Rethinking the role of monitoring in technology-enabled care

As TEC services move away from focussing solely on detecting and responding to alarms and become more data-led, the role of monitoring, and the coordination of interventions, will need to change. This will have an impact on the role of the monitoring centre, the traditional hub of existing telecare services, potentially changing their operating model.

Alarm-based telecare remains the most common form of technology enabled care in use in the UK today supporting an estimated 1.8 million people. It focuses primarily on managing the risks involved with people living alone either in their own home or within a grouped living scheme by providing an efficient means of detecting and responding to an emergency using the telephone network. More recently, it has been extended using mobile telecare devices to provide basic reassurance for when people are out and about in their community.

Either way, the service delivery model usually involves a request for help being received at a monitoring centre by a call handler who then establishes the precise reason for the call, provides reassurance, and determines (and sometimes organises) the best response according to the identified need.

This could range from providing information, contacting a family member, arranging for a carer or responder to attend, or calling for an ambulance or other emergency services. The call handler therefore performs the role of an intelligent filter – to triage the initial call, followed by a switching/notification service – to determine the best possible reaction and response personnel. They will then typically monitor the situation until it is safely resolved, when the call can be closed, and the outcome recorded. It is a reactive service.

As we have discussed in our previous articles, next-generation technology enabled care (TEC) applications are shifting away from alarms and focussing more on how smart consumer products can support people in their day-to-day lives as components in more blended, or hybrid, service propositions. Their role includes monitoring the activities of an individual in their home to identify potential problems or significant changes that could indicate that an intervention or a change in the type or level of support provided might be beneficial to achieve improved outcomes.

However, there is often a level of uncertainty associated with this type of monitoring; it relies on alert notifications, rather than alarms, and requires a more considered and agile response which may not be urgent as might be the case for an emergency.

Rethinking Monitoring

Traditional monitoring services were a product of user need (being able to easily call for help in the event of an emergency) and technological capabilities and limitations (simple manually triggered alarms and analogue telephony). The current reactive alarm-based monitoring model is well established and has its origins in warden-controlled schemes from the last century. These were extended to provide cover to people living in their own homes, made possible by the widespread adoption of landline telephone connections in the 1970s. The exceptional occurrence of alarms and the dispersed nature of the user-base necessitated the use of a local/regional alarm receiving centres (now known as ARCs) to receive alarm calls and provide help as required to typically thousands of monitored connections.

As we move into the digital age, this ARC-based model that has been in place for many decades, is ripe for disruption due to several factors:

• Economies of scale – the bleak financial situation faced by many district councils and local authorities mean that they are increasingly looking to review how they provide non-statutory services. The alarm-based model is dependent on 24-hour staffing, though the number of emergencies during the night and at weekends can be small. Inflation has added to these costs. Consequently, there is a growing trend for consolidation in the monitoring market, as smaller locally run monitoring centres are closed and replaced with a much smaller number of outsourced monitoring services provided by larger national providers.
• The shift to mobile – at the end of 2022, 87% (around 10.9 million) of people aged 65 & over in the UK used a mobile phone, of which 58% (around 7.3 million) were using a smartphone. The convenience and affordability of smartphones is such that they are replacing the need for a fixed line connection for many. It also means that they have a means of contacting a family member easily, including when out and about. As of 2020, only 73% of households had a landline phone that could be used to make and receive voice calls, with the overall trend being a downward one. The perceived complications, as recently covered in the media, of transitioning to digital voice services may speed up this downward trend.
• Supply chain interoperability – an expansion in the market led to demands for a greater variety of dispersed alarm receivers and more sensors and peripheral devices to manage an extended range of needs and risks. No single supplier has been able to provide a full set of devices. Interoperability is now essential to ensure that customers can build solutions without having to worry about compatibility issues.
• Digital transformation – the digital transformation of the UK’s telephone network resulting in the retirement of PSTN and ISDN services has significant implications for equipment in the home, sheltered housing schemes, and monitoring centres. Addressing the various issues requires significant investment in new technologies to ensure that monitoring services can receive and process connections from digital alarm units, whilst also maintaining the ability to support legacy analogue alarms during the switchover period. As such, it will likely accelerate the consolidation in the monitoring centre sector, mentioned above.
• Industry-standard alarm protocols – a focus on proprietary alarm protocols such as TT21 has been replaced by a requirement to support common standards such as BS 8521. This has been extended into the digital domain with both TS50134-9 (CENELEC) and BS8521-2 (Now-IP) protocols being preferred by many services, although some proprietary protocols are still in use. Common standards allow monitoring services to support equipment from multiple suppliers more easily.
• Family monitoring – the lack of a formal responder service in many areas means that it is often a local family member or a friend who is required to attend if there is a problem that may not require attendance by the emergency services. This, together with the increasing cost of telecare service provision and the availability of products that allow people to easily perform their own monitoring and respond to alert notifications on their smartphone, means that the case for needing a separate monitoring service is not always clear. Moreover, the difficulty or perceived stigma associated with accessing local authority provided services, means that this ‘self-monitoring’ approach is likely to be a growth area. Likewise, local authority services wanting to upstream interventions will likely signpost individuals to self-care approaches that can make use of alternative TEC monitoring technologies.
• Diverse data sources – in addition to home sensors, individuals may use wearable devices and allow access to electronic records, data from video and telephone calls, and use of proactive calling approaches to be available for analysis and response.
• Data-led preventive services – the shift to data-led applications and services means that the role of a monitoring service needs to be re-defined. Where alerts can be handled automatically by family members, it may be that monitoring services become more of a backup option for when family members are unable to respond, such as when they are away on holiday.

