2.1 Overview

The PRISMA flow diagram for the literature search was used for a systematic and transparent research process.

2.2 Step 1: Research questions

2.3 Step 2: Searching and identifying relevant studies

We developed a search strategy by identifying keywords using the Population, Concept, Context (PCC) framework (Pollock et al., 2023). Similar terms were explored using Mesh (PubMed), SvenskMesh, thesaurus.com and keywords used in similar articles searched through Google Scholar. Initial search strings were formulated and piloted in the relevant databases. The search strings were further refined by consulting with the librarian at the authors' affiliated institution. A systematic search was conducted in Scopus, Web of Science and PubMed databases. Keywords for search are available in Appendix A in Figure A1. The date of the final search was 27 February 2023. The full electronic search strategy for Pubmed is shown in Appendix B in Figure B2. The records were exported from each database into a web tool, Rayyan (Ouzzani et al., 2016), where duplicates were removed, and articles were screened by title and abstract for eligibility. After the final retrieval of selected articles, an AI-based search was done on the Research Rabbit (RR) (Sharma et al., 2022) as an added check to ensure no important articles are missed to be included. This added check helped us to identify 10 more research studies. After applying the inclusion and exclusion criteria on these studies, five of these were included in the list of primary studies for this scoping review.

2.4 Step 3: Study selection

To select the primary studies, three independent reviewers evaluated the initial set of retrieved research studies, and applied inclusion and exclusion criteria developed based on the research questions. The step-by-step application of the study selection process is shown in Figure 1.

2.5 Step 4: Charting the data

In this data extraction process, a data charting form was developed by identifying variables corresponding to the research questions. Data charting was done by the first author but, during the pilot run, the first three authors had discussions back and forth to finalize the template. Table C1 in Appendix C shows the charting of data done during this scoping review.

2.6 Step 5: Collating, summarizing and reporting the results

The collating and summarizing of the data was performed after the data charting. A general summarizing of the number of studies according to the location and year of publication was performed. The sample size was summarized from small to large, and a median was calculated. In addition to general summarizing, a specific collating and summarizing of the data was conducted in accordance with each research question. The technologies were categorized into wearables and non-wearables, while sensors used in the studies were categorized into different categories. The objective parameters of sleep were listed, and each study was vigilantly read to understand the definition of each parameter. Thereafter, several different parameters were collated into one to avoid any redundancy. For example, SD or total time of sleeping was collated into TST. For the third question, the data were collated into four categories and summarized. Consequently, this led to the fourth question, where the most common objective markers of sleep were chosen and identified common predictors and outcomes associated with them. The results were then reported to present a comprehensive view. The descriptive numerical summary was presented to provide key quantitative information, such as the number of studies reviewed and the distribution of sleep parameters across various studies. In addition, common themes/categories were reported to highlight recurring patterns or trends, such as types of technologies, predictors and outcomes associated with sleep parameters in the studies. Finally, a data charting table was included to organize and report the extracted data from each study.

3.1 General overview

Our search identified 2020 articles, of which 54 articles met the inclusion criteria. A PRISMA flow diagram of the scientific literature and selection criteria is presented in Figure 1. A total of 12 countries were represented in the studies. The two most common locations where the studies took place were the USA (n = 30) and Japan (n = 10). A total of three studies were conducted in Switzerland, two in Australia, and two in Canada, while all the other countries, as mentioned in Figure 2, had only one study each.

This scoping review showed a fluctuating publication trend with a surge in publications around 2021 (n = 10). The graphical representation of this trend is shown in Figure 3. The sample size in the studies ranged from n = 8 (Tsuzuki et al., 2015) to n = 6107 (Patel et al., 2008), with a median of 103. In 35 studies, most of the sample population was female, and only one study had an equal number of both genders (Obayashi, Saeki, Iwamoto, et al., 2014). In most of the studies (n = 21), the majority of the population was Caucasians (white), and in only three studies, the majority of the population was African or African American (black). All the studies were quantitative because of the nature of the research questions. Fifty-one studies were published in a medicine, health or public health journal.

