Loneliness is a common experience in later life and describes a subjective feeling, whereas social isolation is an objective state that is not necessarily associated with felt loneliness [1,2]. While people can be alone and lonely, they can also be lonely while cohabiting with their families [3]. Risk factors for loneliness and difficulties dealing with loneliness at an older age include physical health issues; lack of social resources, such as quality and frequency of contacts; lower educational attainment; and loss of a partner [3-6]. Loneliness can have health implications for individuals and economic impacts on the health care system. Loneliness has been shown to affect stress responses by increasing fibrinogen levels and diastolic blood pressure [5], risk of malnutrition [7], frailty [8], sleep deterioration [5], greater sensitivity toward self-rated health [2], depressive symptoms [9], and higher mortality risk [10].

Loneliness monitoring can help older people communicate with health care providers [11,12] and provide reassurance to family members. Another motivation is to assess loneliness objectively and understand its relation to health outcomes [11,13]. Previous systems used sensors at home, in corridors, and on doors and monitored phone and computer activities [11,12]. In this paper, we focus on the use of smart textiles in wearables and furniture for loneliness monitoring. Smart textiles have the potential to support a comfortable, accessible method of detection for physiological and behavioral data [14-16]. No previous systems designed to detect and systematically assess loneliness have used smart textiles for detection. Only 1 study detected the frequency of speech through a textile band; however, it did not evaluate whether this could be associated with the experience of loneliness [17].

Loneliness has been associated with depressive symptoms [18], sleep disturbance [19], sleep dysfunction, lower sleep quality, and differences in heart rate [20] and blood pressure [21]. Previous monitoring systems for loneliness also found associations with decreased daily phone use and incoming calls [13], time spent outside the home [11,12], the number of computer sessions [11], and speech characteristics [22]. Some of these symptoms could potentially be detected with smart textile sensors positioned on beds [23], embedded electrocardiogram sensors [24], sensors in insoles [25], or pressure sensors [26] on seat covers.

Design considerations for current loneliness monitoring systems concern the need for privacy, system unobtrusiveness [11,12], and durability (ie, battery efficiency) [27]. Having sensors installed intuitively on walls or devices was considered more comfortable and could be more easily remembered by older people than those on wearables [11]. There have been a few proposals for the application of a loneliness monitoring system. One proposal was to implement the technology in community settings to raise awareness for people who are on the verge of becoming lonely and to address the issue or share the data with physicians or nurses [11]. There is also a need to gain consent from older people to use the proposed technology [11]. Furthermore, a consideration is the question of how to serve lower-income and rural populations in accessing this technology [12]. No studies have presented in-depth findings on the needs and preferences of older people or stakeholders in loneliness monitoring systems using smart textiles. Thus, we drew on user studies and insights into design requirements from the literature on smart textile technologies.

Compared to pervasive computing systems, smart textile systems are predicted to acquire denser data, be less obtrusive, and reduce the number of devices required to detect diverse symptoms simultaneously [28]. Greater privacy is also a reason for positioning sensors in insoles for monitoring falls compared to alternative video cameras [25]. Positioning of sensors in furniture, such as a mattress topper, has previously been considered less invasive than worn sensors [23]. Positioning sensors on wearables, such as socks, allows for continuous monitoring including at home [29]. Even in older patient groups, there was a high adherence to the continuous use of wearable sensors, helping continuous monitoring of potential health risks [30]. Other examples of textile sensors include Taekwondo suits (pants and jackets) [26], floor mats [31], waistbands [24], and head ornaments [32]. Previous applications have contributed to the development of strain, pressure [26], electrocardiogram textile sensors, and explored the possibility of sensors with combined functions [26]. There has been a proposal for “intelligent health agents” based on smart textiles, which can support monitoring and treatment through textile technologies, such as thermotherapy, electrical stimulation, or phototherapy [33].

