1 Introduction

Given the rapid progression into a super-aging society in Japan, one of the typical problems in a 24-h society is nursing care for older adults during the night. Specifically, caregivers suffer from the burden of providing long-term care because older adults need periodic care for the night. They have to monitor and check older adults’ sleeping situation periodically, give suggestions or advice when older adults have difficulty sleeping, and control the room environment, such as illumination and air conditioning. Consequently, caregivers themselves experience sleep problems, which may negatively affect their attitude to older adults. Such a situation seems to drive a vicious circle, and it is obviously worse for both caregivers and care recipient.

A lot of caregivers are middle aged, and they often have to quit their job to care for older adults, as providing care and working are hardly compatible. According to an employment status survey in 2017, about 100,000 workers annually leave their jobs, most of whom are aged 55–59 years. Half of them want to keep working even while being caregivers. However, in the present situation, it is difficult for them to provide care for older adults and invest in their regular work [1].

The population of older adults is estimated to keep increasing more and more in the future. Achieving effective and high-quality care support for older adults while reducing the burden for caregivers are pressing concerns in Japan. Special care is due not only for care recipients but also for caregivers; the care stress and irregular daily rhythm experienced by caregivers drain them physically and mentally. Sleep problems among care recipients and caregivers are clearly serious. Caregivers cannot sleep well, because they have to care for older adults all day. Sleep deprivation increases risk of depression, dementia, lifestyle-related diseases, and obesity. In other words, the lack of sleep can cause diseases in many people. Thus, caregivers are at risk of contracting diseases and be new care receivers themselves. Accordingly, the care problem in Japan can be expected to become serious in the future. However, studies on sleep care support system has not emphasized the needs of both care recipients and caregivers in home care.

In this research, we propose a caregiver support system for nursing care of older adults. This system supports the sleep not only care recipients but also caregivers.

2 Caregiver Support System

To reduce caregivers’ burden of the long-term care, we propose a caregiver system that automatically executes a caregiver’s functionality. It aims not to replace caregivers but to support them by reducing the number of their periodic tasks. Caregivers have conversation skills, know-how, experience, and knowledge of nursing care, which has been cultivated over many years. The system should, therefore, achieve functionality by substituting caregivers’ performance through data acquisition and learning mechanisms. Moreover, to foster reliable relationships, mutual understanding, and acceptance among older adults, their family members, and caregivers, it is indispensable for the system to provide them with an information-sharing mechanism on the care recipient’s sleep data. Thus, the goal of this research is to improve and resolve the serious sleep problems of both caregivers and older adults.

Figure 1 shows the system configuration of the proposed system. It is composed of a mattress sensor to measure a person’s biometric data related to their sleep state, Google Home to that provides suggestions and/or advice, Nature Remo to control home appliances linked with Google Home, and a PC to control these devices and execute the system’s functionality.

Fig. 1.
figure 1

System configuration.

Full size image

For the first step of the research, we develop and compare with the deference of effect to the participants in physically and mentally between synthesized voice and human voice of a person who is the participant’s relative through Google home.

3 Related Works

Japan is ranked tenth in the world in the number of people suffering from mental diseases. It is also ranked first in number of beds in mental hospitals. The percentage of people who have sleep disorders in worldwide ranges from 9% to 15%; that in Japan is 20%. At present, the economic loss attributed to sleep loan is JPY 35trillion/year in Japan. Further, sleep loan causes depression, obesity, dementia, and other disorders. On average, it takes three weeks (12 h/day) for people to return the sleep loan [1].

Sleep disorders increase the risk of lifestyle-related diseases and depression. This problem has been observed in Japan owing to changes in lifestyle. Meanwhile, the quality and duration of sleep vary greatly with age and are determined by numerous other factors. Previous studies have shown that sleep disorders commonly occur in older adults.

Nishino and his team emphasized how humans spend a significant part of our lives sleeping, which is essential for our physical and psychological well-being. However, sleep can be easily impaired by psychological and physical disorders [2, 3]. Shimamo to and his team suggested that a decline in the quality and total duration of sleep decreases physical activity levels and increases daytime sleepiness as well as the risk of lifestyle-related diseases and depression [4].

Takadama and his team focused on this problem and proposed a concierge-based care support system to provide a comfortable and healthy life for older adults. The system estimated a user’s daily sleep stage and stores such personal data as big data, thereby enabling care workers and doctors to design personal care plans for specific users more effectively [5,6,7].

Takahara et al. proposed an indirect biofeedback mechanism that helps patients keep track of their sleep quality and condition by monitoring a device that displays a virtual plant. They also proposed a mechanism through which the patient, family members, and medical staff can share indirect biofeedback information. An experiment was conducted in a senior care home using five elderly people and two healthy people as subjects, with family members and medical staff participating in the experiment. The experiment attempted to clarify the usefulness of indirect biofeedback in the improvement of a patient’s sleep. They also aimed to confirm that patients, their family members, and medical staff could deepen their mutual understanding and mutual acceptance by sharing indirect biofeedback information. Consequently, they may be able to judge whether indirect biofeedback through the virtual plant is useful for improving patients’ sleep condition [8].

