2

Mobile Monitoring and Intervention (MMI) Technology for Adaptive Aging

Neil Charness, Walter R. Boot, and Nicholas Gray^1,2

INTRODUCTION AND OVERVIEW

Mobile monitoring and intervention (MMI) technology offers a promising way to provide interventions tailored to individuals and their current context. Ideally, the system would be capable of monitoring relevant aspects of physiology and behavior, making intelligent predictions about when and how to intervene, and then delivering timely interventions. This chapter outlines critical issues to consider for MMI, including whom to target, what measures to target, where to monitor and intervene, when to monitor and intervene, and how to monitor and intervene. We also discuss attitudinal barriers for aging adults and the challenge of promoting adherence to MMI systems.

We review recent studies, most employing smartphones with small, unrepresentative samples that include monitoring and prediction, though not intervention. Although there are many commercial apps for smartphones aimed at supporting health, they have unknown efficacy and generally are not well designed for aging adults, failing to consider changing needs for the young-old, middle-old, and old-old age groups. We find that MMI technology for aging adults is in its infancy, with few good examples showing efficacy or cost effectiveness. To move such technology toward maturity we

___________________

¹ Florida State University. Address correspondence to: Neil Charness, Psychology Department, Florida State University, 1107 West Call Str., Tallahassee, FL 32308-0844; charness@psy.fsu.edu.

² This work was supported in part by NIH/NIA 4 P01 AG 017211, Center for Research and Education on Aging and Technology Enhancement (CREATE).

suggest supporting studies that can enroll larger, more representative samples, and that can track system performance over an extended period (years) to assess efficacy for managing chronic conditions. Such studies might benefit from cooperation between federal agencies such as the National Institute on Aging (NIA) and the National Science Foundation (NSF) and might consider making use of existing longitudinal panels.

FRAMEWORKS FOR MONITORING AND INTERVENTION FOR ADAPTIVE AGING

Our aim is to provide frameworks and recommendations for research on MMI systems by relying on recent (2015+) studies and reviews that assess efficacy for promoting adaptive aging. We focus primarily on studies of aging adults. We define mobile technology as devices that are wearable (e.g., internally: cardiac pacemaker; externally: smartwatches) or portable (e.g., smartphones, tablets that can fit in clothing or in accessories such as purses). In this chapter, we first introduce a framework for identifying the challenges for deploying MMI systems, then discuss attitudinal constraints on adoption. We then discuss frameworks for MMI, focusing on measurement, prediction, and intervention. We evaluate existing mobile apps and how they might promote adherence for diverse aging populations. This chapter ends with a discussion of how the RE-AIM framework can guide the development of MMI systems and closes by outlining potential research priorities.

Sensor-based monitoring technology, both fixed and mobile, offers advantages and disadvantages for intervening to promote improved wellbeing for our aging population. Unlike early “one-size-fits-all” interventions in behavioral clinical trials (e.g., Ball et al., 2002), sensor-guided interventions can generate tailored actions (e.g., Lustria et al., 2013). Usually fixed-location sensor systems (e.g., smart home sensor arrays) have the disadvantage that the user must be in a fixed location, though it is possible to envision blended fixed and wearable systems (Skubic et al., 2014). A significant advantage of MMI is that the system can move with the person. A smartwatch monitoring movement can prompt an immobile wearer to move after a lengthy interval of sitting no matter where they are (home, senior center). A significant disadvantage for MMI is that users must continually wear or carry devices on their person and keep them charged (Reeder and David, 2016).

Table 2-1 lists some of the challenges that arise when making the decision to deploy MMI technology.

Some questions relate to the ethics of MMI—that is, whether (“why” and “what”) and under what circumstances (“where” and “when”) MMI might be initiated. The unit of analysis is important (“who”), usually taken to be the monitored person, such as an older adult living alone. But that

unit of analysis may miss the person–family and person–community contexts for MMI (see the chapter by Fingerman et al.), in line with the finding that caring for family members is a primary human social motivation (Ko et al., 2019). Limiting consideration to the older adult (and family) may also miss the issue of bystander capture: people being monitored who did not consent to being monitored.³ The methodology for monitoring (“how”) is dealt with in other chapters in this volume.

TABLE 2-1 Challenges in Mobile Monitoring and Intervention (MMI) Research and Practice

Underlying many of the questions is consideration of privacy: whether older adults wish to be monitored and if so, what aspects of their behavior/physiology should be allowed, and how monitoring should occur. A population-representative survey of Americans found that older adult cohorts are more aware than younger cohorts about government monitoring but are less likely to view as “very sensitive” contents of email, text messages, and health information, and equivalently less sensitive about their Social Security number (Madden, 2014). Older cohorts are also less likely

___________________

³ An example from one of our monitoring studies (Evans et al., 2016) is a worker who came into a telehealth-equipped home that was monitoring a heart failure patient. He stepped on a wireless weight scale and triggered an alert because of the increased weight over the patient’s baseline.

than younger adults to take appropriate measures to protect their privacy online, such as asking to have information removed, or anonymizing postings (Madden, 2014). A similar population-representative survey showed that older cohorts on Facebook are less likely to change their Facebook privacy settings: 33 percent of those age 65 and older have changed privacy settings compared to 64 percent of those age 18–29 years (Perrin, 2018). A year following entry into a study of unobtrusive monitoring (ISAAC), nondemented older adult volunteers and older adults with mild cognitive impairment (MCI) showed more concerns with privacy (concern that their information could be exploited) than at entry (Boise et al., 2013). However, 72 percent of participants still showed acceptance of monitoring.

If everyone valued privacy more than any potential gains from monitoring, there would be no basis for developing systems that might provide other benefits, such as prolonging independence or preventing harm. A survey of a diverse sample of aging American adults (45 years and older), showed a willingness (particularly among those with disabilities) to trade off privacy in favor of maintaining independence even for rather intrusive monitoring options, such as cameras (Beach et al., 2009). Still, in terms of sharing information from monitored activities, participants indicated they were more willing to do this with family members and health care providers than with researchers and least willing for insurance companies or government. There appears to be some generalizability across populations. In a representative Swiss survey, 57 percent of those age 50 and older who tracked health data (28% of the sample) were willing to share data with researchers (Seifert, 2018). Such willingness to share data provides constraints on how MMI systems might be designed.

