Objective Evaluation of Sleep Disturbances in Older Adults with

Cognitive Impairment Using a Bed Sensor System and

Self-Organizing Map Analysis

Tomoko Kamimura

, Risa Otsuka

, Asaka Domoto

, Hikofumi Suzuki

and Mamino Tokita

Department of Health Sciences, Graduate School of Medicine, Shinshu University, 3-1-1 Asahi, Matsumoto, Japan

Department of Cyber Science Infrastructure Development, National Institute of Informatics, Tokyo, Japan

Department of Global Centre for Advanced Research on Logic and Sensibility, Keio University Tokyo, Japan

Keywords: Sleep Disturbance, Total Sleep Time, Cognitive Impairment, Alzheimer's Disease, Self-Organizing Map.

Abstract: Bed sensor systems are useful for measuring sleep states in cognitively impaired older adults because they

can measure unrestrained individuals. However, there are no criteria for identifying sleep abnormalities using

them. We developed a method to determine sleep abnormalities by analysing data collected by a bed sensor

system using a self-organizing map (SOM). In this study, the sleep states were measured in two cognitively

impaired care-facility residents. These recordings were used to calculate total nocturnal sleep time, wake time

after sleep onset, frequency of leaving the bed, and frequency of awakening in the bed for each day. The data

from these four variables were used to draw an SOM for each individual’s sleep state to identify normal or

abnormal sleep days. We visually determined whether a main cluster was formed in the SOM. If a main cluster

was formed, the days included in the main cluster were defined as the individual's normal days, while other

days were defined as the individual's abnormal days. The above parameters were independently compared

between the two groups, as determined by the SOM. The characteristics of abnormal sleep days identified by

SOM could be explained using these four variables, suggesting the effectiveness of identifying abnormal days

by SOM.

1 INTRODUCTION

Sleep disturbances in older adults with cognitive

impairment adversely affect health status and

increase the burden on caregivers (Webster 2020a;

Okuda 2019; Shi 2018). In addition, it has been noted

that sleep disturbances may exacerbate brain damage

(Irwin 2019; Nedergaard 2020). Therefore, early

detection and management of sleep disturbances are

important for not only maintaining the quality of life

(QOL) of older adults and their caregivers but also for

maintaining the health care system.

Sleep disturbances in older adults with cognitive

impairment have been estimated to range from 20%

to 70% (Guarnieri 2012; Wilfling 2019; Webster

2020b), depending on the assessment method. A

meta-analysis of studies using validated proxy

questionnaires found a pooled prevalence of sleep

disturbance of 20% (95% confidence interval [CI

https://orcid.org/0000-0003-2973-2064

16%-24%) among cognitively impaired individuals

living in care homes.

Sleep assessment methods other than proxy

questionnaires include validated self-assessment

questionnaires, polysomnography, and wearable

actigraphy. However, these methods are often

inadequate when individuals are very old or

cognitively impaired because the results are

unreliable or burdensome to the individuals.

To solve this problem, the sleep state of

cognitively impaired older individuals has been

measured using a bed sensor system that measures

sleep state without restraining the individual, and

abnormalities, such as frequency of leaving the bed

and prolonged or shortened total sleep time, have

been identified (Higami 2018). However, the

characteristics of these sleep states vary depending on

the measurement index, and the lack of criteria for

comprehensively determining abnormalities within

Kamimura, T., Otsuka, R., Domoto, A., Suzuki, H. and Tokita, M.

Objective Evaluation of Sleep Disturbances in Older Adults with Cognitive Impairment Using a Bed Sensor System and Self-Organizing Map Analysis.

DOI: 10.5220/0012715000003756

Paper published under CC license (CC BY-NC-ND 4.0)

In Proceedings of the 13th International Conference on Data Science, Technology and Applications (DATA 2024), pages 253-258

ISBN: 978-989-758-707-8; ISSN: 2184-285X

253

and between individuals remains a challenge.

To tackle this issue, we developed a method to

detect abnormalities by analysing data measured by a

bed sensor system with a self-organizing map (SOM),

which is an unsupervised learning clustering method

of artificial intelligence (AI). The SOM is a two-

dimensional plot of data; the shorter the distance

between the data, the higher the similarity of the data.