3.2 Technologies and sensors

All the technologies and sensors used to monitor sleep are mentioned in the data charting Table C1 in Appendix C. The data were collated into wearables and non-wearables. Note that one of the studies used two devices to monitor sleep (Seelye et al., 2015). Among the 48 studies on wearables, 43 studies used wrist-worn with actigraphy sensors, while three studies used headbands with EEG sensors (Altendahl et al., 2020; Lustenberger et al., 2022; Scullin, 2013) and two used actigraphy sensors around the waist/hip (Full et al., 2020; Miyazaki et al., 2021). Among the non-wearable devices, three studies used near-bed devices, including in-home sensor networks using motion sensors (Seelye et al., 2015) and passive infrared sensors (Reynolds et al., 2022), as well as bedside monitors using radiofrequency sensors (Mahajan et al., 2021). Furthermore, four studies used under-mattress mats to monitor sleep with ballistocardiography (BCG; Gao et al., 2021; Kholghi et al., 2021; Rawtaer et al., 2020; Schütz et al., 2021), while one study used a motion sensor under the mattress along with in-home sensor networks using motion sensors. (Seelye et al., 2015). The study (Rawtaer et al., 2020) did not mention the brand name, except that it used fibre-optic technology with BCG sensors placed under the mattress. All the sensors used are summarized in Figure 4, where the size of each bubble represents the number of studies in which the sensors are used. Most of the studies used manufacturers' software to analyse the raw data from the device (n = 39), while five studies did not mention how the analysis of raw data was conducted. Thirteen studies utilized algorithm-based analysis or a combination of the manufacturer's software and algorithms to determine specific parameters like sleep onset and offset.

3.3 Objective parameters of sleep

Figure 5 shows all the objective parameters used in the studies to evaluate sleep. Our results show that the most common parameters used in the studies are TST (n = 44), WASO (n = 33) and SE (n = 32).

Six studies evaluated the sleep architecture with sleep stages (Altendahl et al., 2020; Lustenberger et al., 2022; Mahajan et al., 2021; Schütz et al., 2021; Scullin, 2013; Tadokoro et al., 2020). These parameters were obtained by using different kinds of sensors, for example, EEG (Altendahl et al., 2020; Lustenberger et al., 2022), BCG (Schütz et al., 2021), Radiofrequency (Mahajan et al., 2021) and Actigraphy sensors (Tadokoro et al., 2020).

Scullin (2013) used a headband device that measured only slow-wave sleep. The studies (Altendahl et al., 2020; Lustenberger et al., 2022; Schütz et al., 2021) assessed non-rapid eye movement (NREM) sleep, encompassing light-stage sleep (Stages 1 and 2), deep-stage sleep (slow-wave sleep, Stage 3), and REM sleep, serving as objective markers of sleep stages. Another study (Tadokoro et al., 2020) assessed the sleep architecture with the same parameters as in studies (Altendahl et al., 2020; Lustenberger et al., 2022; Schütz et al., 2021) in addition to REM latency.

3.4 Types of predictors and outcomes measured in studies of objective sleep monitoring

The total number of associations measured in 54 studies was 75. To understand the overall landscape of outcomes that are studied with sleep parameters, we collated the outcomes in different categories, as mentioned in Table 1.

Health-related predictors and outcomes include human immunodeficiency virus (HIV) status (Redman et al., 2022), cognitive performance (Bernstein et al., 2022), autonomic regulation through heart rate variability (Huang et al., 2022), daily functioning (Lustenberger et al., 2022; Parsey et al., 2015), peritoneal dialyses specific factors like haemoglobin and phosphorous level (Zhang et al., 2021), mild cognitive impairment (MCI; Gao et al., 2021; Rawtaer et al., 2020), Alzheimer's disease (Gao et al., 2021; Lim et al., 2013; Wams et al., 2017), perceived health status (Schütz et al., 2021), markers of cardiometabolic health (Full et al., 2020), neural structure, i.e. white matter of the brain (Altendahl et al., 2020), daytime cortisol level (Morgan et al., 2017), amnesic MCI (aMCI; Wams et al., 2017), inflammatory burden (Smagula et al., 2016), APOE e4 (a risk factor allele for Alzheimer's disease; Drogos et al., 2016), peripheral arterial disease (Nagayoshi et al., 2017), working memory (Miyata et al., 2013; Seelye et al., 2015), number of falls (Stone et al., 2014), obesity (Patel et al., 2014, 2008), episodic memory (Scullin, 2013), sustained attention and vigilance (Miyata et al., 2013), instrumental activities of daily living (IADL) impairment (Spira et al., 2012), grip strength (Spira et al., 2012), gait speed (Spira et al., 2012), subjective pain (Dzierzewski et al., 2010), frailty status (Ensrud et al., 2009), anxiety (Spira et al., 2009), testosterone levels (Barrett-Connor et al., 2008), cognitive decline (Blackwell et al., 2014; Lim et al., 2013) and depression (Maglione et al., 2008; Paudel et al., 2009).