In summary, privacy concerns need to be addressed to encourage aging adults to adopt and use MMI systems. Adoption of “Fair Information Practices” such as the eight principles in the OECD Privacy Framework (2013) is one approach. Another related approach is to provide people with granular control over release of captured information (Caine and Hanania, 2013).

Age and Technology Attitudes

Attitudes toward health monitoring technologies differ across age groups such that older adults tend to be more accepting than younger adults (Beach et al., 2009). Also, they tend to be primarily concerned about self-efficacy, or perceived ability to use the system (Lv et al., 2012).

Although older age had been associated with greater openness to adoption of health monitoring technology, when accounting for disability status, the effects of old age on openness are much smaller than those of disability status (Beach et al., 2009). If, therefore, the imminent threat of losing health or independence is one of the main motivating forces behind adoption of

health monitoring technology in old age, then preventive interventions for older adults may prove to be the most difficult to stage, as they would be met with the most hesitation. Without an obvious and apparent cause for concern, older adults may be reluctant to accept a new technology-based intervention, even though they remain at higher risk of health decline.

A variety of technology adoption models and variants, such as the Technology Acceptance Model (TAM: Davis, 1989), Universal Theory and Acceptance and Use of Technology (UTAUT: Venkatesh, Thong, and Yu, 2012), and the Senior Technology Acceptance Model (STAM: Chen and Chan, 2014), propose that adoption and use of technology depend on the trade-offs between benefits (e.g., perceived usefulness) and costs (e.g., perceived ease of use) as represented in such models (Charness and Boot, 2009). With respect to predicting concerns and actions by people for security and safety online, the protection motivation theory (Tsai et al., 2016) is also a useful framework.

One recent technology adoption model relevant to MMI is the smart wearable acceptance model (Li et al., 2019), which incorporates additional factors such as compatibility with existing electronics, perceived stigma, device performance (e.g., reliability), and health status. Challenges for compatibility with existing electronics might arise, for example, when trying to switch a user from their preferred smartphone to one with a different operating system. Given that older adults learn at about half the rate of younger adults (Charness et al., 2001), asking them to learn a new operating system may result in poor enrollment in, and adherence to, an MMI system.

Intervention Framework

If we assume that older adults, who normatively have a variety of chronic conditions and impairments (Buttorff, Ruder, and Bauman, 2017), are willing to be monitored, and that systems can be devised that provide for adaptive interventions, what type of interventions are people likely to accept? One proposed hierarchy is “PRAS”—prevention, rehabilitation, augmentation, substitution—(Charness, 2019), which suggests that if prevention is insufficient and an impairment develops, people will prefer rehabilitation first, then augmentation to current capabilities (assistive devices, such as walkers, hearing aids), and lastly substitution (e.g., prosthetics that replace a failed function, such as pacemakers, cochlear implants).

STATE OF THE SCIENCE FOR MEASUREMENT, PREDICTION, AND INTERVENTION USING MOBILE SYSTEMS

It is worth noting that any MMI system (a good example of a classical information processing system: Newell and Simon, 1972) will have multiple

components, including sensors, processors, algorithms to interpret sensor data, transceivers (transmitters and receivers), and data storage capabilities (see chapter by Cook). If intervention capabilities are built in, the system will have actuator components that can alert or communicate with the recipient (usually visual, auditory, and haptic output capabilities). A growing platform for monitoring health is the smartphone, which helped initiate the field of mobile health, or mHealth (see chapter by Murnane and Choudhury).

We did not locate any studies of MMI systems that incorporate the full chain of measurement, prediction, and just-in-time intervention for anyone, let alone older adults. A model system illustrating the full chain would be a cardiac pacemaker device. It monitors heart electrical activity, decides that it is irregular, and generates just-in-time pulses to regularize heartbeat). In the absence of studies looking at the full chain, we examine issues around each of the components, discussed next.

Measurement

Much of the literature concerning measurement capabilities of MMI technologies that we uncovered consists of feasibility pilot projects aimed at developing MMI technology systems. Many of these programs do not test such technologies with older adults, probably because of concerns with safety during simulated fall testing (studies that ask people to simulate the range of fall types) and for convenience of development (e.g., use of a student dormitory for Radio-frequency identification [RFID] tag testing). These problems can be seen in a recent review of wearable sensors and Internet of Things (IoT) monitoring for older adults (Baig et al., 2019). The review by Baig and colleagues indicates the range of target behaviors for measurement (the “what” question in Table 1). Those authors found 14 studies (from 12 projects) between 2015 and 2019 that met inclusion criteria from an initial set of 327 studies. Seven had a focus on fall detection using wrist-worn devices or RFID tags. Others concerned monitoring Activities of Daily Living (ADLs) using smartwatches, smartphones, and smart insoles. Other studies reviewed used smart home environments with passive sensors to monitor ADL and Instrumental Activities of Daily Living (IADL) activity. Lastly, geriatric depression and dementia detection (through classifying “forget” events with front door openings) were the goals of two of the studies.

Based on examining some of the studies in that review, we suggest that future measurement system development for MMI systems include older adults in both the development and testing phases, though this may prove problematic for fall simulation studies.

Prediction

A largely untapped research area is prediction/inference using mobile technology for older adults. By fusing data across time from multiple sensors and including data from active monitoring components, such as ecological momentary assessment (EMA) surveys, inferences can be made about behavior patterns (Harari et al., 2016).