The commonly mentioned requirements are comfort [34-36] and wearables being made of breathable [34,37], lightweight, and permeable materials [34]. Furthermore, smart textile applications would need to function during daily and sports activities [35], be discrete [36], be safe on the body, and resist bodily fluids [35]. Aesthetic properties, such as color, fit, and diverse choices, to adapt clothing to the cultural context of the wearer and the possibility of mass production of such design choices should be further considered for greater acceptance [35,38]. Moreover, smart textile wearables must adapt to different body measures, genders, postures, and movements [35]. Further requirements include greater accuracy, reliability, privacy, and safety of the transmitted information [38]. The requirements for the material can vary according to the application context, but they did involve resisting large deformations and strikes in sports [26] and resisting abrasions from moving a topper on a mattress [23].

Furthermore, washability is considered an essential requirement for smart textiles [34]. Current applications can range from approximately 5 washing cycles by hand [26] to 40 cycles at 25 °C by a machine [36]. This implies potential additional requirements for washing [39]. One proposed solution to protect the textile sensors was microelectronic systems that could “be detached from the clothing component before washing” [40]. However, this might also contradict being a user-friendly washing process [38]. Thus, it is recommended that smart textiles are machine washable, support known washing processes, and do not require complicated washing instructions [38]. Power sources are also vital as currently available power sources are often too bulky [41]. Hence, one requirement is reduced energy consumption as one way to use less bulky power sources and the option to use energy from motion and body heat [34,35,38]. Enough power should be provided to last throughout monitoring activities [35]. Another important consideration is the recyclability and potential biodegradability of smart textiles produced without using harmful materials [34,41].

We are an interdisciplinary research group consisting of experts in gerontology, psychology, design, sensor technology, and machine learning, aiming to develop a loneliness monitoring and communication system from a human-centered design perspective [42]. This study aims to understand the needs of older individuals and the expectations of stakeholders to inform the system design.

We obtained user and stakeholder feedback from workshops and focus groups to develop the design requirements and initial design concept of the smart textile loneliness monitoring system for older adults. The study design and topic guides were developed in collaboration with team members who are developing the textile sensors and explored symptoms of loneliness in previous interviews [40]. For instance, sensor development required an understanding of user expectations in the positioning of sensors, washing, and charging. In addition, our initial prototypes developed and presented during the studies were informed by insights from interviews on potential symptoms of loneliness.

On March 27, 2023, and April 12, 2023, we conducted 2 co-design workshops with people aged ≥65 years who had experienced loneliness (n=5). On June 7, 2023, and June 15, 2023, we conducted a second co-design workshop with stakeholders (n=7). Therefore, we derived a set of design requirements implemented in an initial service system scenario and a smart home wearable system prototype. In October 2023, the design requirements were evaluated in a focus group study that included older people and stakeholders (n=7).

Participants for the workshop and focus group studies were contacted based on their previous expression of interest in the DELONELINESS project, as specified by Rees et al [4]. Invitations were sent via email newsletters within a housing network specialized for older people. We engaged individuals who had previously responded through a newsletter in a study involving older participants. Participants were also contacted through a study pool on wearable technologies from the Remote Assessment of Disease and Relapse-Major Depressive Disorder program focused on the passive monitoring of depressive disorders through smartphone data and wearables. Older participants were eligible for the workshops if they were aged ≥65 years, spoke fluent English, and experienced loneliness within the last 10 years since reaching the age of 65. Participants were ineligible if they had any cognitive impairment or dementia.

The severity of loneliness measured by the University of California, Los Angeles 3-item Loneliness Scale [43] has been associated with various health effects and restrictions. Higher and medium levels of loneliness were associated with a higher risk for frailty than lower levels of loneliness [8]. Another study showed that higher levels of loneliness were associated with greater mortality risk [9]. Higher levels of loneliness were also associated with people who might face functional restrictions, which have potential difficulties in engaging in activities [44]. Severe loneliness might have different requirements for a monitoring system and linked interventions. When participants were unable to attend on the day due to ill health, follow-up sessions were organized.

For both the workshop and focus group, stakeholders were recruited through the People in Research website by the National Institute for Health and Care Research to offer opportunities for public engagement and through the researchers’ professional networks. The eligibility criterion for stakeholders was being aged >18 years. We searched for family members of older people who had experienced loneliness, industry representatives, carers, social workers, social prescribers, housing providers, and charity workers.