Environmental predictors and outcomes that were measured in the studies for association with sleep parameters are seasonal variations in patients with MCI (Reynolds et al., 2022), perceived household environment (Johnson et al., 2021), interior housing conditions (Okoye et al., 2021), exterior housing conditions (Okoye et al., 2021) neighbourhood disorder (Okoye et al., 2021), evening indoor temperature (Saeki et al., 2015), initial bed temperature (Okamoto-Mizuno & Tsuzuki, 2010; Saeki et al., 2015), seasonal thermal environment (Okamoto-Mizuno & Tsuzuki, 2010; Tsuzuki et al., 2015), illuminance (Obayashi, Saeki, & Kurumatani, 2014; Tsuzuki et al., 2015), evening light exposure (Obayashi, Saeki, Iwamoto, et al., 2014), daytime-light exposure (McCurry et al., 2011), at-night skin temperature (Okamoto-Mizuno & Tsuzuki, 2010) and relative humidity (Okamoto-Mizuno & Tsuzuki, 2010).

Interventional predictors and outcomes are few that involved intraocular lens placement in cataract patients (Chellappa et al., 2022), auditory deep stimulation (Lustenberger et al., 2022), interval training with sleep hygiene (Mahajan et al., 2021), 3-month light to moderate intensity aerobic exercise (Miyazaki et al., 2021), CogniFit intervention (a cognitive training program; Haimov & Shatil, 2013), yoga intervention (Taibi & Vitiello, 2011), guided sleep education (NITE-D; McCurry et al., 2011), sleep hygiene program with bright and dim light exposure in morning or evening (Friedman et al., 2009), and daytime light exposure (Martin et al., 2008).

Behavioural predictors and outcomes consisted of mobility in terms of daily steps (Gao et al., 2021), behavioural and psychological symptoms of dementia (Cho et al., 2021), in-bed behaviours (Johnson et al., 2021), number of baths in 3 days (Tai et al., 2021), bathing duration (Tai et al., 2021), time interval between bathing and bedtime (Tai et al., 2021), and walking (McCurry et al., 2011).

Predictors and outcomes regarding time and place are measured by comparing sleep between two different times or places. These include sleep measures pre- and post-lockdown (Kholghi et al., 2021), nursing home residents versus community-dwelling residents (Kume et al., 2016; Olsen et al., 2016), and sleep in assisted living facilities versus sleep at home (Martin et al., 2008). Social predictors and outcomes are measured only in one study where partnership status and relationship quality are assessed in association with sleep parameters (Chen et al., 2015).

3.5 Predictors and outcomes associated with objective parameters of sleep

In our 54 studies, we found that several objective parameters of sleep were associated with several diverse predictors and outcomes mentioned in the previous section. Table 2 shows the predictors and outcomes that were associated with TST, WASO and SE. TST was associated with 14 health-related, four environmental, two interventions, three behavioural, two time and place, and one social predictors and outcomes. WASO was associated with nine health-related, four environmental, four interventions, three behavioural, one time and place, and one social predictors and outcomes. SE was associated with 10 health-related, four environmental, two interventions, two behavioural, one time and place, and one social predictors and outcomes.

The 10 common predictors and outcomes associated with all three most common parameters are shown in Figure 6. Two of these were health-related, frailty (Ensrud et al., 2009) and grip strength (Spira et al., 2012); three were environmental, which included exterior housing conditions (Okoye et al., 2021), seasonal thermal environment (Okamoto-Mizuno & Tsuzuki, 2010; Tsuzuki et al., 2015) and illuminance (Obayashi, Saeki, & Kurumatani, 2014; Tsuzuki et al., 2015); two were behavioural associations, like in-bed behaviours (Johnson et al., 2021) and mobility (Gao et al., 2021). One intervention, 3-month light to moderate intensity aerobic exercise intervention (Miyazaki et al., 2021), living place, i.e. nursing homes versus community-dwelling homes (Kume et al., 2016; Olsen et al., 2016), and a social outcome of partnership status (Chen et al., 2015) were associated with TST, WASO and SE. The objective parameters with sleep stages are associated with auditory deep sleep stimulation (Lustenberger et al., 2022), daily functioning (Lustenberger et al., 2022), Alzheimer's disease (Tadokoro et al., 2020), MCI (Tadokoro et al., 2020), neural structure, i.e. the white matter of the brain (Altendahl et al., 2020), and episodic memory (Scullin, 2013).