By fusing data intelligently, systems can generate “mood sensors.” One study invited people to respond to EMA prompts about current mood (Sandstrom et al., 2017) and used phone sensors to determine where they were or directly queried their location with an EMA probe. That study relied on the general public (Android smartphone users) downloading an app (n = 12,310) and, given age-related technology adoption lag, enrolled a sample where at least 78 percent of those who reported a birth year were below the age of 45, a young to middle-aged sample.

More sophisticated inferences have been drawn through modeling, using various classifier algorithms and deep learning on data sets that contain large amounts of temporally tagged personal data in order to forecast depressive affect in young and middle-aged adults (Suhara, Xu, and Pentland, 2017) and loneliness in older adults (Sanchez et al., 2015). However, having to use a supervised machine learning procedure (see chapter by Rajkomar) somewhat limits the scalability of the approach, because of the need to have a human in the loop to label/classify patterns.

Intervention

Behavioral research studies we reviewed that use mobile device data typically do not intervene based on building up behavioral prediction models of study participants. Intervention is a logical next step. Perhaps because of lack of federal regulation, commercial enterprises have already entered the intervention space. Facebook experimentally manipulated mood for hundreds of thousands of its members by changing the information that a user saw in their news feed (Kramer, Guillory, and Hancock, 2014).

Nonetheless, once a model has been validated—for instance, that depressive affect has been detected and that it is predicted to worsen in a few days (e.g., Suhara et al., 2017)—it would make sense to provide referrals to professionals, or as research and technology advance, instantiate validated interventions, particularly to head off conditions that are potentially life-threatening. One such example is a model that predicts that a suicide or homicide attempt is likely, and intervenes accordingly by providing immediate access to a therapist. Suicides show a sharp increase at older ages for men, and older cohorts have also experienced some of

the largest suicide increases between 1999 and 2017 (e.g., > 50% for age 45–64; Hedegaard et al., 2018).

Another behavioral domain where prediction and intervention might be valuable for older adults is falls, given that about 29 percent of older adults reported experiencing a fall in the past year, and about 37 percent of those falls were serious enough to require medical treatment (Bergen, Stevens, and Burns, 2016). Balance and gait can be monitored and risk of falling assessed and detected (e.g., for fixed sensor systems: Rantz et al., 2015). If predicted risk rises above some threshold, the system could prompt the monitored person to seek help, or possibly, could provide validated rehabilitation exercises.

Loneliness and social isolation might present another such domain for MMI. About 20 percent of adults in the U.S. (16% of those age 65 and older) and in the U.K. report significant loneliness (DiJulio et al., 2018), with death of a loved one and health problems given as the top two reasons for loneliness. If a trend that indicates significant loneliness is detected, interventions might be offered via a software suite that aims to improve social connectivity, such as in the PRISM clinical trial (Czaja et al., 2017).

Another area for MMI is the management of chronic conditions. About 81 percent of those age 65 and older have multiple chronic conditions (Buttorff et al., 2017). Total population prevalence was about 60 percent for one or more such conditions in the U.S. A study of heart failure (Evans et al., 2016) is an example of where smart monitoring (examining data to detect deviations from baseline for blood pressure, weight, heart-failure questionnaire items) was used to generate text messages to home health nurses who contacted participants.

MMI might also provide help in managing medication schedules. The greater the number of chronic conditions, the greater the number of prescriptions (Buttorff et al., 2017), possibly leading to complicated medication schedules, though medication adherence is sometimes better in older adults than middle-aged ones (Park et al., 1999). Monitoring (e.g., smart caps for bottled prescriptions) and intervention (prompts to the target person) can be used to help people with medication adherence problems to take medications as prescribed, with prompting more successful (d = .5) than not prompting (d = .2; Conn et al., 2016).

Finally, supporting those with cognitive impairments due to normal aging and disease (e.g., mild cognitive impairment, dementia) may provide for greater independence and mitigate caregiver burden. If a significant trend of increasing cognitive impairment were detected through long-term individual monitoring, a prompt to seek professional care could be provided. Possibly, short-term interventions to assist people with dementia and their caregivers with everyday tasks could be organized by using Quality of Life Technology interventions such as virtual coaches (e.g., Schulz, 2012).

A recent study showed promise in using mobile monitoring to differentiate MCI and mild dementia from normal aging (Chen et al., 2019).

IS THERE AN APP FOR THAT?

Google and Apple online stores feature hundreds of thousands of applications (apps) aimed at addressing nearly all aspects of health and disease, many with the goals of supporting MMI, including apps to help monitor and manage medication adherence, weight, nutrition, physical fitness, blood pressure, diabetes, sleep, and mood. Some apps track these activities and variables through self-report or sensors within the smartphone itself, while others rely on external sensors, including smartwatches, fitness trackers, telehealth devices, and web-cameras.

Two critical general issues include the safety and efficacy of interventions. (For other ways of evaluating apps on dimensions such as engagement, functionality, aesthetics, information quality, and subjective quality, see Choi et al., 2018.) Do these apps really benefit the user by improving their health and well-being, and if so, are these improvements long-lasting? And are there any potential negative consequences of use (e.g., risk of harm)? Unfortunately, there is not a large, high-quality evidence base to review, especially when it comes to long-term health outcomes. Further, the large and rapidly increasing number of health apps prevents regulatory agencies from thoroughly evaluating these issues for many technology-based interventions.

In the United States, the Food and Drug Administration (FDA) regulates medical devices. Recent guidance released by the FDA clarifies that health apps that fall under the category of medical device may be regulated only in cases in which there exists a potential risk to the user’s safety should the app not work as intended (FDA, 2019). An app that uses gamification to motivate the engagement in physical therapy might fall under the definition of a medical device, but the risk of malfunction is unlikely to result in serious harm to the user. In contrast, an app that makes use of a mobile device’s camera to image a skin lesion, and then uses an AI algorithm to make a classification of whether the lesion is dangerous, would be an example of a health app that the FDA would regulate. Should the algorithm be ineffective, the user can be harmed (e.g., cost of missing cancerous lesion or stress induced by a false alarm). Based on this guidance, many health apps are not FDA regulated, meaning that their efficacy is uncertain, and there is little incentive for app developers to conduct efficacy trials.