The reason for using this method is that it is suitable

for comparing the similarity of data with

multidimensional features in an exploratory manner.

In this study, to obtain basic data for the

development of a method for the detection of sleep

abnormalities in cognitively impaired older adults

using a bed sensor system and SOM analysis, we

detected abnormalities based on the similarity of

sleep states within individuals using SOM to extract

their characteristics and confirm whether these

abnormalities can be expressed using conventional

sleep indicators.

2 METHODS

2.1 Participants

Participants of this study were residents of a geriatric

healthcare facility in Japan. The inclusion criteria

were (1) a cognitive function score validated by the

Japanese Ministry of Health, Labor and Welfare

(Tago 2021) of II b (i.e., symptoms, behaviours, and

communication difficulties that interfere with daily

life are sometimes observed at home, but the patient

can be independent if someone pays attention to

them) or worse; (2) the need to monitor sleep and

leave the bed during the night and use a bed sensor

system; and (3) sleep data had been collected for at

least 21 consecutive days.

2.2 Equipment

The bed sensor system used in this study was a Nemuri

Scan (Paramount Bed Corporation), which was

installed under the bed mattress. Equipped with a

highly sensitive pressure sensor, the system detects the

body movements of the examinee on the bed through a

mattress and calculates an activity score every minute

that reflects the intensity and frequency of body move-

ments, excluding movements caused by respiration and

heartbeats. Nemuri Scan uses a proprietary algorithm

to detect one of three states per minute: the examinee

leaves the bed, awakens in the bed, or sleeps in the bed.

These results are directly output as comma

separated value (CSV) files, as well as daily sleep

indices, such as total sleep time and time awake after

falling asleep, calculated from the data. The validity

of these indicators has already been verified (Kogure

2011). Only the CSV data were used in this study.

2.3 Data Analysis

An SOM was created using CSV data from 6:00 p.m.

to 8:00 a.m. during the measurement period for each

individual. Days with missing CSV values during the

measurement period were excluded from the analysis.

Four variables (see Table) were calculated daily using

the CSV data: total nocturnal sleep time, wake time

four values were used to determine the SOM as the

sleep state of the day.

In the SOM, we visually determined whether the

main cluster, which accounts for the majority of the

data, was formed, and defined days included in the

main cluster as the individual's normal sleep days and

days not included in the main cluster as the

individual's abnormal sleep days. R studio ver.

1.3.1093. was used to create the SOM.

To confirm whether the abnormalities detected by

the SOM can be expressed using conventional sleep

indicators, the values of each of the above four

variables were compared between the two groups of

normal and abnormal sleep days in each case, as

determined by the SOM. Comparisons between the

two groups were made using the Mann–Whitney U

test, and a p-value <0.05 was considered statistically

significant using SPSS ver. 26.

Table 1: Four variables were calculated daily using the CSV data.

DATA 2024 - 13th International Conference on Data Science, Technology and Applications

254

2.4 Ethical Consideration

This study was approved by the Medical Ethics

Committee of Shinshu University School of

Medicine.

3 RESULTS

Data were collected from a 97-year-old woman for 28

days (case A) and a 91-year-old man for 51 days (case

B). There was 1 day of missing data in each case

during the measurement period, and the data of that

day were excluded from the analysis. The purpose of

using the Nemuri scan was as follows. Case A was at

risk of falling due to anaemia, cognitive decline, and

a history of hip fracture, and required monitoring the

transfer of a portable toilet at night. Case B was at risk

of falling or getting lost when going to the toilet at

night due to Alzheimer's disease and a history of

lumbar spinal stenosis and needed to be monitored

when going to the toilet at night. The main cluster was

formed in the SOM for case A (Figure 1). This cluster

consisted of data from 20 normal sleep days. Eight

abnormal sleep days fell outside the cluster.

The median (interquartile range) of total nocturnal

sleep time, WASO, frequency of leaving the bed, and

frequency of awakening in the bed on the abnormal

days in case A were 570.0 (482.5-602.5) minutes,

103.0 (69.8-121.8) min, 2.0 (0.5-3.8) times, and 11.0

(6.0-12.8) times, respectively. The median of each of

the four variables on abnormal days was lower than

that on normal days (p<0.01, Figure 1).

The main cluster was also formed in the SOM of

Case B (Figure 2). This cluster consisted of data from

35 normal sleep days. There were 16 abnormal sleep

days outside this cluster.