Key findings

Our findings provide knowledge of the overview of home-based objective sleep monitoring in older adults. The majority of studies on objective sleep monitoring at home in older adults were conducted in the USA. Among the European states, only Switzerland, the UK and Norway were represented, indicating a research gap in these regions. In recent years, especially during and after 2021, there has been an increase in studies focusing on home-based objective sleep monitoring in older adults. The need for telehealth solutions and home-based monitoring systems rose with the implementation of social distancing measures and the trend towards remote service. This change was necessary to maintaining healthcare delivery while lowering the risk of viral transmission, and it increased interest in creating and evaluating new technology that could facilitate these adjustments (Warrier & Sood, 2021). These immediate adaptations within healthcare research during the pandemic might be the reason for the reflected surge of research articles during 2021.

The review pointed out that actigraphy is the most used in-home sleep-monitoring method for older adults. The actigraphy technology collects data generated by movement or activity, and is either a wrist-worn or hip/waist-worn device. Although actigraphy is convenient and comfortable, a study shows that sleep parameters derived from actigraphy are less accurate than EEG (Rocknathan et al., 2017). In a study, it was found that actigraphy tends to overestimate light sleep bias more in the older age group than in the younger age group (Ong et al., 2023). EEG sensors monitor changes in brain activity, which in our review was performed through a headband device. The headband devices are comfortable and patient-friendly, have a high level of accuracy, and provide accurate physiological signals (Arnal et al., 2020). Compared with the gold-standard PSG, EEG headbands provided the lowest discrepancies in sleep parameters among other sleep monitors. These headbands are also recommended for evaluating poor sleep quality or when the highest degree of accuracy in sleep staging is required (Ong et al., 2023). There are other technologies than headbands that use EEG for sleep monitoring, such as ear-EEG sleep-monitoring plugs and patches (Mohamed et al., 2023). Our scoping review revealed a research gap in the use of these devices for older adults in their home environment. Another method of sleep monitoring we found in our studies is BCG. A BCG measures “the recoil force of the body in reaction to cardiac ejection of blood into the aorta and provides information on the timing of each heartbeat” (Starr et al., 1939). All of the BCG sensors, according to our review, are installed as under-mattress mats. The use of these devices has the potential to empower individuals to improve their sleep quality. On one hand, the use of screens and connectivity is generally perceived as disruptive to sleep. On the other hand, novel digital technologies could potentially facilitate sleep health and aid in the identification of disruptions of sleep parameters and sleep disorders (Jaiswal et al., 2019). A further review study is required that investigates the user-friendliness, non-intrusiveness, acceptance and compliance of these technologies in older adults (LeBlanc et al., 2022; Sadek & Mohktari, 2018).

The objective parameters that were most commonly found in our study were TST, WASO and SE. Sleep quality is measured using various technologies that assess different aspects of sleep. These technologies generate diverse parameters used to determine the objective parameters of sleep, such as TST, WASO and SE. TST refers to the total duration of sleep, WASO measures the total amount of time a person spends awake after falling asleep and is measured in minutes, and SE is the percentage of time spent asleep while in bed (O'Donnell et al., 2009). The literature supports that TST decreases with age from childhood to late adulthood, but then plateaus after 60 years old (Li et al., 2022). WASO, which is a measure of sleep maintenance, also worsens as aging occurs but remains unchanged after 60 years (Ohayon et al., 2004). SE remains consistent from childhood to adolescence and declines significantly with age in adulthood, and it declines very slowly further on with advancing age (Ohayon et al., 2004). There is significant night-to-night variability in objective sleep parameters in older adults (McCarter et al., 2022; Parsey et al., 2015). Even among those averaging the recommended 7–9 hr of sleep, 40% of nights fell outside this range (Scott et al., 2024). Daily monitoring of sleep parameters can be a powerful tool for older adults and their healthcare providers. It can help identify potential sleep-related disorders (Lechat, Scott, et al., 2023) at an early stage, and health factors that affect the variability in these parameters (Lechat, Loffler, et al., 2023). Single-night parameters may result in misdiagnosis and undertreatment of patients with mild sleep disorder symptoms during the monitoring period (Lechat et al., 2022). Therefore, these sleep parameters, through daily monitoring, provide a more comprehensive and accurate picture of an individual's sleep health.