Specific to the issue of older adults, health apps (and peripheral devices associated with them) for the most part are not developed and designed considering the needs, preferences, and abilities of older adults. This can be seen in human factors evaluations of existing health-related apps. Morey

and colleagues (2019) reviewed popular apps with the aim of supporting medication adherence and managing heart failure. Expert evaluation uncovered deficiencies that would make these apps challenging for older adults to use, including small and hard-to-see buttons, difficult-to-navigate menus, confusing terminology, and other usability problems. Similar issues were identified in evaluating pain management apps (Bhattarai, Newton-John, and Phillips, 2017). Usability challenges have been noted in user testing as well (e.g., Wildenbos et al., 2019). Many studies have identified difficulties using hardware and software among older adults experiencing normative age-related changes in perception and cognition. Design guidelines do exist for how to reduce these challenges (Czaja et al., 2019). However, such challenges are likely to be greater for older adults experiencing cognitive impairment (e.g., MCI and dementia).

Reminder Efficacy

MMI technology has the potential to greatly benefit the success of interventions at home and in the community by promoting adherence to healthy behaviors. Across a variety of domains, including pharmacological, behavioral, exercise, and nutrition interventions, adherence can be quite poor, resulting in a gap between the potential and actual benefit of a treatment. For example, 50 percent of individuals prescribed a medication for chronic conditions do not take that medication as prescribed (Brown and Bussell, 2011). MMI technology can serve two potential roles: 1) it can monitor whether a behavior (e.g., medication bottle was opened) has occurred, and 2) it can provide reminders to engage in behaviors (e.g., taking a medication at a certain time).

There is a long history of study of methods to improve adherence, for example, to health-related behaviors, and this has resulted in the publication of several systematic reviews. Although these reviews often focus on a broad age range, they are informative with respect to anticipating important issues older adults may face. With respect to medication management, Nieuwlaat et al. (2014) conducted a comprehensive review of general methods to improve adherence, and this was followed by a specific review of all adherence interventions that were mediated by technology (Mistry et al., 2015). Technology-based reminders included various telephone, text messaging, and software-based reminders, and remote monitoring included the use of telehealth devices and electronic drug monitoring. In general (for technology and nontechnology-based adherence interventions), this 2014 Cochrane report arrived at the pessimistic assessment: “Even the most effective interventions did not lead to large improvements in adherence or clinical outcomes” (p. 2). For a variety of reasons, one might expect technology-based adherence interventions to be more successful, but this more

focused review concluded that there was limited evidence for effectiveness, and that adherence-promoting technology “will need to improve if clinically important effects are to be realized” (p. e190). Both Nieuwlaat et al. and the Cochrane report noted the poor quality of many studies that have been conducted to date. Additional, high-powered, well-designed studies (with appropriate control group contrasts) are clearly needed. Further, as discussed later, “one-size-fits-all” interventions should be contrasted with personalized, customizable, and adaptive interventions to explore whether these types of interventions provide additional benefit.

Simons et al. (2016) reached similarly pessimistic conclusions about the efficacy of “brain training” cognitive interventions for impact on everyday functioning, for far transfer measures (e.g., driving safety) compared to near transfer ones (improved performance on the training games). Given concerns with sample inclusion/exclusion rules that tend to exclude comorbid older adults (He et al., 2016), small sample sizes, lack of adequate control groups, and lack of long-term assessment, a cautious conclusion is that the Scottish verdict “not proven” best describes the efficacy of MMI systems.

Beyond efficacy, there is also the issue of cost effectiveness. The largest-scale clinical trial (N = 3230 people with diabetes, COPD, or heart failure) conducted by the National Health Service in the U.K. (Steventon et al., 2012) showed that a telehealth intervention for chronic conditions was not cost effective compared to usual treatment (Henderson et al., 2013), primarily because of equipment costs. Technology costs usually diminish over time (a recent exception being the cost of “flagship” smartphones in the past few years), potentially altering that conclusion as technologies become more affordable.

INTERVENTION STRATEGIES

Traditional intervention strategies often follow a one-size-fits-all approach, with the dose of the intervention identical or similar across individuals and changing infrequently over time. An exercise intervention, for example, might have individuals engage in a walking program in which participants are asked to walk a certain amount of time for a certain number of days each week. Likewise, an individual with hypertension could be prescribed medication at a dose that is adjusted over time based on occasional blood pressure readings. These interventions have the potential, unfortunately, to ignore the varying needs and attributes of the individual and might be insensitive or slow to adapt to the time-varying intervention context.

Just-in-time adaptive interventions (JITAIs) represent an exciting new approach that can be implemented through a combination of mobile and sensor-based technologies (Nahum-Shani et al., 2017). JITAIs are character-

ized by their ability to monitor the state and the context of the individual and, based on this information, provide the appropriate amount and type of intervention at the right time. For example, when sedentary behavior is detected by a worn accelerometer, an app-based JITAI might suggest that the individual engage in physical activity. Further, the system might suggest a specific activity based on the time and weather conditions. Although there appears to be great promise to the approach (Wang and Miller, 2019), additional higher-powered studies are needed to determine the success of JITAIs over other approaches (Hardeman et al., 2019), and to address the unique issues involved in designing successful JITAIs for older adults.

LIMITATIONS FOR USE OF MOBILE AND SENSOR TECHNOLOGY IN HEALTH

Readiness in Aging Populations

When designing a technological intervention, it is important to consider whether the target population is likely to have basic computer experience, or a home broadband connection. In early 2019, only an estimated 53 percent of older adults owned a smartphone (Pew Research Center, 2019), meaning that technologies incorporating the use of a mobile application may not be practical for everyone without significant training for smartphone use. Additionally, older smartphone owners are much less proficient than younger ones (Roque and Boot, 2016). Likewise, although 73 percent of older adults (aged 65+) use the internet, only 59 percent report having a home broadband connection (Pew Research Center, 2019), which is critical for telehealth, mostly done with videoconferencing. Also, only 48 percent of “older-old adults” (aged 75+) use the internet, compared to 78 percent of “younger-old adults” (aged 65–74) (Czaja et al., 2019). Thus, computer and technology literacy are a barrier to adoption, though older adults can significantly benefit from computer literacy interventions, and more specifically, eHealth literacy interventions, resulting in positive changes to health care (Xie, 2011).

