A Step Toward Better Care: Understanding What Caregivers and Residents in Assisted Living Facilities Value in Health Monitoring Systems

The past several decades have seen significant advances in monitoring older adults' health and well-being. However, creating viable, practical monitoring systems for informing caregivers requires understanding which behaviors and signs to track and what approaches best present that information. To investigate how technology can be leveraged to better augment caregivers' workflows, we take a multi-stage, qualitative approach to gain insights into the needs of caregivers and the older adults receiving care. Specifically, we use a series of domain expert interviews, cognitive walkthroughs, and semi-structured interviews with residents, and we synthesize our takeaways using thematic analysis at each phase. Our results show that this type of monitoring technology has great potential to reduce the effort needed by caregivers to complete their responsibilities and communicate with their teams. Additionally, we found that older adults are receptive to the technology but their privacy and autonomy must be prioritized for the sake of their mental wellbeing. These insights will facilitate greater intelligent interface development for Person-Centered Care by identifying important design considerations and vital features that require system support.


INTRODUCTION
The number of older adults is expected to increase significantly over the coming decades: the United Nations expects that there will be nearly 2.1 billion adults over the age of 60 by 2050 [92].This trend is significant, as older adults generally have more complex and varied care needs than working-age adults [39,48].In the United States alone, approximately 50% of older adults have at least one chronic condition (including 11 million older adults with five or more chronic conditions) [62].Moreover, for older adults in long-term care facilities, over 75% have had two or more of the most common chronic conditions, and 40% require assistance with three or more activities of daily living (ADLs) [11].As it stands, caregivers already have difficulty meeting these care needs [70,83,89], with nearly 70% of older adults being dependent on some form of caregiving [35].This paradigm will only become more challenging for caregivers as the older adult population increases and the "caregiver support ratio"-the number of potential caregivers aged 45-64 years for each person aged 80 years or older-drops from 7-to-1 in 2010 to 4-to-1 in 2030 and 3-to-1 in 2050 [35,79].Furthermore, for caregivers of older adults with chronic conditions, studies have found that the associated stress from their responsibilities leads to worsening physical health, mental health, and quality of life [69].Studies have also found that this population increase of older adults is expected to significantly increase overall healthcare costs [86].If these factors together are left unchecked, the burden of care-i.e., the physical, emotional, social, and financial problems associated with providing care-will eventually become societally untenable.
Advances in both machine learning techniques and wearable and environmental sensing have made advanced healthcare interfaces designed to alleviate these challenges an increasingly viable path forward.Specifically, caregivers who leverage healthcare interfaces that better optimize the logistics of caring for older adults-such as automatically documenting resident status and sending timely insights into older adults' health and well-being-could feasibly provide more effective Person-Centered Care (PCC) [60] and facilitate Aging in Place [98].Realizing this possibility, multiple studies have investigated the design of streamlined healthcare interfaces for caregivers, focusing on the automatic tracking of data streams such as: vitals [28,50,75,90], gait and balance [34], location [94], and various ADLs [14,[18][19][20]67].More specialized studies for elderly living have also explored the utility of automatically recognizing both vitals and activities [77].The culmination of the research outcomes of these previous works encourage approaches that reduce the burden of care of caregivers and the societal costs in healthcare [38,40,56,76,78,97].
While this significant body of work has investigated the design of tools to automate aspects of the caregiving process, few works have explored how these approaches could be used to both reduce the burden of care on caregivers as well as fundamentally improve the care provided [17,70].Studies that do focus on the needs of caregivers largely focus on the needs of informal caregivers, who have varied needs and experiences in comparison to formal caregivers, who have formal training, generally more structured roles and communication needs, and look after more older adults each with their own unique needs.Specifically, existing studies have primarily focused on codifying the needs of informal caregivers [17,82,83], understanding how informal caregivers interact with existing technological approaches [46,70], or developing and evaluating systems for specialized use cases [29,64,84].Conversely, there has been a lack of investigation into similar questions for formal caregivers, namely: what insights would formal caregivers like to have into the health and status of older adults (i.e., residents of formal caregiving communities), how would they like these insights delivered to them, and how would they ultimately use these insights to improve their caregiving workflow.
To that end, we aim to achieve the following goals in this work: ( (2) identify potential points of improvement in the current caregiving paradigm, (3) explore how a health monitoring system could supplement the current caregiving workflow, (4) understand resident views of a technology-based system to improve caregiving To answer these questions, we propose an exploratory multi-stage qualitative research methodology grounded in the design of and interaction with a prototype health monitoring system.Specifically, we glean insights from the thematic analysis of semi-structured interviews with both caregiving staff and older adults, as prior work demonstrating that their input in the developmental stages is crucial to the practical utility of these systems [24,25,33,71].We conduct two rounds of interviews with each population group, with the second round of interviews being conducted after interaction with a prototype health monitoring system.Caregiving staff were asked to interact with AutoRounds, a prototype health monitoring interface designed based on insights gleaned from thematic analysis of the first round of interviews.Older adults were asked to wear a smartwatch on each wrist, simulating participation in a health monitoring system that tracks their activity and daily health.
From our work, we claim the following contributions of our paper: (1) Multi-stage, Qualitative Approach: We outline a multi-stage, qualitative approach that involves domain experts at each stage (i.e., domain expert interviews, health monitoring system design, and cognitive walkthrough evaluation).( 2) Key Features: We identify key features of a health monitoring system designed for caregivers that could leverage advanced health monitoring.(3) Health Visualization Design: We present a technology probe interface based on identified features that serves as an example for how future systems should communicate advanced health monitoring information.(4) Domain Insights: We present the results of semi-structured interviews and cognitive walkthroughs conducted with domain experts.(5) Resident Insights: We present insights from the qualitative analysis of semi-structured interviews about residents' attitudes toward participating in a remote health monitoring system.(6) Lessons and Considerations: We share lessons learned and future considerations for interfaces geared towards improving the state of elderly care in long-term care facilities.