However, the other aspect involves how long-term data trends are analysed and acted upon to provide a truly holistic and preventive approach. There are also population-based analyses that may be of value to local and national healthcare commissioners and providers. New markets such as monitoring the condition of housing and their environment, and the short-term monitoring of the health and wellbeing of people recently discharged from hospital, are also opening opportunities for monitoring centres to expand their customer-base.

All the above means that the appeal of the ‘linear’ ARC service model is waning and is likely to be replaced by a new distributed model, Figure 1, which offers both an emergency alarm approach (to the left of the user), and a preventive approach (to the right) where new and emerging services and agencies will offer interventions aimed at delaying or preventing future threats to the health and wellbeing of monitored individuals. This is a more agile model that adapt to support a wide variety of support arrangements.

Data-led TEC Technology ‘Stack’

Data-led TEC applications require more advanced data processing to interpret data as meaningful events and to create actionable insights. This is necessary:

• for data analysis,
• to manage uncertainty,
• to monitor long-term trends,
• to identify possible alert conditions and
• to present all this information to a range of users in a simple and meaningful way.

This results in a technology ‘stack’ that is required to achieve these key features, Figure 2.

We have covered the technologies in the home in previous landscape articles and will now focus our attention on the monitoring platform and associated services.

Cloud Services

Cloud services are applications and technical infrastructure that reside on the Internet, and which are therefore accessible wherever there is suitable connectivity. The digital transformation agenda means that TEC applications are increasingly making use of cloud infrastructure to provide a range of services from data storage for dispersed sensors all the way through to hosted monitoring platforms. The latter are increasingly being deployed as they offer several advantages over the traditional self-hosted or ‘enterprise’ approach:

• Applications and data are hosted in secure data centres with dedicated backup and redundancy to provide reliable and dependable systems at scale.
• This results in simpler IT requirements for the monitoring provider (no need for in-house servers with local database management, computer telephony integration, business continuity, backup management, cybersecurity, etc.).
• It supports a move away from the need for comparatively large upfront capital expenditure to a simple revenue-based approach using a subscription-based model (monitoring ‘as-a-service’). As well as upfront savings from not having to invest in expensive hardware (and its eventual disposal) and associated software, there are also recurring savings from no longer requiring ongoing maintenance and support.
• Access to the monitoring platform is possible using a computer or laptop with network connectivity and either a dedicated software application or increasingly a simple web browser. It encourages ‘monitoring-on-the-go’ by staff who will be responding to alerts.
• Subscription-based access means organisations can more readily upscale or downscale their monitoring capacity according to changing demands. They can also more easily upgrade to the latest version of the platform, and incorporate optional features as required.
• This flexible access to the monitoring platform enables more versatile remote/home and mobile (app-based) working arrangements, with alert notifications sent to the most appropriate responder based on criteria such as skill set, availability, location, user subscribed services, etc.
• Access to data and applications can be more easily shared across stakeholders using API’s.
• The need for large, secure premises capable of hosting all the IT equipment and call handling staff required for 24/7/365 operation is reduced.
• Operations can be dispersed with call handling performed by mobile staff in their own vehicles enabling them to escalate response by visiting a property during their shift.
• Services may hand over monitoring to national centres at night or at weekends and operate local interventions only during office hours.
• Flexible operation and partial outsourcing might provide a lifeline to smaller services that would not be possible if all monitoring centres had to provide large-scale monitoring services. Smaller, local monitoring services can offer more limited or specialist function, especially for operation during office hours only.