The Challenge of Subgroups with Low Tech Adoption

Not all older adults aged 65 and older share the same knowledge about and access to technology products. We have already seen that more specific age groups can be established within the classification of older adults, and these subgroups have different levels of technology usage. In addition to age, education/income and ethnicity are also important factors.

Across age groups, 56 percent of people with an income lower than $30,000 have a home internet connection, compared to 92 percent of those

who make over $75,000. Older adults who have retired may be living with a restricted budget. It is estimated that 9 percent of the older adults in America live below the poverty level (Czaja et al., 2019). Therefore, even those who are willing and cognitively able to adopt new technology and participate in an intervention may not be able to afford to do so. Among older adults, racial minorities are more likely to face the challenges of poverty, as are women and those who live by themselves. Racial minorities and those with lower socioeconomic status are also more likely to rely on a smartphone for internet access, without having a home broadband connection (Pew Research Center, 2019).

Thus, we cannot make broad assumptions about readiness and acceptance of technology. As MMI technology continues to develop, it will be important to consider that older adults may need to adopt an entire infrastructure of technology (e.g., home network, broadband subscription, specific smartphone), and not just that which is necessary for the MMI system itself.

FUTURE DIRECTIONS FOR MOBILE TECHNOLOGY SUPPORTING ADAPTIVE AGING

Several outcome criteria can be envisioned for assessing effectiveness of MMI systems as they mature, drawing on the RE-AIM framework (Glasgow, Vogt, and Boles, 1999) that was developed in the public health intervention field. RE-AIM criteria include reach (the percentage and risk characteristics of persons who receive or are affected by a policy or program), efficacy (positive and negative outcomes for the intervention), adoption (proportion and representativeness of settings, implementation (fidelity of delivery of the program: effectiveness = efficacy × implementation), and maintenance (long-term maintenance of behavior change).

Assuming that researchers can demonstrate MMI efficacy with typical, unrepresentative (He et al., 2016) older adult samples through short-duration, high-internal-validity studies (e.g., phase three clinical trials), what challenges would remain? Pragmatic clinical trials (Ford and Norrie, 2016) are a way to evaluate implementation and adoption. Current home monitoring studies and interventions rely on volunteers, and older volunteers are more likely to have higher levels of education and income, as well as better health and social integration, and less likely to be minority than white (Howell, 2010). Further, in our studies (e.g., Evans et al., 2016) lower SES homes and apartments presented challenging environments for deployment of monitoring equipment. Internet access, a necessity for MMI systems, can be costly and difficult to arrange in rural settings. Broadening participation by underserved populations in pragmatic trials is a worthy goal. Also, once a system either receives FDA approval or earns a best clini-

cal practice designation, ensuring that it is affordable and implementable is an important next step.

Even if a system proves to be efficacious and cost effective, often-overlooked features of cutting-edge technology are maintenance and obsolescence. Maintenance can be problematic in part because companies abandon commercial product lines, or go out of business. A good example was a recent RCT pilot study that showed significant improvement in fitness relative to a wait list control for sedentary middle-aged and older adults. It used a Jawbone UP24 monitor (wearable fitness tracker) in conjunction with an iPad app, and a thigh-worn ActivPAL monitor (Lyons et al., 2017). Jawbone discontinued the fitness tracker, so it became an “orphaned” device. To what extent was that specific hardware and software platform necessary for efficacy?

Further, systems based on mobile devices need to contend with additional challenges. U.S. consumers apparently change smartphones about every two years (Ng, 2019) though that period is lengthening, perhaps in response to smartphone cost increases and slowing improvement in functionality. Our suspicion is that aging adults may change phones less frequently, based on evidence that of those age 65 and older, 53 percent own smartphones and 39 percent own nonsmart cellphones compared to ages 18–29, where 96 percent own smartphones and 4 percent own nonsmart cellphones (Pew Research Center, 2019). This would mean that older adults are likely at greater risk for device obsolescence. Mobile operating system changes by Apple (iOS) and Google (Android OS) can “break” applications, so apps must be maintained and updated. Considering lifespans from onset of chronic conditions, a 10- to 20-year MMI program is conceivable. Focusing on technology functions rather than devices (e.g., Skubic et al., 2014) can address obsolescence.

Finally, maintenance, in the RE-AIM sense, assumes that people will continue to use the MMI system over extended periods of time (years) to support positive changes. Chronic conditions, such as hypertension, require vigilance, and as noted earlier, adherence to taking a prescribed medication is very poor in the general population and for older adults. There is little information available about how best to motivate aging adults to adhere to treatments over long-term intervals, especially when payment to participants is unavailable.

A recent study (Scherbina et al., 2019) of 2,783 iPhone users age 18 and older (M = 48 years, a middle-aged sample) used a smartphone app to try to increase physical activity over a four-week period; the app offered four different intervention types for one week each (crossover design) following a one-week baseline period. All conditions increased step count about 10 percent for those who completed at least one intervention; however, that represented 1,075 people only—a 60 percent attrition rate that

does not bode well for long-term adherence. Only 493 people completed all interventions, representing an attrition rate of 83 percent.

SUGGESTIONS FOR FUTURE MMI STUDIES AND RESEARCH PRIORITIES

We agree with earlier conclusions (e.g., Joe and Demiris, 2013) that too many studies are very short term pilot or feasibility studies. It was difficult to locate robust studies demonstrating MMI efficacy using older adult populations. None followed the full chain of measurement, prediction/inference, and just-in-time intervention, so the following could be priority areas.