The median (interquartile range) of total nocturnal

sleep time, WASO, frequency of leaving the bed, and

frequency of awakenings in the bed on the abnormal

days of case B were 432.0 (379.3-515.0) minutes,

196.5 (156.0-258.5) minutes, 8.5 (6.0-11.8) times,

and 14.0 (7.8-24.8) times, respectively. The median

total nocturnal sleep time, WASO, and frequency of

leaving the bed on abnormal days were significantly

worse than those on normal days (p<0.05, p<0.01,

p<0.01, respectively). respectively, Figure 2).

Since several clusters were found in the abnormal

days group in case B, we performed a supplementary

analysis comparing the status of the four variables in

the three subgroups (B1, B2, and B3; Figure 3).

All three subgroups had longer WASO than the

normal group, but the days belonging to B1 were

characterized by the longest WASO, with relatively

more frequent leaving the bed and awakening in the

bed. The days belonging to B2 were characterized by

the most frequent leaving the bed and less frequent

awakening in the bed, and the days belonging to B3

were characterized by less frequent leaving the bed

and most frequent awakening in the bed.

Figure 1.

Objective Evaluation of Sleep Disturbances in Older Adults with Cognitive Impairment Using a Bed Sensor System and Self-Organizing

Map Analysis

255

Figure 2.

Figure 3.

4 DISCUSSIONS

This study suggests that it is possible to detect

abnormal sleep days in older adults with cognitive

impairment by measuring using a bed sensor system

and performing SOM analysis of the data. We

determined that it is possible to simultaneously detect

abnormalities in individuals with multiple patterns of

abnormalities, such as in case B in this study.

Furthermore, the abnormalities discriminated by

SOM analysis could be explained by comparing

several conventional sleep measures, suggesting the

validity of the SOM analysis results.

DATA 2024 - 13th International Conference on Data Science, Technology and Applications

256

Previous studies have shown that sleep

disturbances in older adults with dementia are

characterized by frequent leaving the bed (Higami

2018), prolonged or shortened total sleep time

(Higami 2018), decreased sleep efficiency (Cote

2021), and greater inter-daily circadian variability

(Cote 2021). Another study reported the

characteristics of sleep disturbances that vary

according to the cause of cognitive impairment

(Fukuda 2022), dementia severity (Blytt 2021), and

the presence or absence of complications such as pain

and depression (Blytt 2021). However, to the best of

our knowledge, this is the first study to detect

abnormal days based on intra-individual sleep

variability in an older adult with cognitive

impairment who had sleep problems and showed that

there are different types of abnormalities from day to

day.

In the future, we would like to investigate the

relationship between the disease and sleep

disturbances, such as whether the characteristics of

nocturnal sleep shown in case B of Alzheimer's

disease in this study indicate variability other than the

diurnal variability of the circadian rhythm, which is

one of the characteristics of this disease.

One limitation of this study was the small number

of individuals in whom abnormal sleep was detected.

Therefore, it is necessary to verify the reliability of

our findings by including a greater number of older

adults with cognitive impairments in subsequent

studies.

Another limitation is that the state of normal sleep

is not necessarily generally normal since the study

focused on the detection of abnormalities.

An important limitation of our measurement is

that it is unable to detect sleep disturbances in

individuals immediately after the start of data

collection because a certain amount of data

accumulation is required to detect abnormal sleep. To

address this issue, we are currently investigating the

possibility of extracting standard sleep patterns by

accumulating data from multiple cases, including

older adults without cognitive impairment, and

conducting an SOM analysis. If this standard pattern

can be extracted, it may lead to the early detection of

abnormalities in each individual by comparison with

the standard pattern.

5 CONCLUSIONS

The characteristics of abnormal sleep days identified

by SOM could be explained using these four variables

i.e. total nocturnal sleep time, wake time after sleep

onset, frequency of leaving the bed, and frequency of

awakening in the bed for each day suggesting the

effectiveness of identifying abnormal days by SOM.

Using SOM analysis, we also showed that there are

different abnormalities from day to day in older adults

with cognitive impairment.