Caregiving for Older Adults
The type and extent of care that an individual needs generally depends on the level of assistance that they require for performing everyday tasks and activities.Such activities are commonly referred to as ADLs and can be divided into two categories: basic and instrumental [1,59].Basic ADLs are activities oriented toward taking care of one's own body and are completed on a routine basis.Examples of basic ADLs include personal hygiene and grooming, bathing and showering, toilet hygiene, dressing, eating and feeding, functional mobility, personal device care, and sexual activity [8].
Instrumental ADLs are activities that are characterized by more complex daily interactions; some examples include health and home management, driving and community mobility, child rearing, meal preparation and cleanup, and shopping [1].Occupational therapy work by Amini et al. [1] provides the following details of how different levels of older adults' ADL performance ability can affect their choice of living communities in the United States.Older adults who can perform all but one or two instrumental ADLs can live in independent living communities.These communities have caregivers available to provide assistance, but they generally allow residents to maintain the bulk of their independence.As individuals become unable to adequately perform more ADLs on their own, they may choose to live in assisted living communities or skilled nursing facilities.These communities provide older adults with more regular nursing care to aid with these difficulties in performing ADLs but this often comes at the cost of reduced freedom and autonomy.Continuing Care Retirement Communities (CCRCs) are a combination of all three of these living options, allowing residents to transition between the living options as their ability to perform ADLs and their corresponding need for assistance changes while allowing them to maintain a sense of community.
Within these facilities, the most common way in which healthcare professionals interact with residents' health data is through either paper documentation or Electronic Health Records (EHR) [30].Increasingly over the past several decades, EHR has supplanted paper documentation [30].However, studies have found that EHR systems are difficult to navigate [22], time-consuming [22,72], lower interaction both among caregivers and between caregivers and patients [54], and fundamentally do not substantially improve and may even lead to a decrease in the quality of care [41,63].
In this work, we collaborate with the personal care housing section of a CCRC located in the Northeastern USA, which uses paper documentation-but not an EHR system-for interacting with residents' data; and a memory care facility located in the Southwestern USA, which does use an EHR system for interacting with residents' data.Personal care homes are very similar to assisted living facilities with the distinction that assisted living facilities offer more skilled healthcare services to residents.Memory care facilities are a specialized form of assisted living facilities that are designed to accommodate residents who are largely independent but have specific and serious memory-related needs.

Recognizing Elderly Activities, Vitals, and Locations
In addition to directly supporting caregivers, researchers have also explored tracking the activities, vitals, and locations of older adults to facilitate better care [20,27,47,49,52,55,67,77,87,94,99,100].For activities, much of the work has been done in the field of activity recognition, which utilizes either wearable inertial sensors or environmental sensors and machine learning techniques to recognize human activities.While much of the literature in this space has focused on recognizing the activities of the working-age population, several studies have looked at recognizing the activities of older adults [20,49,52,67].A few studies have taken this a step further and tested these systems in long-term care facilities, e.g., the Center to Stream HealthCare In Place at the University of Missouri has done a significant amount of work in this space [77,87,99].For vitals, researchers have sought to automatically monitor a range of vitals including body temperature, heart rate, respiration rate, blood pressure, pulse oxygenation, and blood glucose commonly using either wearable or implanted sensors [47].Examples of such sensors applied to elderly care include works by Xie et al. [100] and Diraco et al. [27] who both looked at heart and respiratory rates.For location, Global Positioning Systems (GPS) and indoor positioning systems (IPS) have been well-studied as a means of identifying the location of humans and objects outdoors and indoors respectively [97].With respect to elderly care, a big focus has been on addressing the problem of wandering, especially among individuals who have dementia [55,94].Wang et al. [97] provides a substantial review of advances in the monitoring of each of these three types of information.
These advances in sensing are increasingly being leveraged by commercially available wearable devices such as smartwatches and fitness trackers.While early devices were largely limited to recognizing steps, ambulation, and basic exercises, more recent iterations can recognize a wider range of activities (e.g., washing hands, sleeping, an array of workout types [3,5,68]), vitals signs (e.g., heartbeat and blood pressure [65]), and adverse events (e.g., fall detection [4]).Combined with consistent advances in battery life and sleeker form factors, these devices will likely continue to become more and more ubiquitous as tools for both personal and caregiver-provided healthcare management.
In this work, we aim to take a step toward understanding the practical viability of these systems as tools for elderly care.On a broad level, many works have sought to understand how older adults perceive technology designed to help with the aging process.These have demonstrated that significant barriers to acceptance among older adults need to be overcome before these technologies can be accepted.Specifically, older adults have raised concerns over usability, cost, the need for specialized training, security, safety, increased amount of work, and privacy [45,61,73,74,80,103].In addition, studies have found that people prefer human healthcare caregivers over the use of more device-based care, noting that people had concerns over the lack of human contact such devices might engender [45,61,88,95].Recent work by Caldeira et al. [13] found that there was a significant stigma associated with technology designed to facilitate aging in place, as older adults often associated such technology with frailty and were unwilling to admit that they could possibly need any assistance.Our work adds to this body of literature, as few studies have explored the acceptability of monitoring systems in particular and grounded their findings in interactions with actual systems.

Caregiver Support Systems
There has been a number of studies that have sought to understand the challenges of providing care within the HCI and CSCW communities.Although the specific populations they have studied have varied, these works have all emphasized the significant and unique challenges caregivers face.Works have highlighted the increased mental strain on caregivers [17,83], the isolation that many feel [58,83], and the difficulty many face in coordinating care requirements [17,21,82,83].In particular, a number of researchers have sought to understand and improve the state of communication between caregivers, between caregivers and patients, or between caregivers and family members of the patient [2,31,81,85,101].
Notably, much of this work centers on the needs of informal caregivers, seeking to both understand their unique care practices and needs [17,57,58,66,82,83] and identify corresponding best practices for design for this population [9,16,17,23,43,85,101].Unlike formal caregivers, informal caregivers are often required to learn on their own how to provide care and juggle those responsibilities with their existing personal and professional lives [17,82,83].Furthermore, they are often only providing care to a single individual and may not have access to resources that a formal caregiver in a long-term care facility will have.Thus, solutions designed for informal caregivers to improve the quality and experience of providing care are unlikely to perfectly address the needs of formal caregivers.
Works that have looked at formal caregivers are less prevalent in the literature; however, studies have looked at the needs of this population in specific contexts.A common theme among these works is the need for technologies to support greater collaboration and communication among caregivers and between caregivers and the older adults they are providing care for [2,7,12,44].Other works have proposed specific tools to improve upon formal caregivers' ability to provide care such as a medication management application [85] and a nursing activity recognition system [42].One area of formal caregiving that has received significant attention from the HCI community is dementia care [29,31,32,37,64,91].
Our work focuses on formal caregivers in assisted living, a population that is not as well studied in the CSCW and HCI literature.While there are overlaps between this particular subset of caregivers and other types of caregivers with respect to the types of care they provide and the means by which they provide them, there are enough differences that their specific needs bear identifying and designing for.Compared to informal caregivers, formal caregivers generally provide care for a larger number of individuals, who can have a wide variety of physical and cognitive capabilities, personalities, and unique needs.A similar comparison can be made between caregivers in assisted living and caregivers in skilled nursing and dementia care.In these latter facilities, individuals need a significant amount of attention and assistance, regardless of their specific physical and cognitive capabilities.As such, health monitoring systems designed to increase resident independence are not as relevant in skilled nursing and dementia care.
All of this is to say that there is both a need for and a lack of studies investigating systems that leverage the recognition and monitoring of key health indicators of older adults to support formal caregivers in providing care.To that end, in this work we seek to understand what indicators are critical to their workflow, explore how those indicators could be communicated to them, and how such systems would be perceived by both caregivers and older adults.