There were several predictors and outcomes that were found to be associated with the most common objective parameters of sleep in older adults. They were health-related, environmental, interventions, behavioural, time and place, and social. Home-based sleep monitoring can be used to explore the link between sleep and various factors by measuring night-to-night sleep variability and objective measures of sleep across multiple nights. Analysis of sleep parameters and their association with multidimensional associations reveals a complex interplay that impacts healthy aging. Health-related associations such as frailty and grip strength are associated with TST, SE and WASO. Grip strength is often used as a measure of overall physical health and functional ability, while frailty, which encompasses physical, cognitive and social elements, has also been linked to reduced grip strength (Pana et al., 2021). Sleep quality and grip strength have a bidirectional association, with sleep quality affecting grip strength and vice versa (Han et al., 2023). The interconnectedness of objective sleep, frailty and grip strength highlights the importance of sleep in the overall wellbeing of older adults (Ravyts & Dzierzewski, 2022). Environmental associations such as exterior housing conditions, seasonal thermal environment, and illuminance are also found to be associated with sleep parameters TST, SE and WASO. Several environmental factors, both indoors and outdoors, play an important role in sleep quality. Exterior housing conditions, such as poorer housing quality, were found to be associated with shorter SD (Troxel et al., 2020). Being exposed to continuous ambient noise levels of 39 dBA or higher is connected to more frequent awakenings at night, shorter duration of sleep, diminished sleep quality, and a decrease in REM sleep (Marks & Griefahn, 2007). Light at night affects the sleep quality through increased awakenings and reduced slow-wave sleep (Cho et al., 2013). By facilitating an environment conducive to healthy sleep, public health workers can contribute to better sleep for older adults. Apart from environmental associations, in-bed behaviours such as watching TV or using screens have been found to be associated with TST, WASO and SE. A recent review highlighted that the relationship between technology use and sleep may be bidirectional. Technology use can affect sleep by displacing bedtime and delaying sleep onset, while sleep problems may lead to increased technology use as a way to fill time, provide cognitive distraction or aid in emotional regulation (Bauducco et al., 2024). These results about the association of objective parameters and sleep in older adults indicate that different behaviours, the environment, and the time before and after the pandemic can affect brain activity and, thus, inevitably affect sleep. Objective sleep monitoring, simultaneously with other technologies that help in relaxation and maintaining sleep hygiene, can be useful in improving sleep quality (Leonidis et al., 2021). The result of the analysis provides valuable insights to guide the development of future research. They can guide interventions to address key factors that significantly impact the quality of sleep among older adults. For example, interventions of changing in-bed behaviours or exterior environmental conditions may improve the SD, SE and sleep maintenance of older adults (Koch et al., 2006). Recognizing the associations between specific sleep stages and conditions such as Alzheimer's disease, MCI, neural structure and episodic memory can have significant implications for cognitive health. Addressing sleep disturbances in older adults could help maintain cognitive function and potentially lower the risk of cognitive decline (Nyholm et al., 2024).

There has been limited research on daily monitoring of sleep stages in older adults in home environments. Monitoring sleep stages, such as REM sleep and NREM sleep (N1, N2 and N3 or deep sleep), provides detailed insight into the quality and depth of your sleep (McCarter et al., 2022). The disorders in specific stages of sleep might be an indicator of a potential health condition. For example, increased time in stage N1 sleep and less time in stage REM sleep are associated with worsening cognitive performance in older men over time (Song et al., 2015). Disturbances in NREM sleep are associated with the development of Alzheimer's disease. This is due to the insufficient functioning of the brain's glymphatic system, leading to the accumulation of amyloid beta, a hallmark of Alzheimer's disease (Astara et al., 2023). Therefore, distinguishing sleep stages is important in older adults to monitor and diagnose sleep disturbances and health outcomes. Non-EEG-based technologies are currently not accurate enough to reliably distinguish sleep stages when compared with the gold-standard of PSG (Imtiaz, 2021; Yuan et al., 2024). Recent advances in wearables using EEG technology have shown high accuracy, especially in sleep staging, supported by feasibility studies and validation across populations. These devices offer the potential for more accurate at-home sleep monitoring, with high compliance and user-friendliness (de et al., 2024).

There are several strengths of this review. This review entailed a comprehensive search, which was reviewed and improved with the help of a librarian. The meticulous following of the methodology led to a detailed extraction process, which was discussed on a number of occasions among the three authors. To our knowledge, this is the first scoping review regarding home-based objective sleep monitoring at home in older adults, and its associations with diverse predictors and outcomes.

There are some limitations of our study that are also important to consider. Our search of the articles was limited to English, which may have led to the exclusion of important studies. The exclusion of the articles after 27 February 2023 and articles that published algorithms based on machine learning or described technologies without exploring the predictors and outcomes related to sleep parameters might have led to leaving out recent technologies. Furthermore, the exclusion of studies focused on sleep disturbances and clinical sleep disorders also limits the applicability of the findings to populations with sleep disorders. One limitation of this review is the inability to provide validation of the technologies provided, as this was out of the scope of the review. Future studies should comprehensively study validations, and disclose funding and conflicts of interest to enhance clarity and depth of information.