enables a wide variety of applications and services. A selection of some of the data parameters that are collected are shown in Table 27.1

Table 27.2 shows a data sample from a smart meter during a single period. This sample shows the granularity of the data collected compared with traditional meters, where the readings are submitted collectively over a much larger period (e.g., monthly, quarterly, or yearly). It displays the data parameters obtained at each 30-min interval, totaling of 48 individual readings in a 24-h period (10 readings are shown). Customer key identifies the individual smart meter device within the AMI; time of reading indicates the time and date of the reading, while general supply highlights the amount of on peak electricity being used in kW h.

Figure 27.4 shows readings taken from an individual's smart meter over a single 24-h period. Each of the 48 individual readings represents the total amount of electricity being consumed in kW h at each 30-min interval. This frequency of readings makes it possible to identify certain daily activities as shown above. Any change in energy usage will enable the identification of any change in routine and habit.

In addition to smart meters, smart plugs that monitor individual electronic devices in the home can be used to establish a more accurate profile. Data from these smart plugs enable the identification of each electrical device and which appliance is responsible for the electrical load at each 30 min reading. Smart plug reading frequencies can be reduced from minutes to seconds to provide a more detailed analysis of a person's behavior while showing the exact duration of use for each electrical device.

Smart plugs can interface directly with the smart meter using ZigBee Smart Energy; the ZigBee Alliance forms a collection of device descriptions and functions that allow energy providers to manage and monitor energy loads to optimize consumption. Over a million ZigBee electric meters are deployed by many utility companies in the United States with UK Department of Energy and Climate Change (DECC) announcing SMETS 2, which cites ZigBee Smart Energy 1.x. Table 27.3 shows an example of the data generated by home plug readings over 1 h.

27.2.3 AMI Implementation Challenges

Using smart meter data to examine the behaviors of vulnerable people enables patients to live independently while safe in the knowledge that they are being actively monitored. This however presents many technical, ethical, and privacy challenges.

Firstly, the scale and size of data collected from smart meters and the AMI presents real and complex difficulties in terms of storing, structuring, and analyzing the data. New methods for analyzing and modeling data will need to be developed with the focus on using cloud platforms and data centers. Cloud platforms, such as Azure, currently have the ability to analyze large datasets and assign virtually unlimited resources in order to process the data in a timely manner.

Secondly, due to the vast scale of the smart grid, ensuring standards, interoperability, and continuity throughout the system is a challenge. This is largely due to the integration of interchangeable components from a variety of different providers [23]. Additionally, there is an ever-increasing interdependency between control systems, such as SCADA, and other commercial networks.

Thirdly, there are many ethical and privacy concerns associated with the smart meter role out, which could potentially leave consumers vulnerable to exploitation [24]. For example, criminals could process data that is generated by the AMI to identify when households are unoccupied, helping to facilitate burglary or some other crime. Additionally, being able to identify appliances would allow burglars to identify and target households with the most electronic devices. Criminal activities are not the sole concern however. Law enforcement agencies could exploit the data for a variety of purposes. These include categorizing properties being used for the purpose of producing drugs due to the implementation of heat lamps, hydroponics, and so on or proving/disproving premises occupation. Commercial entities, using targeted advertising at a specific household, or simply being able to know when someone is home to take a sales call, are further possible applications.

Being able to identify legitimate and beneficial applications is imperative for long-term success of the AMI. People justifiably have concerns on how their data will be utilized and accessed. Any misuse of data or incidents of security breach will lead to large-scale resistance of their use. This resistance is already being seen in the United States, for example, where groups of consumers are actively refusing smart meter installations [25]. In Philadelphia, some customers have opted out of having the smart meters installed because of its two-way communication capabilities. Residents are concerned that government organizations will use the information for spying on their activities. These concerns will need to be addressed by providing transparency on how the collected data will be used. In one significant case, the First Chamber of the Dutch parliament rejected two smart metering bills in 2009 because of privacy concerns, forcing the government to add significant privacy protections to the revised bills that were passed in 2011 [26].

27.2.4 Patient Behavior and Uses

As previously discussed, smart meter data enables active in-home monitoring. By analyzing past behaviors, an improved prediction of worsening conditions is made possible. Analyzing the data in this manner facilitates early intervention and an improved outcome for the patient by ensuring that their medical and care needs are sufficient. Being able to detect and predict these changes requires a detailed understanding of the symptoms and behaviors that are expected for each condition. The following section discusses the features for each condition and the monitoring applications.