Study Design
To get a baseline understanding of how monitoring systems can support formal caregivers, we conducted a set of semi-structured interviews with nurses and nursing assistants who worked in the personal care section of a CCRC in the Northeastern USA.We interviewed seven individuals via convenience sampling (code-named P11-P17), whose demographics can be found in Table 1.
Interviews lasted approximately twenty minutes and were audio-recorded for later transcription.Interview scripts are available as supplemental material.
In conducting our interviews, we focused on addressing the following goals: (G1) Paradigm Understanding: The first goal is to understand the current health management paradigm for caregivers in a senior living community.(G2) Technology Comfort: The second goal is to gauge caregiver comfort with using technology for health management.(G3) Utility Potential: The third goal is to assess the potential utility of an application that is designed to assist caregiving staff in monitoring residents' health statuses and well-being.

Qualitative Data Analysis
We used thematic analysis [10] to analyze the interview data.All interview transcripts were uploaded to Dedoose [26]-an application designed to facilitate mixed-methods research-for our qualitative analysis.Our qualitative data analysis team (who are co-authors) was composed of four experts: an occupational therapist, a mixed-methods research analyst, and two computer science researchers with formal qualitative analysis training.Each person first separately coded two of the interviews to develop an initial set of codes.Then, the team discussed the codes that they had discovered and combined their findings into a single codebook.This codebook was then used by two of the four coders to code the remaining interviews, iteratively making updates to the codebook as needed.Afterwards, the resulting codes and interview excerpts were reviewed to extract themes, which were later reviewed and approved by all four coders.A total of five themes emerged from our thematic analysis, summarized in Table 2, are discussed in detail in the following subsections.

Integrating technologies to improve current workflow (G2, G3
). Participants were receptive to the idea of using technology to address the challenges they experience while fulfilling their responsibilities.In particular, they expressed desires for systems that would automate repetitive tasks (e.g., sorting medication), reduce search time (e.g., having to knock on every door), and help them determine how to divide their time (e.g., know how many people need assistance and to what degree they need aid).Moreover, although participants noted that they did not use smartphones and tablets as part of their current caregiving workflow, all participants expressed that they would be comfortable using them for caregiving if required.However, P12 noted: "I could see a problem with [nursing staff] not understanding how to use it.That's where I could see a problem [...] Some who are older nurses who are used to everything being written not using computers as much and things like that." Participants did note that there are a few technologies currently used by the facility as part of the care provided.These consist primarily of tools and systems designed to keep residents safe and are not directly interacted with by caregiving staff.Notably, residents do not like or fully understand these technologies, and their resistance towards or improper use of these technologies limits the technologies' effectiveness.When referring to medical alert pendants (necklace devices for alerting staff of medical emergencies), P15 mentioned: "...a lot of them don't know how to [push the button].The ones here, they don't wear it [...] probably like one or two [residents do] [...] They don't know what it is." 3.2.2Wide range of behaviors that caregiving staff need to monitor (G1).Over the course of their shifts, caregiving staff are responsible for ensuring residents are safe and well, while still allowing them to have as much independence as possible.In particular, nursing assistants brought up two of the main aspects of the care that they provide.The first is either calling or visiting residents to ensure that they get out of bed in the morning, and the second is ensuring that residents remember or are prepared to leave their rooms on time to go to the cafeteria for meals.More broadly, participants noted that a key component of their responsibilities was monitoring residents over the course of their shift.P14 described this aspect of her responsibilities: "I'm a nursing assistant.I have a lot of experience.I feel like we are like the third eye.For a lot of these residents, we see what others don't [...] we are there to observe.[...] Well it's not only about get in there and do it, helping them get dressed like making sure they go to the meals and appointment.[...] we gotta make sure what's going on and see if they are showering, or their hair is greasy, or they're washing their hair maybe we need to step in and give them [a] shower or like you know [...] helping them 'cause we here in assisted living.[...] We gotta keep an eye on them and be you know...observant." Participants expressed a desire for the tracking of a variety of activities, vital signs, incidents, and moods, as well as general location.The most common activities interviewees brought up were toileting and bathroom habits, dressing, eating, and ambulation.Although eating was frequently mentioned, it often came up in the context of knowing that residents had left their room and successfully made it to the cafeteria on time for meals.Other activities mentioned were showering, brushing teeth, fluid intake, hair care, medication intake, getting to activities and appointments, and a general knowledge of residents' routines.For example, P17 explained: "Their bathroom habits, their bowel movements is a big thing...Are they able to dress themselves?Are they able to clean themselves?Do they need cueing to come to meals?Do they need assisting walking or like coming down to different activities or appointments or do they need guiding throughout the day?[...] It also might be helpful to have the PRN [pro re nata, as needed] drugs that residents ask for during the day.Observations and interactions with residents are shared with coworkers, as P16 commented: "'Cause then that's something that could be you know passed on to the next shift or whatever.I mean that actually will slip you know people's mind too because it's just something that's PRN [pro re nata, as needed].It's not like it's an everyday thing." The most frequent incident that participants wanted to see detected was falls, but they also brought up coming down the stairs when residents were not supposed to, abnormalities in vital signs, and breaks in routines as worth tracking.P12 noted: "I don't know if it would be possible, but like unusual behaviors if you're used to a resident who's normally up and about.Say for instance I come in at 6:30; there's certain residents that you know 6:30 [they are] drinking coffee.What about one day you come in and one of those residents is not out there drinking coffee?It should alert you that something might be wrong you know." In that sense, participants frequently brought up knowing the general location of residents.This was most commonly brought up in relation to knowing whether residents had made it to meals or had wandered off to a location they were not supposed to be.In terms of mood tracking, participants mentioned being able to track when residents were agitated or depressed would be helpful.