Data Analytics

Data-led TEC works by sensing and analysing an individual’s movements, activities, lifestyle, and other health-related parameters. The technology was originally developed for people living alone and many applications are based on monitoring a person’s activity within the home (sometimes referred to as lifestyle monitoring). Activity data is derived typically from one or more of the following sources:

• Movement, and location, of individuals in the home – typically using PIR motion sensors (or IoT presence sensors) or from other sensors that have a room location assigned to them (e.g. a fridge door monitor registered as being in the kitchen).
• Change in status of doors – magnetic contact or vibration sensors that monitor the use of doors, drawers, cupboards, fridge/freezer, larder etc.
• Interaction with objects in the home – electrical use sensors that monitor the use of appliances such as the kettle and microwave oven and/or movement/presence sensors that detect interactions with objects such as beds, chairs, medication boxes and blister-packs or toilet, etc.
• Ambient environment – room temperature, humidity, ambient light levels, but also noise levels, air quality, and the concentration of gases such as carbon dioxide that can be used to monitor occupancy.
• Home energy use – total electricity, gas, and water consumption throughout the home (i.e. not on a per-device basis).
• Telephone use – the number of incoming and outgoing calls and their duration.
• Personal activity – wearables that monitor the activity levels of an individual (e.g. step counts, stairs climbed, calories expended); and
• Physiological parameters – wearables or other devices that can monitor vital signs such as body temperature, heart rate, breathing rate, sleep duration and quality, as well as accurate blood pressure measurements, and continuous blood glucose monitoring.

These data are processed and combined to derive useful information about the behaviour and health status of the person being monitored. This can be thought of in two stages:

• matching data to events, activities, or domestic tasks of interest; and
• looking for significant patterns in how certain data trends change over time.

This processing can be relatively simple, or it can benefit from the use of advanced and/or machine learning algorithms to improve the accuracy of the information derived. The way in which activities are identified from data varies depending on the data available and the way in which it is processed.

It may be apparent that the more data points that can be assigned to a particular activity, the more confidence can be assumed for the derived assumption. However, in practice only a small subset of devices is typically installed in a person’s home. This is to reduce costs and the perceived levels of intrusion. Therefore, more longer-term statistical analysis is often necessary to build up a profile of behaviour over time to learn the most likely reasons for each sequence of events. It is the job of machine learning algorithms to reduce the uncertainty involved with gaps in these data and to ensure that false negatives and positives are kept to a minimum.

Data analytics is also required to process how significant activities or monitored parameters are changing over time. This can be used to help calculate important thresholds for generating an alert for when a change is thought to be significant. These will then be used in conjunction with a rule-based ‘notification engine’ to help bring key actionable insights or issues to the attention of an interested party, such as an observer or a responder.

Rules-Based Notification Engine

The notification engine is responsible for raising alerts if monitored parameters move outside of the normal range to a point where they are of significance, warranting further investigation (an ‘amber alert’) or are so far outside agreed norms to warrant immediate concern (a ‘red alert’). The level of sophistication provided by notification engines vary. Basic alerts can be set up using procedural rules with pre-defined thresholds to offer a basic level of reassurance in the absence of additional details (e.g. “If the user does not get out of bed by 9am, then raise an alert”). Each alert needs to be directed to an interested party, and a protocol established for response and, potentially, escalation.