Participant characteristics in the studies have been summarized in Table 1 and Textbox 1.

We chose a qualitative approach with the aim of gaining more in-depth insights into participant perspectives and needs, which serves well to inform the initial system design. Previous qualitative research on design requirements for technologies for susceptible people and with stakeholders have been based on sample sizes such as 12 patients and 4 stakeholders [45], 10 experts to develop and 4 to evaluate a system [46], or 15 stakeholders [47]. For a qualitative evaluation in user research, we followed the recommended participant number of 3 to 15 [48].

The workshops and focus groups were transcribed by the first author and, in part, by a professional transcription service with the consent of the participants. Our analytical approach was an inductive qualitative content analysis [53]. Content analysis is a “technique for making replicable and valid inferences from texts...to the contexts of their use” [50]. A qualitative approach differs in coding, the determination of categories, and how researchers interpret data frequencies [54,55]. Categories are often developed inductively close to the dataset at hand and reading data in depth rather than by applying automatized analysis approaches based on predetermined codes [54,55]. According to the approach [53], we first applied open coding by reading the material, assigning codes, and developing categories with the help of NVivo (Lumivero) software. Then, the codes were classified and grouped into categories (eg, “accuracy in detection,” “economical inclusivity,” and “instructions for carers”). The categories developed were further grouped into superordinate categories (eg, “general system requirements,” “wearable requirements,” “installation and maintenance requirements,” “material requirements—wearable,” and “material requirements—furniture”). For part of the material, the coding was done independently by a second researcher. Differences in coding were resolved through discussion. This led to additional codes in a few categories. The categories were further presented to 2 coauthors. This led to the summary and further grouping of codes, such as soft and lightweight materials. Moreover, it led to renaming some categories, such as “machine washing” rather than “direct washing.” We considered frequencies and how often different participants mentioned a concept. The detailed quotes and frequencies in each category are enlisted in Multimedia Appendix 2. This served to structure the observed requirements. The frequency count is not intended to be statistically significant or generalizable but to better compare different requirements and their potential importance.

The study was approved by the research ethics committee at King’s College London (LRS/DP-22/23-34602) on January 16, 2023. Participants were provided with an information sheet and the consent form before the initiation of the study for their information and they were informed about being able to opt out at any time in the study and asked to sign the consent form on the day the study started. In all data processing, data were labelled with a unique non–personally identifiable participant identification number. This made the data pseudoanonymous. A password-protected file contained the name and contact details with an ID that is stored separately. Participants were not provided with any direct incentives to participate. However, travel and lunch costs were reimbursed.

The general system requirements addressed the overall system function or were mentioned concerning multiple aspects of the system.

This research aimed to understand user and stakeholder requirements for a smart textile loneliness monitoring system. Overall, a key finding is the participants’ concern regarding what applications and interventions become linked to the sensor monitoring system for loneliness. Previous work found that older adults’ acceptance of smart textile technology and smart home devices depends strongly on perceived usefulness [56-58]. Whether the monitoring system was perceived as useful often depended on what actions would be taken in response to the monitoring. Previous rationales for developing loneliness monitoring provided in the literature were the importance of loneliness as a health concern [11,59]; the identification of loneliness, as directly communicating it might be associated with stigma [11,12,22]; potential bias in self-reporting and the nonobjective reporting of one’s state, also due to cognitive decline, to help prevent loneliness early on [11]; and research purposes to better understand the relationship between loneliness and health outcomes [11,13]. While questioning older participants and stakeholders, we found that participants perceived particular applicability in dementia, particularly as family members who wished to be notified of the status of loneliness of their loved ones. This might require future research to better involve such susceptible populations in the development of loneliness monitoring systems for accuracy. In contrast, it raises issues of consent and the state at which it would be achievable. Other potential applications considered by participants were buddy systems, being part of a group to improve technology development for older people to provide agency, and an artificial intelligence–recommendation system.

A proposition has also been made to potentially detect changes in individuals earlier on, in terms of loneliness, than is currently possible [13]. These suggestions elicit questions about available infrastructure and about an efficient distribution of the system. In case a loneliness monitoring system is intended warn people of loneliness in its early states [13], it could require a wider distribution than what might be affordable. A suggestion was to involve people who are in regular contact with potentially affected people to support a more targeted distribution.