Usage of mental notes and verbal communication (G1).
Participants noted that they had no formal mechanism for documenting care provided or changes in resident abilities or needs on a short-term basis.P15 noted: "So we don't really document it, we just know.We just, if we see somebody declining, we'll let the nurse know but we know that we have to help them more." Instead, caregiving staff rely on a combination of observational skills, working memory, and verbal communication.Specifically, this paradigm depends on caregiving staff noticing changes in a resident's behavior or abilities, making a mental note of those changes, and efficiently communicating that information to the nurse to make any changes to the resident's care.P12 explained: "Well, we normally rely on the nursing assistants to give us that information and feedback because they're the ones that actually do the immediate morning hands-on care 'cause the nurses are pouring meds at that time and things like that.So, we rely on the information to come from our nursing assistants [via word of mouth]." 3.2.4Presenting data over time through simple UI (G2, G3).Given this large amount of information about residents that caregiving staff expressed needing to monitor, it is vital that any software application that provides this information also displays it in a way that is useful and easy to understand.Participants noted the need for immediate information when specific incidents occur as well as long-term information (i.e., the ability to view trends in activity performance or vital Proc.ACM Hum.-Comput.Interact., Vol. 8, No. CSCW1, Article 13.Publication date: April 2024.signs).Regarding the former, participants commonly mentioned they would like to receive alerts or notifications when incidents occurred.P17 noted: "One thing it would be great to have [...] [is] something letting you know when someone is getting in trouble [...] OK, [for example] somebody who really shouldn't be toileting themselves alone.If we, I mean that's how most falls happen, like send it to text when they're getting up without bringing [one of us] along.Like letting us know that, that person is getting up." Regarding the latter, P16 noted: "Yeah so then we can see those changes like say we tracked it today and then a month later we could see what their progress was or what their decline was.You know?That would help like every shift out too. " In terms of how this information would be presented, participants stated that the application should be simple, user-friendly, and minimize the amount of language (i.e., to be accessible to caregiving staff who are not native English speakers).However, there was not a consensus on more specific design elements: participants wanted information presented in a variety of ways from numbers to graphs to symbols and color-coding.3.2.5 Resident resistance is the main challenge to providing care (G1).Understandably, assisted living facilities engender unique social challenges due to the nature of the relationship between residents and caregiving staff.Although the staff wants to preserve the dignity of residents while ensuring that they are clean, safe, and healthy, residents can be resistant to this assistance and care, especially given that they are used to a lifetime of being independent.P13 summarized the following: "And residents who come from-some residents who come from independent living, depending on their lifestyle when they were there, they'll come and we want them to be active we want them to be you know doing things that are scheduled to be done and like my colleague said it's a lot of resistance sometimes.It doesn't happen a lot here but yeah, we have that experience as well and that's one of our greatest challenges."

DESIGNING A PROTOTYPE MONITORING SYSTEM
To investigate how well these themes generalize to other facilities we decided to develop a prototype monitoring system based on these themes and see how well it met the needs of caregivers at a separate facility.To achieve this, we extracted insights, designed solutions, and determined the necessary interaction techniques with the data from the themes derived from our first round of interviews.We describe these insights and solutions in more detail in the following subsection and describe the features of the prototype monitoring system in the remaining parts of this section.

Design Guidelines Based on Interview Findings
Through these interviews, we found that for caregivers to adequately provide care they need to be cognizant of a resident's safety, location, ADL performance, health status, and mood, confirming the utility of a health monitoring system.As participants unanimously stated that they were comfortable with smartphones and tablets, many of them noting that they had their own personal smart devices, we decided to design with these devices in mind (S1).In this work, we chose to develop a tablet application to maximize screen size.To alleviate the difficulties associated with the current paradigm of keeping track of the wide range of behaviors, we decided that one of the focal points of the application would be an interactive screen that shows these various behaviors for the residents caregivers are providing care for (S2, S3).We also decided to include a map of the facility that displayed the locations of residents (S3).As caregivers highlighted the importance of communication and their current reliance on both direct observation and verbal communication, we decided to include a notetaking system (S4).Finally given that caregivers noted they wished to view trends over time and wanted to be alerted when adverse events occurred, we designed the system to show ADL performance and vitals over multiple time periods (S5) and provide alerts that triggered immediately upon the occurrence of an adverse event (S6).When designing the application, we placed an emphasis on minimizing text in the UI and allowing users to take multiple paths to accomplish tasks within the application due to specific concerns regarding second language learners and people with varying tech competencies.A summary of the themes, insights, and solutions can be seen in Table 2. Notably, our application did not address the theme of resident resistance, we explore the needs of residents further in a separate study in Section 6 (S7).

AutoRounds Overview
Based on the insights from the exploratory semi-structured interviews, we developed AutoRounds as a React web application (see https://auto-rounds.vercel.app/).We populated the application with nine imaginary residents (with corresponding fictitious health data) for the purpose of the study.These residents were given a number of traits that are representative of residents within an assisted living facility (e.g., mobility needs, health conditions like high blood pressure, prescription medications, and interpersonal relationships).The health data includes logs for vitals and ADLs over the course of three months and were generated to be stable in healthy ranges with exceptions being situations such as one of the imaginary residents having high blood pressure.The application has four main components: a List of Residents view, a specific Resident view, a Map view, and a Notes view.We discuss each of these in detail in the following subsections.

List of Residents View
The List of Residents View (shown in Figure 1) shows all the residents in the facility, providing basic information such as age, room, current location, mobility assistance requirements, and wearable battery life.Additionally, it allows users to search for a particular resident or filter by location or building within the facility.This view is related to Solution (S1).

Resident View
Clicking on a resident from the List of Residents view takes users to the Resident view for the selected resident.The user is first presented with a Summary page (Figures 2a & 2b) which shows the most recent measurements for both vitals and ADLs as well as common locations the resident typically spends time in within the facility.The vitals section allows users to view blood pressure, heart rate, blood glucose, and weight while the ADL section allows users to view dressing, eating, fluid intake, medication intake, physical activity, showering/bathing, and toileting.Each of these items has an option to manually add in new data by clicking on a plus button located to the right of the latest/aggregate value.The medication intake portion of this screen displays a list of the resident's prescriptions and allows caregivers to check off when a dose was taken/administered.The system allows users to manually record administration of both prescription and over-the-counter (OTC) medications.This view also provides a simple graphic that can be viewed at a glance to see when ADLs were performed over the course of the day.
In addition to the Summary page, an individual resident's page has Vitals and ADLs pages (Figures 2c & 2d) which allow users to view the performance history of both vitals and ADLs respectively over varying time periods in a form of graphs or calendars.These views are related to Solutions (S2), (S3), and (S5).

Notes View
We allowed users to enter a new note from anywhere in the application by placing the "Add Note" button in the header of the application, envisaging caregiving staff wanting to quickly add notes for residents while performing their duties.Users are able to view notes in two different locations: they can view resident-specific notes within the Notes page of an individual resident's page or they can view the notes for all the residents on the separate Notes page (shown in Figure 3).Notes are able to be edited if a mistake is made or context changes, resolved if the task is completed, and deleted if the note is no longer relevant.This view is related to Solution (S4).

Map View
As shown in Figure 4a, the Map View provides users with a real-time look at where residents are within the facility.The sidebar shows this information textually, also allowing users to search for specific residents or locations.This view is related to Solution (S3).When a fall occurs, the resident's location is highlighted on the map (Figure 4b & 4c).This feature is related to Solution (S6).