27.2.4.1 Active Monitoring for Behavioral Changes with Dementia

There are a common set of features of Alzheimer's disease and other dementias. These include agitation, anxiety, depression, apathy, delusions, sleep and appetite disturbance, elation, irritability, disinhibition, and hallucinations [27]. The severity of each symptom differs at various stages of the disease. Therefore, any system would need to be fully adaptable to these changes, as patient's progress through the different stages of the illness [28]. Particularly, as later stages of the disease are regarded as important as (if not more important than) the earlier stages, they tend to harbor unique characteristics and events, which occur. These affect the lives of the patients and their carers. Behavioral problems, such as agitation, become more pronounced in the later progressive stages of the disease.

Currently, the Mini–Mental State Examination (MMSE) [29] is used by clinicians to help diagnose dementia and assess its progression and severity. The 6-Item Cognitive Impairment Test (6CIT) is also used for similar purposes [30]. These tests would need to be performed at regular intervals to identify the stage of the disease. The process involves identifying the correct characteristics from patient datasets.

Figure 27.5 highlights the MMSE in more detail, showing the need for changes in the monitoring techniques, as the severity of the disease increases. The feature vectors would need to change regularly, along with the algorithms used, in order to maintain system accuracy for each stage of the disease.

Figure 27.5 displays, although each person with dementia experiences the illness in their own individual way, their common traits in behavior that can be identified [31]. These symptoms are listed below and can be detected through their electricity usage:

• • Loss of mobility – People with dementia gradually lose their ability to perform everyday tasks. They will usually perform tasks at a much slower rate and are more likely to fall due to a reduction in mobility.
• • Eating – People with dementia often lose weight due to a number of factors, which range from forgetting to make meals to finding it hard to eat. Sufferers often need constant help and encouragement to ensure they consume enough food and liquids. Failing to monitor and ensure that the patient consumes regular meals can increase the likelihood of further falls and other complications.
• • Unusual behavior – A sufferer's behavior can drastically change especially in later stages of dementia. A person might become more agitated and confused in the late afternoon and early evening, so extra attention is needed during these periods. In addition, some sufferers might experience hallucinations and delusions and may alter how they interact with their environment. Some patients might become more restless because they often need more physical activity; by contrast the patient might have long periods of physical inactivity.
• • Side effects of medication – Drugs that are prescribed to dementia sufferers can have severe side effects, which can often increase a person's confusion. Patients can often be prescribed with doses that are too high or drugs that are no longer appropriate to the patient's needs.

Being able to detect changes in a patient's habits, routines, and features, as highlighted above, will ensure the active monitoring of their well-being.

27.2.5 Current Assistive Technologies

In this section, an investigation into current devices (assistive technology), which enable or aid independent living for people with self-limiting conditions and their associated benefits, is presented [37]. The term assistive technology refers to any device or system that allows an individual to perform a task that they would otherwise be unable to do or increases the ease and safety with which the task can be performed. There are many benefits associated with assistive technologies from the ability to promote independence and autonomy, both for the person with dementia and those around them [38]. Additionally they help manage potential risks in and around the home.

Surprisingly [39], assistive technology covers a wide range of equipment from simple low-tech devices, such as handrails and grips, to high-tech equipment that includes power wheelchairs and robots. Limited technology exists, which enables proactive monitoring for people living with these conditions. Monitoring technologies include video monitoring, fall detectors, and health monitors [40]. Environmental (passive) motion sensing devices, such as a smart floor, have pressure sensors located in the floor that track the movement and location of the resident within the home. While most of these solutions help with the monitoring of physical impairments, none cater for mental and inactivity monitoring, which is crucial for conditions like Alzheimer's disease and depression.

27.3.1 Case Study

In this section, a case study is presented for one application and condition, which highlights an individual's habits and routines using smart meter data. Being able to recognize changes and patterns in behavior aids in the ability to ensure a better outcome by ensuring a more personalized healthcare treatment/package for the patient. Normally, people who suffer mental illness exhibit certain behavioral changes during periods of heightened severity. One of these most usual changes is the alterations in sleep patterns. Typically, a sufferer will awaken much earlier than normal. Figure 27.7 shows the total energy consumption between the hours of 1:30 a.m. and 4:00 a.m. over a 1-year period. Each of the larger peaks displays an increase in electricity during the early hours of the morning.

This type of result highlights changes in the person's sleep patterns, which could provide an indication to a healthcare professional whether intervention is required. This method can also be used to identify long periods of inactivity that keep on recurring. In many conditions early intervention is important where inactivity is prolonged and reoccurs frequently.