Potential Research Priorities for MMI Study Design

• Future studies need to address weaknesses such as small, unrepresentative older adult samples, lack of adequate control groups, and lack of long-term assessment. This may entail funding for a large, multisite study like ACTIVE (Ball et al., 2002).
• Effective MMI systems can be facilitated by partnerships between the research community and industry to enhance usability, scalability, and deployment.
• Given that multimorbidity becomes the norm in old age, MMI studies need to relax exclusion rules to enhance generalizability of results.
• MMI systems should be designed to honor/respect privacy rights.

Potential Research Priorities for MMI Technology Acceptability

Even if an MMI system can show efficacy and cost effectiveness, its value for enhancing well-being in our aging population will be in jeopardy if it is not adopted and used.

• Studies of adoption and use of MMI systems need extended time frames (e.g., decades) to assess longer-term efficacy and cost effectiveness commensurate with lengthened life spans burdened by later life chronic diseases.
• Studies need to incorporate diverse samples including young-old, middle-old, and old-old users; those with disabilities; and disadvantaged groups to gauge comparative effectiveness of MMI versus home-based sensor technology.
• It would be ideal to tap into existing longitudinal studies, such as National Health and Aging Trends Study (NHATS), Health and Retirement Study (HRS), National Health and Nutrition

• Examination Survey (NHANES) to create subsample MMI study opportunities.
• Encourage interdisciplinary MMI teams encompassing engineering, computer science, data science, health, and behavioral science through interagency projects.

REFERENCES

Chen, R., Jankovic, F., Marinsek, N., Foschini, L., Kourtis, L., Signorini, A., Pugh, M., Shen, J., Yaari, R., Maljkovic, V., Sunga, M., Song, H.H., Jung, H.J., Tseng, B., & Trister, A. (2019). Developing measures of cognitive impairment in the real world from consumer-grade multimodal sensor streams. KDD ’19—Proceedings of the 25th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining (pp. 2145–2155). New York, NY: ACM. https://doi.org/10.1145/3292500.3330690.

Choi, Y.K., Demiris, G., Lin, S-Y., Iribarren, S.J., Landis, C.A., Thompson, H.J., McCurry, S.M., Heitkemper, M.M., & Ward, T.M. (2018). Smartphone applications to support sleep self-management: Review and evaluation. Journal of Clinical Sleep Medicine, 14(10), 1783–1790. http://dx.doi.org/10.5664/jcsm.7396.

Conn, V.S., Ruppar, T.M., Enriquez, M., & Cooper, P. (2016). Medication adherence interventions that target subjects with adherence problems: Systematic review and meta-analysis. Research in Social and Administrative Pharmacy, 12(2), 218–246.

Czaja, S.J., Boot, W.R., Charness, N., Rogers, W.A., & Sharit, J. (2017). Improving social support for older adults through technology: Findings from the PRISM randomized control trial. The Gerontologist, 58(3), 467–477. https://doi.org/10.1093/geront/gnw249.

Czaja, S.J., Boot, W.R., Charness, N., & Rogers, W.A. (2019). Designing for older adults: Principles and creative human factors approaches (3rd ed.). Boca Raton: CRC Press.

Davis, F.D. (1989). Perceived usefulness, perceived ease of use, and user acceptance of information technology. MIS Quarterly 13(3), 319–340.

DiJulio, B., Hamel, L., Muñana, C., & Brodie, M. (2018). Loneliness and social isolation in the United States, the United Kingdom, and Japan: An international survey. Henry J Kaiser Family Foundation. Available: http://files.kff.org/attachment/Report-Loneliness-and-Social-Isolation-in-the-United-States-the-United-Kingdom-and-Japan-An-International-Survey.

Evans, J., Papadopoulos, A., Tsien Silvers, C., Charness, N., Boot, W. R., Schlachta-Fairchild, L., & Crump, C. (2016). Remote health monitoring for older adults and those with heart failure: Adherence and system usability. Telemedicine and e-Health, 22(6), 480–488. https://doi.org/10.1089/tmj.2015.0140.

Food and Drug Administration (2019). Policy for device software functions and mobile medical applications. FDA-2011-D-0530. Available: https://www.fda.gov/regulatory-information/search-fda-guidance-documents/policy-device-software-functions-and-mobile-medical-applications.

Ford, I., & Norrie, J. (2016). Pragmatic trials. New England Journal of Medicine, 375(5), 454–463. https://doi.org/10.1056/NEJMra1510059.

Glasgow, R.E., Vogt, T.M., & Boles, S.M. (1999). Evaluating the public health impact of health promotion interventions: The RE-AIM framework. American Journal of Public Health, 89(9), 1322–1327.

Guo, X., Zhang, X., & Sun, Y. (2016). The privacy–personalization paradox in mHealth services acceptance of different age groups. Electronic Commerce Research and Applications, 16, 55–65.

Hardeman, W., Houghton, J., Lane, K., Jones, A., & Naughton, F. (2019). A systematic review of just-in-time adaptive interventions (JITAIs) to promote physical activity. International Journal of Behavioral Nutrition and Physical Activity, 16(1), 31.

Harari, G.M., Lane, N.D., Wang, R., Crosier, B.S., Campbell, A.T., & Gosling, S.D. (2016). Using smartphones to collect behavioral data in psychological science: Opportunities, practical considerations, and challenges. Perspectives on Psychological Science, 11(6), 838–854.

He, Z., Charness, N., Bian, J., & Hogan, W. R. (2016). Assessing the comorbidity gap between clinical studies and elderly patient populations. Conference Proceedings, 2016 IEEE-EMBS International Conference on Biomedical and Health Informatics (BHI) (pp. 136–139). New York, NY: IEEE. https://doi.org/10.1109/BHI.2016.7455853

Hedegaard H., Curtin, S. C., & Warner, M. (2018). Suicide mortality in the United States, 1999–2017. NCHS Data Brief, no 330. Hyattsville, MD: National Center for Health Statistics. 2018. https://www.cdc.gov/nchs/data/databriefs/db330-h.pdf.