Study Design
To understand if our themes would generalize to other facilities, we conducted a user study with caregivers at a separate senior living community in which they directly interacted with the interface  and subsequently participated in a semi-structured interview to understand their thoughts on how such a system could be useful as part of their workflow.We formalized the caregivers' interaction with the interface by conducting a Cognitive Walkthrough with Users (CWU) [36,51,53] with a target group at a memory care facility.To make up for the limitations of cognitive walkthroughs [53], we combined the CWU with a Concurrent Think-Aloud.While Retrospective Think-Alouds have been shown to reveal more usability issues [93], our studies were conducted during the workday within the environment in which the system would be used.As a result, our study benefited from the increased cost-effectiveness of Concurrent Think-Alouds [51].Additionally, this design allowed the participants to interact with the interface in as close to a live deployment as possible during this phase of the design without interfering with the care of the actual residents.
At the beginning of the cognitive walkthrough, participants were given a set of five simple tasks focused on UI interactions and navigation to help them get accustomed to the interface.After completing these tutorial tasks, participants were given a set of tasks that were centered around a common workday in an assisted living facility and involved one or more of the fictional residents.To complete the tasks, participants were allowed to use the interface and/or describe physical actions as if they were actually within the facility.As such, tasks generally had multiple viable paths to completion, not just within the application itself but also via a combination of physical and digital actions.After completing the cognitive walkthrough, participants were asked to fill out a System Usability Scale (SUS) survey [6], evaluate the ecological validity of the cognitive walkthrough tasks [102], and answer open-ended questions via a semi-structured interview.All of the task definitions and interviews questions used in the cognitive walkthroughs are available as supplemental material.
Prior to conducting these walkthroughs, the tasks and interface were reviewed by three healthcare domain experts and one HCI expert-all of whom were not involved in their development-to confirm that the instructions were clear and that the tasks were ecologically valid.Two of the healthcare domain experts are nurses who actively work in or work with memory care and assisted living facilities.The other healthcare domain expert is an occupational therapist who works closely with skilled nursing, long-term care, and assisted living facilities.One of the nurses and the occupational therapist are authors on this paper.This process led us to make our tasks more specific and make minor changes to our interface such as adding a navigation feature to the map screen to search for residents or locations and adding text underneath the icons on the sidebar for readability.
We conducted the cognitive walkthrough study with staff at a memory-care facility located in the Southwestern USA.We recruited eight participants (code-named P21 through P28), all of whom were involved in the care of residents living in the facility; their demographics are shown in Table 3.These participants were recruited via convenience sampling and were compensated with a giftcard worth 25 USD for their time.These participants had several different roles at this facility, enabling us to assess the utility of the interface from a variety of perspectives and identify any gaps in feature support not covered in our initial semi-structured interviews.

System Usability
We used the SUS scores as a preliminary assessment of the validity of the cognitive walkthrough.If the interface was deemed unusable by SUS standards, then revisions would be necessary before the cognitive walkthrough evaluation would be valid.The SUS scores had an average of 91.9 (SD=10.6)with four of these participants giving the interface a perfect score, three giving it an excellent score, and one giving it a good score using categories as defined by Bangor et al. [6].Because the average score is strongly positive, the design of the interface was found to be satisfactory for the cognitive walkthrough.

Ecological Validity
We asked the participants to evaluate the ecological validity of the tasks they were asked to complete following the practice used by Zhang et al. [102].That is, they were asked to rate how realistic each task was on a four-point scale: not at all, somewhat, realistic, and very realistic.If the tasks were deemed largely unrealistic by the participants, then revisions to the cognitive walkthrough would be necessary to ensure that the health monitoring system is being evaluated in the context of actual caregiver tasks.Four of the participants rated every task as very realistic.Three rated most of the tasks as very realistic and one or two tasks as realistic or somewhat realistic.One participant rated five of the tasks as realistic, three as somewhat realistic, and rated the first practice task to look at a resident's ADL history in the interface as not realistic.In the post-interview, all of the participants agreed that their responsibilities were represented by tasks they were asked to complete during the study.