More sophisticated approaches involve modifying initial baseline thresholds by learning the behaviour of the individual so that, for example, the time in the previous example can be set to a more appropriate value based on monitored data and with a tolerance set to a statistically relevant period. Thus, the rule might become “If the user does not get out of bed between 8.36 and 9.14 am, on a Monday, Thursday or Friday, then raise an alert” – whether this is defined explicitly or the result of a machine learning algorithm. The same is true for other monitored data so that more intelligent thresholds based on learned patterns of behaviour mean that the system can respond more accurately and speedily to abnormal situations. This processing relies on the use of statistical modelling, which is why many automatic alerts based on significant deviations from normal behaviour are not enabled until sufficient data is available. The longer that a system is in use for, the more reliable the alerts should become. Alert notifications may be generated using a variety of approaches including on-screen notifications, SMS or other instant messaging services, in-app notifications, e-mail and to other platforms via third-party APIs.

Data Visualisation

Most leading TEC monitoring platforms provide a data dashboard for offering important insights about the status of an individual or a group of individuals (e.g. all active monitored individuals or those limited to a geographical area/customer-base). Key parameters are often highlighted so that they can be monitored for instant review so that monitored values that have a range are often marked as being normal (green), worthy of investigation (amber) or significant, requiring intervention (red). Further investigation is possible by drilling down to view and consider the underlying data. An example dashboard UI is shown in Figure 3, which is an example from the Archangel platform.

 

Monitoring & Response

Monitoring involves responding to alerts and calls, as well as observing and interpreting changes and trends in data. For a family member, this might involve checking in on the data dashboard for a loved one each evening to confirm that all is well – and responding to any alerts that are raised (e.g. “Check-in: Mam hasn’t made herself a cup of tea by her usual time.”). For a monitoring service, this might involve anything from telecare alarm calls through to an alert that suggests that Mrs Jones is at an increased risk of falling and would benefit from a review. The response required for both types of events are different and will require different response protocols. There are a multitude of products available that support this kind of functionality. Each uses their own cloud-based system, and each will have varying success and longevity in the market. Competition is rife but not all providers will survive, making the choice of which platform to use complex. The question of how to manage deployed devices and continue to access long-term data if a particular platform is withdrawn from the market represents a significant challenge that commissioners need to address before investing in digital monitoring systems.

Monitoring Platforms

It follows from the above that choosing the right monitoring platform is critical for implementing successful digital TEC services. It is the hub of the entire service, responsible for collecting and interpreting data, triggering alerts, and ultimately ensuring the safety and wellbeing of users. Choosing the wrong platform can result in problems with supporting new digital and legacy analogue telecare products, adopting new sensor/IoT technologies, missing emergencies, inaccurate data interpretation, ineffective or late interventions, and problems with sharing information to enable effective collaboration with partner organisations.

 

Distributed monitoring & response capabilities

 

Technology enabled care (TEC) services have evolved over the past 30 years from basic, active alarm systems: first to passive systems based on smart sensors and, more recently, to 3rd and 4th generation systems that use activity monitoring data as well as alarms. The monitoring centres employed as the hub of operations changed little during this time, except for more standardised reporting of performance and improved database management, which were possible through general improvements in IT and telecommunication systems. The retirement of the analogue Public Switched Telephone Network, and the introduction of an all-digital (or IP-based) arrangement, have provided a boost to the transformation agenda for services that will enable higher levels of integration through 4G monitoring platforms and future 5G designs. These will enable a new era of Ambient Assisted Living and digital health which should lead to system-wide efficiencies and more effective independent living at lower cost.

It is now pertinent to consider the required developments of future platforms that will be compatible with changes in the TEC landscapes that we have considered in previous articles. These will be required to enable the advanced data collection and analytics that are envisaged, and to avoid issues such as a lack of interoperability that have limited previous generations of telecare systems.

Platforms will also need to consider the impact of financial and resource pressures on local authorities and on other commissioners or providers of TEC services. These are already leading to consolidation of some monitoring services resulting in the closure of many small centres and the creation of large national centres. The latter may be less effective in responding to local and individual needs unless their systems allow more agile, dispersed operational models.

This should lead to a specification for a next generation monitoring centre that is agnostic to the field equipment and to other sources of monitored data that are likely to be integrated to improve the understanding of everyone’s circumstances and status. We have, in this article, identified several of the key high-level requirements. Commissioners and TEC monitoring service providers should consider these requirements carefully before investing in a new platform.

This is the fifth of a series of six articles describing the changing TEC landscape. The final article will consider applications for the integrated TEC systems that we believe will soon be possible when other technologies mature, and when the benefits of Integrated Care Systems become more apparent.

Article written by T-Cubed