Henderson, C., Knapp, M., Fernández, J-L., Beecham, J., Hirani, S. P., Cartwright, M., Rixon, L., Beynon, M., Rogers, A., Bower P, Doll H, Fitzpatrick R, Steventon A, Bardsley M., Hendy J., Newman S.P., & Whole System Demonstrator evaluation team. (2013). Cost effectiveness of telehealth for patients with long term conditions (Whole Systems Demonstrator telehealth questionnaire study): Nested economic evaluation in a pragmatic, cluster randomized controlled trial. British Medical Journal, 346, f1035. https://doi.org/10.1136/bmj.f1035.

Joe, J., & Demiris, G. (2013). Older adults and mobile phones for health: A review. Journal of Biomedical Informatics, 46, 947–954. http://dx.doi.org/10.1016/j.jbi.2013.06.008.

Ko, A., Pick, C.M., Kwon, J.Y., Barlev, M., Krems, J.A., Varnum, M.E.W., Neel, R., Peysha, M., Boonyasiriwat, W., Brandstätter, E., Crispim, A.C., Cruz, J.E., David, D., David, O.A., de Felipe R, P., Fetvadjiev, V.H., Fischer, R., Galdi, S., Galindo, O., Golovina, G., Gomez-Jacinto, L., Graf, S., Grossmann, I., Gul, P., Hamamura, T., Han, S., Hitokoto, H., Hřebíčková, M., Johnson, J.L., Karl, J.A., Malanchuk, O., Murata, A., Na, J., O, J., Rizwan, M., Roth, E., Salgado, S.A.S., Samoylenko, E., Savchenko, T., Sevincer, A.T., Stanciu, A., Suh, E.M., Talhelm, T., Uskul, A.K., Uz, I., Zambrano, D., & Kenrick, D.T. (2019). Family matters: Rethinking the psychology of human social motivation. Perspectives on Psychological Science, 15(1), 173–201. https://doi.org/10.1177/1745691619872986.

Kramer, A.D.I., Guillory, J.E., & Hancock, J.T. (2014). Experimental evidence of massive-scale emotional contagion through social networks. Proceedings of the National Academy of Sciences, 111(24), 8788–8790. https://doi.org/10.1073/pnas.1320040111.

Li, J., Ma, Q., Chan, A.H., & Man, S.S. (2019). Health monitoring through wearable technologies for older adults: Smart wearables acceptance model. Applied Ergonomics, 75, 162–169. https://doi.org/10.1016/j.apergo.2018.10.006.

Lustria, M., Noar, S.M., Cortese, J., Van Stee, S.K., Glueckauf, R., & Lee, J. (2013). A meta-analysis of web-delivered tailored health behavior change interventions. Journal of Health Communication, 18(9), 1039–1069. https://doi.org/10.1080/10810730.2013.768727.

Lv, X., Guo, X., Xu, Y., Yuan, J., & Yu, X. (2012). Explaining the mobile health services acceptance from different age groups: a protection motivation theory perspective. International Journal of Advancements in Computing Technology, 4(3), 1–9.

Lyons E.J., Swartz, M.C., Lewis, Z.H., Martinez, E., & Jennings, K. (2017). Feasibility and acceptability of a wearable technology physical activity intervention with telephone counseling for mid-aged and older adults: A randomized controlled pilot trial. Journal of MIR Mhealth Uhealth, 5(3), e28. https://doi.org/10.2196/mhealth.6967

Madden, M. (2014). Public perceptions of privacy and security in the post-Snowden era. Pew Research Center. Available: https://www.pewresearch.org/wp-content/uploads/sites/9/2014/11/PI_PublicPerceptionsofPrivacy_111214.pdf.

Mistry, N., Keepanasseril, A., Wilczynski, N.L., Nieuwlaat, R., Ravall, M., Haynes, R.B., & Patient Adherence Review Team. (2015). Technology-mediated interventions for enhancing medication adherence. Journal of the American Medical Informatics Association, 22(e1), e177–e193.

Morey, S.A., Stuck, R.E., Chong, A.W., Barg-Walkow, L.H., Mitzner, T.L., & Rogers, W.A. (2019). Mobile health apps: Improving usability for older adult users. Ergonomics in Design: The Quarterly of Human Factors Applications, 27(4), 4–13. https://doi.org/1064804619840731.

Morrow-Howell, N. (2010). Volunteering in later life: Research frontiers. Journal of Gerontology: Social Sciences, 65B(4), 461–469. https://doi.org/10.1093/geronb/gbq024.

Nahum-Shani, I., Smith, S.N., Spring, B.J., Collins, L.M., Witkiewitz, K., Tewari, A., & Murphy, S.A. (2017). Just-in-time adaptive interventions (JITAIs) in mobile health: Key components and design principles for ongoing health behavior support. Annals of Behavioral Medicine, 52(6), 446–462.

Newell, A., & Simon, H.A. (1972). Human problem solving. Englewood Cliffs, NJ: Prentice-Hall.

Ng, C., She, J., & Ran, R. (2019). A Compressive Sensing Approach to Detect the Proximity Between Smartphones and BLE Beacons. IEEE Internet of Things Journal, 6(4), 7162–7174.

Nieuwlaat, R., Wilczynski, N., Navarro, T., Hobson, N., Jeffery, R., Keepanasseril, A., Agoritsas, T., Mistry, N., Iorio, A., Jack, S., Sivaramalingam, B., Iserman, E., Mustafa, R.A., Jedraszewski, D., Cotoi, C., & Haynes, R.B. (2014). Interventions for enhancing medication adherence. Cochrane Database of Systematic Reviews, (11), article number CD000011. doi: 10.1002/14651858.CD000011.pub4.

The OECD Privacy Framework (2013). The Organisation for Economic Co-operation and Development (OECD). Available: http://www.oecd.org/sti/ieconomy/oecd_privacy_framework.pdf, accessed 12/13/2019.