Qualitative Data Analysis
We conducted qualitative analysis on the think-aloud comments and interview data in a similar fashion to the analysis conducted during the initial design phase of this work (described in Section 3.2).The four emergent themes are listed in Table 4 and discussed in detail in the following subsections.From our interviews, we found that the most fundamental change that a monitoring system engendered was awareness without direct in-person monitoring.This sentiment was most frequently expressed by participants during two tasks in particular.The first of these tasks asked participants to ensure that all individuals were in the dining room for breakfast.P21 explained how she would use the system for this task as follows: "So the first thing I would do because I like to cheat, is that I would look at the map.And I will see if everybody's in the dining room.[...] this would be great for me being a director because I will be able to sit at my desk and I will be able to log on to the maps and I will be able to see the location of my residents and the ones that are not in the dining room.I don't have to get up and go to their apartment and check, I can give my walkie-talkie.So there, and I'm going to call my staff to say, why isn't ABCD in the dining room?Because I can clearly see that they're still in their rooms, so that would work for me." In other words, participants found that having this awareness without the need for in-person monitoring would allow them to provide more direct and efficient care by reducing the burden of information gathering.In the current caregiving paradigm, much of their workflow entails seeking out information in order to know what care they need to provide.Information is typically gathered from communication with other caregivers, searching through documentation, or observing the residents directly.P24 explained this burden as follows: "We have to kind of do everything on our own by ourselves, if you have to track, you have to go back and retrieve archives or whatever information [you need].And we have pagers that actually help, but [they] can't tell you specifically where the person is.What we have right now [is] if the pager goes off, it says maybe [main street], the person might be close to me, but not specifically be there" In contrast, participants interacting with the health monitoring interface found that this information was available in a centralized location and frequently used the Resident Summary, Map, and Notes views to gather the information needed to complete caregiving tasks.This reduction in the onus of information gathering was especially notable during a task in which participants had to respond to a fallen resident.P21 explained the problem with the current paradigm as follows: "So when you go in every two hours on your shift, that's every two hours that you're taken away from your resident [...] I can go and check on you right now and you fall [right after] I just came in and woke you up and checked on you.[...] I'm not going to find you for another hour and forty-five minutes." In contrast, participants noted how the combination of the fall alerts and the ability to view the location of the fallen resident was immensely helpful, providing them with immediate awareness of the situation.P24 described this as follows: "It might like, instead of you running around looking everywhere, you can just go directly to where they're at.It might save a lot, like save someone from actually getting into a bad situation." 5.4.2Simplifying but not completely automating workflow.In addition to the responsibilities of gathering information and making decisions in order to administer appropriate care, many facilities require caregivers to document these actions.In this particular facility, several caregivers noted that the software had rigid requirements that forced them to adapt their workflow to fit the needs of the software, rather than the software adapting to support the caregiving expertise of the caregivers.P28 noted how the existing software required caregivers to fill out irrelevant documentation: "It actually a whole bunch of stuff that don't even [apply], yeah?[...] On there it got 'Did you put Paula's hearing aids in today?' No, I ain't put Paula's hearing aids in today 'cause Paula ain't got no hearing aids: she don't even wear hearing aids.
Crazy stuff like that.That's a question that could've been, can be deleted altogether, avoided." Thus participants found that the use of a health monitoring system would simplify their caregiving workflow by both reducing the burden of information gathering and removing the need for redundant and irrelevant documentation.Notably, participants did not feel that this system would fully automate the caregiving workflow.Participants commented that they would still have to physically visit and assist residents in need of care.P22, discussing the task of ensuring that all residents were in the dining room for breakfast, noted: "You can see what's going on right there [on the screen], but you still need to go and see them because you know that he's had a shower, but he's not dressed so you still need to go see them.This helps you to know what has been done with them, so that's good.So you would need to go see them physically.So I don't if that's a good thing or not." 5.4.3Details about recent care and current status are vital.Participants frequently made use of the current status and knowledge of recently provided care to inform the way they completed tasks.During a task in which participants were asked to prepare a fictional resident for a doctor's appointment, participants leveraged information provided in the ADL summary to see what care the "other caregivers" had already provided in the fictional facility.Describing their thought process as they provided care, they would list what ADLs the resident had performed already, confirming that he was clean, toileted, and fed.During the task in which they had to ensure residents were in the dining hall for breakfast, several of the participants checked the ADL summary to see if residents were in the process of getting ready or were still in bed and needed to be roused.The element of care that received perhaps the most attention was medication.In comparison to other activities, participants paid more attention to the details of this activity, wanting to know what medication was given, how much was given, and when it was administered.P21 explained: "So if we gave him medication, then it needs to be specifically noted at the time that way someone else don't come in and give him the same medication that I just gave him." 5.4.4Reducing communication inefficiencies.From our interviews, we found that facilitating communication between caregivers was a highly valued feature within the system, as it offered a mechanism for alleviating the burden of communication on individual caregivers in the current formal caregiving paradigm.There were several ways in which participants noted this system would improve communication.One of these was cross-shift communication where the relieving staff need to rapidly adjust to the current state of the facility without any context because they did not encounter the staff who just finished their shifts.P24 described this problem succinctly: "Yeah, we just have to, you know, verbally communicate what actually happened during the day to the oncoming shift.So if you're not there to do it, it's like breaking communication.[...] I have had it happen to me probably two or three times [in the] last two weeks 'cause I had a fall and the nurse relieving me, she came after 10:00, I had to leave.So that was a break in communication because I couldn't tell [her] what happened to go ahead and do documentation." Participants also noted that this system would render some forms of communication unnecessary.P22 gave the following example of this: "I like that I could put a note in so the CNA's could see that they got the PRNs and then you know what they're updated on things that they would come and ask me about.Like you know what 'cause they come and ask me "Did you order this, did you order that?" and here I can just put in a note that I ordered Alex's medication, it should be here tonight." Lastly, participants found that the communication system provided valuable context to the information provided by the health monitoring system.In other words, the communication system on top of a health monitoring system allowed caregivers to provide context to the displayed information and direct the attention of other caregivers to a particular piece of information to inform them of care they provided.

Study Design
In practice, monitoring systems are unlikely to succeed without acceptance from both caregivers and older adults.To that end, we conducted a user study with five residents (code-named P31-P35) in an assisted living facility where they participated in a simulation of the envisioned health monitoring system.This study was conducted in the same CCRC in the Northeastern USA in which we conducted the first round of caregiver interviews.The staff at this CCRC recruited internally to guarantee that these residents were cognitively fit to consent and participate in the study; two of the participants had motor impairments with one using a wheelchair and one using a walker.The participants' demographics can be seen in Table 5.The study itself consisted of a pre-interview, a week-long period during which residents participated in a simulation of a health monitoring system by wearing smartwatches, and a post-interview.Interviews lasted 20-30 minutes and were audio-recorded for later transcription and analysis.The pre-and post-interview scripts are available as supplemental material.
As described in Section 2.2, researchers have shown that both wearable inertial sensors and environmental sensors can be used as part of monitoring systems; in this work, we opted to use smartwatches as our monitoring interface.In comparison to environmental sensors, wearable sensors are not tied to any single environment and not subject to lighting and occlusion.Furthermore, as Caldeira et al. [13] notes, smartwatches are general purpose devices that are less subject to the stigma commonly associated with other aging-in-place technology.The goals of the pre-interviews were to understand participants' preexisting experience using technology and to gauge their potential acceptance of a technology designed to monitor their ADL performance.To that end, we begun the interview by asking participants what types of technology they used, what they liked and disliked about technology, how their use of technology has changed after moving into personal care, if and how they use technology to manage their health, and how they think technology could possibly support them through the aging process.In the last portion of the interview we asked them to imagine that a system was put in place within their facility that recognized their ADL performance and allowed caregivers to monitor that performance.Then we asked them how they would feel about the use of such a system and how it could both benefit and disadvantage both residents and caregivers.
6.1.2Watch Study.We simulated interaction with a health monitoring system by asking participants to wear two Polar M600 smartwatches, one on each wrist, every day for a week while they went about their normal schedule.In an actual implementation of a smartwatch-based health monitoring system, sensor data from these watches would be input into machine learning algorithms that predict the status of the resident; this information would be displayed on the interface used by the caregivers.However, in our implementation sensor data from the watches was not used in this manner; rather, we simply collected the accelerometer and gyroscope data generated by participants during the study.Analysis of this sensor data is outside the scope of this work.Additionally, participants did not have to directly interact with the smartwatches with the exception of putting them on in the morning and taking them off at night, i.e., the data collection happened passively without participant interaction.A researcher visited participants every night of the study to ensure that the watches were charged overnight.Participants were provided a printed-out logbook with entries for writing down the times when ADLs were performed and any notes.