REFERENCES

Blytt, K. M., Flo-Groeneboom, E., Erdal, A., Bjorvatn, B.,

& Husebo, B. S. (2021). Sleep and its association with

pain and depression in nursing home patients with

advanced dementia - across-sectional study. Frontiers

in psychology, 12, 633959.

Cote, A. C., Phelps, R. J., Kabiri, N. S., Bhangu, J. S., &

Thomas, K. K. (2021). Evaluation of wearable

technology in dementia: a systematic review and meta

analysis. Frontiers in medicine, 7, 501104.

Fukuda, C., Higami, Y., Shigenobu, K., Kanemoto, H., &

Yamakawa, M. (2022). Using a non-wearable

actigraphy in nursing care for dementia with Lewy

Bodies. American journal of Alzheimer's disease and

other dementias, 37, 15333175221082747.

Guarnieri, B., Adorni, F., Musicco, M., Appollonio, I.,

Bonanni, E., Caffarra, P., Caltagirone, C., Cerroni, G.,

Concari, L., Cosentino, F. I., Ferrara, S., Fermi, S.,

Ferri, R., Gelosa, G., Lombardi, G., Mazzei, D.,

Mearelli, S., Morrone, E., Murri, L., Nobili, F. M.,

Sorbi, S. (2012). Prevalence of sleep disturbances in

mild cognitive impairment and dementing disorders: a

multicenter Italian clinical cross-sectional study on 431

patients. Dementia and geriatric cognitive disorders,

33, 50–58.

Higami, Y., Yamakawa, M., Shigenobu, K., Kamide, K., &

Makimoto, K. (2019). High frequency of getting out of

bed in patients with Alzheimer's disease monitored by

non-wearable actigraphy. Geriatrics & gerontology

international, 19, 130–134.

Irwin, M. R., & Vitiello, M. V. (2019). Implications of sleep

disturbance and inflammation for Alzheimer's disease

dementia. The Lancet. Neurology, 18, 296–306.

Kogure, T., Shirakawa, S., Shimokawa, M., & Hosokawa,

Y. (2011). Automatic sleep/wake scoring from body

motion in bed: validation of a newly developed sensor

placed under a mattress. Journal of physiological

anthropology, 30, 103–109.

Nedergaard, M., & Goldman, S. A. (2020). Glymphatic

failure as a final common pathway to dementia. Science

(New York, N.Y.), 370, 50–56.

Okuda, S., Tetsuka, J., Takahashi, K., Toda, Y., Kubo, T.,

& Tokita, S. (2019). Association between sleep

disturbance in Alzheimer's disease patients and burden

on and health status of their caregivers. Journal of

neurology, 266, 1490–1500.

Shi, L., Chen, S. J., Ma, M. Y., Bao, Y. P., Han, Y., Wang,

Y. M., Shi, J., Vitiello, M. V., & Lu, L. (2018). Sleep

disturbances increase the risk of dementia: a systematic

review and meta-analysis. Sleep medicine reviews, 40,

4–16.

Objective Evaluation of Sleep Disturbances in Older Adults with Cognitive Impairment Using a Bed Sensor System and Self-Organizing

Map Analysis

257

Tago, M., Katsuki, N. E., Yaita, S., Nakatani, E.,

Yamashita, S., Oda, Y., & Yamashita, S. I. (2021). High

inter-rater reliability of Japanese bedriddenness ranks

and cognitive function scores: a hospital-based

prospective observational study. BMC geriatrics, 21,

168.

Webster, L., Powell, K., Costafreda, S. G., & Livingston,

G. (2020a). The impact of sleep disturbances on care

home residents with dementia: the SIESTA qualitative

study. International psychogeriatrics, 32, 839–847.

Webster, L., Costafreda Gonzalez, S., Stringer, A.,

Lineham, A., Budgett, J., Kyle, S., Barber, J., &

Livingston, G. (2020b). Measuring the prevalence of

sleep disturbances in people with dementia living in

care homes: a systematic review and meta-analysis.

Sleep, 43, zsz251.

Wilfling, D., Dichter, M. N., Trutschel, D., & Köpke, S.

(2019). Prevalence of Sleep Disturbances in German

Nursing Home Residents with Dementia: A Multicenter

Cross-Sectional Study. Journal of Alzheimer's disease :

JAD, 69, 227–236.

DATA 2024 - 13th International Conference on Data Science, Technology and Applications

258