Park, D.C., Hertzog, C., Leventhal, H., Morrell, R.W., Leventhal, E., Birchmore, D., Martin, M., & Bennett, J. (1999). Medication adherence in rheumatoid arthritis patients: Older is wiser. Journal of the American Geriatrics Society, 47(2), 172–183.

Perrin, A. (2018). Americans are changing their relationship with Facebook. Pew Research Center. Available: https://www.pewresearch.org/fact-tank/2018/09/05/americans-are-changing-their-relationship-with-facebook/.

Pew Research Center. (2019). Mobile Fact Sheet, June 2019. Available: https://www.pewresearch.org/internet/fact-sheet/mobile/.

Rantz, M. J., Skubic, M., Abbott, C., Galambos, C., Popescu, M., Keller, J., Stone, E., Back, J., Miller, S.J., & Petroski, G.F. (2015). Automated in-home fall risk assessment and detection sensor system for elders. Gerontologist, 55(S1), S78–S87. doi:10.1093/geront/gnv044.

Reeder, B., & David, A. (2016). Health at hand: A systematic review of smart watch uses for health and wellness. Journal of Biomedical Informatics, 63, 269–276. http://dx.doi.org/10.1016/j.jbi.2016.09.001.

Roque, N.A., & Boot, W.R. (2016). A new tool for assessing mobile device proficiency in older adults: The Mobile Device Proficiency Questionnaire. Journal of Applied Gerontology, 37(2), 131–156. https://doi.org/10.1177/0733464816642582.

Sanchez, W., Martinez, A., Campos, W., Estrada, H., & Pelechano, V. (2015). Inferring loneliness levels in older adults from smartphones. Journal of Ambient Intelligence and Smart Environments 7(1), 85–98. https://doi.org/10.3233/AIS-140297.

Sandstrom, G., Lathia, N., Mascolo, C., & Rentfrow, P. (2017). Putting mood in context: Using smartphones to examine how people feel in different locations. Journal of Research in Personality, 69, 96–101.

Scherbina, A., Hershman, S.G., Lazzeroni, L., King, A.C., O’Sullivan, J.W., Hekler, E., Moayedi, Y., Pavlovic, A., Waggott, D., Sharma, A., Yeung, A., Christle, J.W., Wheeler, M.T., McConnell, M.V., Harrington, R.A., & Ashley, E.A. (2019). The effect of digital physical activity interventions on daily step count: A randomised controlled crossover substudy of the MyHeart Counts Cardiovascular Health Study. The Lancet Digital Health, 1(7), e344–e352. https://doi.org/10.1016/S2589-7500(19)30129-3

Schmidt, L.I., & Wahl, H.W. (2018). Predictors of performance in everyday technology tasks in older adults with and without mild cognitive impairment. The Gerontologist, 59(1), 90–100.

Schulz, R. (Ed.) (2012). Quality of life technology handbook. New York: Taylor & Francis/CRC Press.

Seifert, A., Christen, M., & Martin, M. (2018). Willingness of older adults to share mobile health data with researchers. GeroPsych: The Journal of Gerontopsychology and Geriatric Psychiatry, 31(1), 41–49. https://doi.org/10.1024/1662-9647/a000181.

Simons, D.J., Boot, W.R., Charness, N., Gathercole, S.E., Chabris, C.F., Hambrick, D.Z., & Stine-Morrow, E.A.L. (2016). Do “Brain Training” programs work? Psychological Science in the Public Interest, 17,108–191. https://doi.org/10.1177/1529100616661983.

Skubic M., Jimison, H., & Keller J., Pepescu, M., Rantz, M., Kaye, J., & Pavel, M. (2014). A framework for harmonizing sensor data to support embedded health assessment. 36th Annual International Conference of the IEEE Engineering in Medicine and Biology Society, Chicago, IL, 2014, pp. 1747–1751, https://doi.org/10.1109/EMBC.2014.6943946.

Steventon, A., Bardsley, M., Billings, J., Dixon, J., Doll, H., Hirani, S, Cartwright, M., Rixon, L., Knapp, M., Henderson, C., Rogers, A., Fitzpatrick, R., Hendy, J., & Newman, S. (2012). Effect of telehealth on use of secondary care and mortality: Findings from the Whole System Demonstrator cluster randomised trial. British Medical Journal, 344, e3874. https://doi.org/10.1136/bmj.e3874.

Suhara, Y., Xu, Y., & Pentland, A. (2017). DeepMood: Forecasting Depressed Mood Based on Self-Reported Histories via Recurrent Neural Networks. Proceedings of the 26th International Conference on World Wide Web.

Tsai, H.S., Jiang, M., Alhabash, S., LaRose, R., Rifon, N.J., & Cotten, S.R. (2016). Understanding online safety behaviors: A protection motivation theory perspective. Computers & Security, 59, 138–150. https://doi.org/10.1016/j.cose.2016.02.009.

Venkatesh, V., Thong, J.Y.L., & Xu, X. (2012). Consumer acceptance and use of information technology: Extending the Unified Theory of Acceptance and Use of Technology. MIS Quarterly, 36, 157–178.

Wang, L., & Miller, L.C. (2019). Just-in-the-Moment Adaptive Interventions (JITAI): A meta-analytical review. Health Communication, 1–14. https://doi.org/10.1080/10410236.2019.1652388.

Wildenbos, G.A., Jaspers, M.W., Schijven, M.P., & Dusseljee-Peute, L.W. (2019). Mobile health for older adult patients: Using an aging barriers framework to classify usability problems. International Journal of Medical Informatics, 124, 68–77.

Xie, B. (2011). Effects of an eHealth literacy intervention for older adults. Journal of Medical Internet Research, 13(4), e90.

Zhao, Y., Ni, Q., & Zhou, R. (2018). What factors influence the mobile health service adoption? A meta-analysis and the moderating role of age. International Journal of Information Management, 43, 342–350.

__________________