Qualitative Data Analysis
We analyzed the interview data using thematic analysis.Three computer science researchers who are co-authors on this work independently coded the interviews, discussed the extracted codes, and settled on final themes.These themes can be found in Table 6.With the exception of P33, participants had no desire to learn or use unfamiliar technologies, regardless of their reported comfort level with using technology.Participants either felt that they were too old to learn how to use new devices or expressed that they were content with their daily life and saw no need for new technology.For instance, P32 noted that a close relative bought her an expensive laptop but that she could not get used to it.When asked why she explained: "If I had gotten to it maybe two years ago I probably would have been fine.But I just did not want to bother with; I was at a loss [as] to how things worked.Now maybe if I was more interested in technology, I would have spent the time to do it.It's too late for me because I've come to technology so late." Participants currently use some assistive technology within the facility, most notably an emergency pendant that is worn around the neck.None of the participants liked wearing the pendant, considering it ugly and noting that it has a very limited range.P34 attributed her unwillingness to wear the pendants to personal dislike of the process: "I don't, no.That's just stubbornness.I have it and they keep checking it [...] I believe at least every two months or something they make sure everything works.I don't wanna hang it around." On the subject of health management, participants were generally aware that they were being monitored by caregivers.The most noticeable form of monitoring to them was location during meal times, as caregivers are required to ensure that residents are present in the cafeteria during meals.Other participants noted that caregivers would come around and check vital signs regularly, ensure that residents took medications at appropriate times, and would have a staff meeting in the mornings presumably to discuss residents' statuses.Participants had mixed feelings about the potential implementation of an ADL-monitoring system.On the one hand, P31 noted that if it was done, he was confident that it would be done for a good reason: "If it's important to have it done, then I have it done.[...] Whenever they ask for something this, that, and the other thing, it's important." P35 echoed this idea: although initially comparing such a system to Big Brother, she followed up by noting that it could be beneficial for many of the older adults living within the facility as well as their caregivers.On the other hand, a couple of the participants were against the idea, feeling as if it was too much of an invasion of privacy.6.2.2 Post-Interview Themes.Despite asking residents to wear smartwatches on both wrists from morning to evening every day for a week, participants had no major complaints about wearing the watches on a daily basis.P31 went as far as to note: "I didn't have any feeling one way or the other, they became a part of me.[...] Well the first day I had this difficulty in [using them], but after that it was very easy, then they were excellent really.Yep, I had no problem with it." However, participants did bring up minor issues they had, noting they would like different options for watches as they were a little too big for their wrists, an ugly color, or difficult to put on given their poor eyesight.When asked if their opinions regarding the implementation of an ADL-monitoring system had changed, participants largely echoed their initial viewpoints, albeit with the added emphasis of wanting to maintain their independence.Both participants who had felt that the system was a step too far noted that they were okay with using such a system but still had reservations about the specific implementation and the privacy implications.P34 was perhaps the most apprehensive about the widespread implementation of the idea, noting: "I know at our eating table, there are four of us, and we change often with other people, but I know one in particular.I think she was asked but she didn't want any parts of it because it's, she feels technology knows enough about her.So, I think that's a big hurdle for tech people." While P33, who already wore a smartwatch daily, expressed that he would be amenable to having his activities tracked, he was unsure of how others might react.However, he did offer a potential way of implementing such a system: "Well, I think people would raise the issue of privacy.I don't know, I just really don't know how other people would feel about that.They may feel that the nurses are acquiring too much information.But they can also be motivated to keep up somewhat of a daily routine.You know, if you know you're being watched you might be more motivated to perform.Despite these opinions, participants acknowledged that such a system would be useful for caregivers.P35 provided the following example use case: "So, let's say somebody was getting weaker and [...] all of a sudden, you're not seeing them going around and doing their laps.That I could see, it would help because they're getting older every-and they're sitting more during the day.And these people aren't necessarily talking with their friends on the phone or on the iPad."

DISCUSSION
In this work, we hoped to gain an understanding of how best to design monitoring systems to support and augment the current caregiving paradigm for older adults in assisted living facilities.To that end, we interviewed caregivers in formal caregiving environments at two facilities in the United States to gain insight into their workflows, and we also interviewed older adults to gain insight into their opinions regarding the usage of monitoring systems.Through our conversations and observations, it became clear that although these systems can offer clear benefits, unless they are designed carefully they are unlikely to be welcomed and ultimately adopted long-term.Thus in this section, we offer a synthesis of the themes we found from all of the interviews we conducted and have described thus far.

Monitoring Systems Would Streamline the Existing Caregiving Workflow
The core responsibility of caregivers is to ensure that older adults are healthy and well.This entails ensuring that they are completing their ADLs regularly and adequately, their vitals are normal, and their needs are met on a day-to-day basis.In the current caregiving paradigm, this responsibility often requires a reduction in the independence and privacy older adults are accustomed to.For older adults, this can be frustrating and irritating and manifest as resistance and combativeness: issues that caregivers uniformly described as their biggest day-to-day challenge.With these perspectives in mind, both caregivers and older adults expressed that monitoring systems would be useful.For caregivers, this would streamline their workflow, allowing them to apply care as needed instead of on a schedule.For older adults, this would likely reduce the intrusiveness of care.
Our interviews revealed two areas of monitoring that would be particularly impactful.The first of these was location.Knowing where the older adults were in the facility would remove the need to manually find residents and allow caregivers to ensure that they have not wandered off to somewhere they are not supposed to be.The second of these was adverse events, especially falls.Having a system that notifies them when and where a fall occurs would allow caregivers to find residents quickly after the fall occurs.Caregivers and residents largely described existing systems for fall detection and other emergencies as ineffectual for a combination of reasons.Residents uniformly expressed a dislike for the fall pendants they were required to wear, a sentiment confirmed by caregivers.Caregivers noted that residents frequently found ways to get around wearing them and many residents didn't know how to use them.

Communication Improvements Would Go a Long Way
While communication has long been established as critical to successful and optimal elderly care [15], we found that there are opportunities to improve the state of communication between caregivers in both of the facilities in which we conducted our studies.At the CCRC, where we conducted our first round of interviews, caregivers noted that they did not use any form of EHR, instead making mental notes of care provided and changes in status and verbally communicating that information to other caregivers as needed.In the memory care facility, while the use of EHR reduced the emphasis on mental notes, caregivers still noted a reliance on verbal communication in certain situations, most notably in between shifts.While a system that purely monitors older adult behaviors would alleviate some of this need for mental notes and communication, augmenting this system with features that facilitate communication would increase the overall utility of the system.