Our findings showed that some older adults felt that the monitoring process did not solve their loneliness directly. Some also thought that they would not need a system to tell them when they are lonely.

Other propositions are informing family members or caregivers [12], overcoming stigma to receive help where necessary, and assessing current risks for greater mental health consequences [12]. Some declined the idea of contacting their family members in response to detected loneliness due to fears of burdening them. Stakeholders also noted that time constraints by health care professionals or social work professionals would not change via the introduction of a monitoring system. In contrast, 2 family members wished they had such a system to inform them about the well-being of susceptible family members (eg, those with mild dementia) across distance.

Similar to previous literature, participants expressed a preference for natural, soft, and lightweight materials useful in daily activities [34,37]. They also expressed a need for comfort [34-36] and discretion [36]. We found that furniture-embedded smart textiles (on a mattress or sofa) were rated more positively regarding comfort and ease of remembrance than smart apparel (eg, a smart vest), which has been predicted in previous literature on loneliness monitoring systems [11,23]. This was not only due to the individual preferences in clothing (whether one was used to wearing a vest or a watch) but also due to the greater potential effort in washing and charging. Therefore, it should be possible to consider individual preferences in the choice of sensor devices for monitoring [58]. In addition to the cultural sensitivity for clothing and wearable styles [35,38], we observed the same considerations in the context of textiles on furniture. In our study, we observed that colors that could elicit negative emotions should also be avoided, particularly for mental health applications. This aligns with findings from previously developed websites, robots, and devices to address the loneliness of older people [60-63]. Many vests and smart textiles developed for older people do have a plain design [64]; however, there have been exceptions that experimented with more friendly motifs of cats and flowers on smart textiles for patients with dementia [65]. Participants required the current system to function along with their existing furniture and clothing as they were unlikely to be replaced. Embedded electronics and their positioning must be compatible with other wearable health care devices, such as pacemakers or stoma bags. Regarding considerations of reliability and accuracy [37,38], participants also noted the possibilities of manipulation by them and the potential impact through people’s movement.

Washing and charging procedures are key concerns for the usability of the proposed technology. Current washing cycles [26,36] are not sufficient for user requirements, as participants did not agree with a life span of approximately 3 months. Most people did not accept special washing requirements, such as hand washing, or the use of a special washing cycle [39]. The proposed solution of detachable components [38,40,65], which could protect electronics from being washed as part of the overall textile, appeared more reasonable for furniture but less applicable for wearables that must be washed often. Participants emphasized the importance of washability of all parts (in particular, the materials close to the skin). Smart textile wearables and furniture covers are required to be washed frequently, due to the risk of spilling liquids at an older age or the possibility of incontinence. For maintenance, there will also need to be notifications and instructions for charging and washing and the requirement for fewer plugs, which is supported by previous findings from technologies addressing loneliness for older people. While current charging options are developed for smart textiles to become less obtrusive, they need to consider the potential limited dexterity of older people to handle them while washing or charging.

There has been little understanding of the needs of stakeholders and older people for the development of smart textile loneliness monitoring systems. Our insights were based on people envisioning the system by presenting a prototype and use case. Therefore, it was not possible to determine the proposed system’s actual usability. In the future, we would like to undertake system testing in the field.

This study presented key design requirements for a loneliness monitoring system embedded in smart textiles in wearables and smart home furniture from the perspectives of older people and stakeholders, touching on questions of usefulness and usability. Sensors on smart home furniture seem to face fewer difficulties in terms of individual habits, washing, and charging than applications that are wearable. Overall, the linked intervention in response to monitored loneliness remains a question that needs more in-depth understanding, including what individuals perceive as most valuable in connection with such a sensing system. We found additional aspects in the system design that have not found much consideration in previous work, such as the need for user cooperation, compatibility with other worn medical devices, user expectations regarding life span and washing requirements, questions about identifying the recipients of the system, and initial insights into possible linked interventions and purposes of loneliness monitoring systems.