Simplicity is Critical
In our discussions with both caregivers and older adults, we found that simplicity was highly valued.At the memory care facility, many of the caregivers we spoke to expressed frustration with the existing management system, finding many of its features superfluous and/or hindersome to the care they wanted to provide.Furthermore, at both facilities, we talked to multiple caregivers who were not native English speakers and they expressed that they would find any system with a lot of English text difficult to use.While caregivers would more than likely be trained to use any system used as part of their caregiving workflow, designing simple interfaces and user experiences with non-native language speakers in mind would go a long way towards ensuring monitoring systems are intuitive and can be used quickly and accurately during the caregiving process.
Simplicity is equally important for older adults.It has been well-documented that the adoption of technologies designed for older adults to improve their quality of life has been limited.This population generally has low-tech literacy, has concerns over the stigma and potential invasion of privacy associated with use of aging technology, and has a wide range of physical and cognitive abilities that can limit their ability to use such systems [13,96].Our findings are consistent with this body of work, as we found that older adults were adept at using specific technologies (often for leisure or communication with friends and family) but were largely reluctant or unwilling to try new technology.However, in contrast to these findings, we found that older adults did not mind the smartwatches we asked them to wear beyond expressing a preference for a different color or size.As an example, P31 in his pre-interview stated "Anything that's really new, I don't even attempt to do" but in the post-interview noted "Well the first day I had this difficulty in [using the watches], but after that it was very easy, then they were excellent really.Yep, I had no problem with it." We expect that this response was due to the familiar form factor and simple user experience.Furthermore, as general purpose devices that older adults could potentially use for their own personal wants and needs, smartwatches would be a well-received medium for aging in place technology [13].
We would like to emphasize that simplicity should not come at the cost of flexibility or opportunities for personalization.Just as some of the older adults wanted different colors or sizes of watches, some of the caregivers expressed a desire to view the application on their phones or on a computer.Furthermore, our studies looked at formal caregivers as a collective, but we expect that in reality caregivers with different roles will have slightly different needs from a caregiving interface.Designers and researchers should explore how to further personalize the health monitoring interfaces to meet the individualized needs of each group of people.

Striking a Balance between Older Adults' Wellbeing and Privacy
For any older adult living in a senior living community at any level, caregiving is an inherently intrusive endeavor.Caregivers need to observe, monitor, and often assist older adults in performing ADLs and fundamentally ensure that they are safe and well on a daily basis.In that sense, while a certain level of intrusiveness is unavoidable, monitoring systems represent a shift in the ways in which that intrusiveness manifests.The existing paradigm centers around caregivers physically and actively interacting with older adults as frequently as they feel is needed or are able, while monitoring systems track individuals constantly behind the scenes and allow caregivers to physically interact only as needed.
Older adults we interviewed commented on both of these types of monitoring.When asked if they were aware of how caregivers currently monitored them, they were able to list a variety of ways in which caregivers were keeping track of their health and wellbeing.Commonly in our interviews, this awareness would be followed by an explanation of how they had attempted to set boundaries in an attempt to maintain their autonomy.They were just as aware of how a proposed health monitoring system would infringe on their privacy, often following up this observation with an explanation of how comfortable or uncomfortable they were with caregivers having more information about them.Fundamentally neither paradigm of assessing the health and wellbeing of older adults will ever be completely popular among older adults and caregivers.With that in mind, for monitoring systems to see widespread adoption they need to strike a balance between having enough information to ensure that older adults are well and safe and protecting their privacy and sense of control over their day-to-day lives.

FUTURE WORK
The work we presented herein represents a step towards better care; however, there remains a number of areas in which further study is required before health monitoring systems become viable and practical caregiving tools.One immediate area of study would be confirming the utility of the systems presented herein through longer-term and more realistic studies.As noted in Section 7.3, all formal caregivers within any given facility do not have homogeneous duties and needs, and as such, these studies should place an emphasis understanding how such systems can benefit each group of formal caregivers within the facility.Furthermore, in this work we worked with a personal care home and a memory care facility; conceivably a similar system could work anywhere in the range of formal care options available, from independent living to skilled nursing.Thus, studies should explicitly explore the viability of these systems in these other formal care settings as residents in these facilities have different care needs and their caregivers have varying responsibilities.Finally, although we made a distinction in this work between formal and informal caregivers, researchers should explore the practicality and potential benefit of providing the tracking of ADLs, vitals, location, and adverse events to informal caregivers.

CONCLUSION
In this paper, we investigated the design requirements of a health monitoring system intended to improve the quality of care formal caregivers are able to provide to older adults in assisted living settings.To that end, we conducted a multi-stage qualitative study with caregivers and residents in assisted living facilities.Results from the thematic analysis of an initial round of interviews with caregivers were used to guide the development of AutoRounds, a prototype health monitoring interface designed to present resident information and status to caregivers in an easy-to-use tablet interface.A second group of caregivers was asked to interact with AutoRounds to refine the themes and values elicited from the first round of interviews.Additionally, we conducted thematic analysis on interviews with residents conducted before and after week-long use of smartwatches that were worn to simulate participation in a health monitoring system.By synthesizing our themes from the analysis of these interviews, we provide insights into what caregivers and older adults think of health monitoring systems.

13 : 11 Fig. 1 .
Fig. 1.List of Residents View: Displays all residents within the facility Fig. 2. Individual Resident View: Displays all the data associated with the selected resident

Fig. 3 .
Fig. 3. Notes View: Displays notes for all residents within the facility

6. 1 . 3
Post-Interviews.The goal of the post-interviews was to understand how participants felt about the experience of wearing smartwatches over the course of one week and what impact it had, if any, on their opinions of using technology for health management.To that end, we asked residents what they liked and disliked about wearing the smartwatches on a daily basis, if their opinion on technology use to monitor ADLs has changed, and what they think older adults in general might think about such a system.
Proc.ACM Hum.-Comput.Interact., Vol. 8, No. CSCW1, Article 13.Publication date: April 2024.A Step Toward Better Care 13:21 [...] It might be what difference does it make how much activity I [do]?It's not the nurse's business.So I think they have to be oriented to the fact that what you eat, what you do, and activities [you do are] the institution's business.That's why you're here." Proc.ACM Hum.-Comput.Interact., Vol. 8, No. CSCW1, Article 13.Publication date: April 2024.

Table 2 .
List of themes, design insights, and solutions from the caregiver interviews

Table 3 .
Cognitive Walkthrough Participant Demographics

Table 4 .
List of themes from qualitative analysis from cognitive walkthroughs 5.4.1 Awareness without direct monitoring.

Table 5 .
Demographics of Residents Interviewed

Table 6 .
List of themes from qualitative analysis on resident interviews Through our interviews, we found that most participants used some level of technology on a daily basis, typically in the form of specific apps and functionality on smartphones and/or tablets.Participants used technology to communicate with their friends and family members, read the news, shop online, and assist in maintaining an exercise routine.Participants did not use technology to track their own health, with the exception of P33 who tracked a number of health metrics using a combination of his iPhone and Apple Watch.P33 was by far the most comfortable with technology, actively seeking out and learning new technologies to use in his personal life.By contrast, one participant, P31, had no ability to use technology at all.When asked about his technology use he explained: "I try to limit it [my use of technology] because my ability is not very good for technicalities and stuff like that.They're not in my heart and mind.[...] Anything that's really new, I don't even attempt to do.I stay away from it."