4 Proposed Method

To develop the proposed system, we addressed the following research issues:

  1. A.

    Learning mechanism of an individual’s sleep state

  2. B.

    Conversation control to enable a care recipient to undertake behavior modifications leading to sleep improvement

  3. C.

    Information-sharing mechanism to foster mutual understanding and acceptance between the care recipient, caregivers, and related people

4.1 Learning Mechanism of Sleep State

The proposed system can acquire a care recipient’s sleep data, including “sleep score,” “heart rate,” “body motion,” “sleep time,” and “nocturnal awakening,” through a sleep mattress sensor developed by TANITA. In this study, we incorporate a learning mechanism to extract the regularity of sleep quality from the relationship between the care recipient’s physical condition/age/disability/daily activity and the above sleep data using genetic programming (GP).

GP is a systematic method for prompting computers to automatically solve a problem. GP starts from a high-level statement of what needs to be done and creates a computer program to solve the problem without requiring the user to know, specify, or structure the solution in advance [9].

In this research, the characteristics of sleep are notably expressed in an introductory sleep phase for 90 min. We focus on and adopted the introductory sleep phase as an indication of sleep quality evaluation.

4.2 Conversation Control enabling Behavior Modifications Leading to Sleep Improvement

Figure 2 shows the proposed system.

Fig. 2.
figure 2

Structure of the proposed system.

Full size image

In this research, we aim to solve issues in generating suggestions and/or advice based on the evaluation of a care recipient’s sleep state, controlling the conversation with a care recipient in the proposed system, and controlling the room environment, such as illumination and air conditioning. Moreover, utilizing a bedside Google Home is important given that older adults mainly use voice communication.

Based on veteran caregivers’ conversation skills and experiences, several basic pieces of advice on such topics as optimal wake-up time, bedtime, and bath time, as well as typical conversation templates, are preset. The lights in the bedroom are controlled depending on the optimal waking and sleep time.

  • Tools in this research

In this research, we use the following tools for nursing care of older adults.

  • Nature Remo

Nature Remo is a smart remote controller that can control every home electronic appliance though Wi-Fi. This can be accessed by Google Home or a mobile phone.

  • Google Home

Google Home is a smart speaker. Nature Remo can control every home electronic appliance by voice via Google Home. In the experiment, when the participant asks Google Home about his/her sleep information, Google Home provides feedback of the previous day’s sleep information, suggests their ideal wake-up time/bed time, and turns on/off the lights automatically.

  • TANITA SL-511

TANITA SL-511 is a sleep mattress sensor that has built-in high-precision body motion detection sensor. It detects the user’s body motion, breathing, and heartbeat from under the bed in real time. The system sends the data to a server.

We employ Sleep Score, which is calculated by TANITA algorithm based on data from the mattress sensor. We also develop a learning mechanism based on Sleep Score and other data; feedback of the data is provided as information sharing to care recipients and their supporters via a web application.

4.3 Information-Sharing Mechanism

Information sharing between people is useful for mutual awareness in general, as shown in Fig. 3.

Fig. 3.
figure 3

Visualization of mutual awareness.

Full size image

For example, a patient’s friend comes to see the patient; a staff member learns of this fact; and the patient also knows that the staff member is aware of the fact. Mutual awareness denotes that parties know that they share given information with one another.

However, sometimes medical staff might be in strong position, whereas patients might be in a weak one, in the sense that the patient depends on the medical care given by the staff. Meanwhile, the patient pays for the medical care and services given by the staff. Information that should be shared by others, such as medical staff and family members, is the patient’s personal information.

Information sharing between the patient and medical staff should be carefully designed considering the points mentioned above. It might be quite significant for a patient to be aware of being understood and accepted by others through information sharing [8]. Meanwhile, a patient’s extremely personal information that cannot be usually seen and known by others may need to be protected. In general, the patient does not want others to know his/her extremely personal information. In addition, direct numerical feedback, displayed as drastic numerical changes, might be perceived as unfamiliar data and give users a negative feeling.

As such, we consider the significance of indirect representation, that is, indirect biofeedback. Information represent as indirect biofeedback and shared by others is the patient’s personal information, but it is not too specific or too comprehensive, enabling acceptability on the part of the patient that the personal information is seen and known by others.

Figure 4 shows the model of mutual acceptance that we aimed to create between care recipients and others.

Fig. 4.
figure 4

Visualization of mutual acceptance.

Full size image

Once the personal information that is too specific or too comprehensive to the patient could be shared, the resulting situation should be expected to be beyond mere mutual awareness, i.e., mutual acceptance [8].

Figure 5 shows a model of information sharing to foster mutual understanding and acceptance not only between care recipients and caregivers but also between them and related people, such as the recipient’s family. They can monitor the change in the care recipient’s sleep state and then confirm sleep improvement.

Fig. 5.
figure 5

Visualization of information sharing

Full size image

The sleep quality of a care recipient is deeply related to his/her relationship with the caregiver. To improve a care recipient’s sleep quality, it is necessary for a care recipient and a caregiver to improve their relationship. Thus, in this research, we facilitate the sharing of information of both care recipient and caregiver using a web application. The system shares their information to foster mutual understanding and acceptance.

5 System Structure

The Table 1 shows the contents of speech through Google home.

Table 1. The contents of speech.
Full size table

The contents of synthesized voice speeches are changed from person’s speeches a bit with asterisks because of a feeling of strangeness with the speeches by synthesized voice. Fig. 6 shows the system image of this experiment.

Fig. 6.
figure 6

The system image.

Full size image

The PC and Google home are connected via the same Wi-Fi. We construct a server and upload the MP3 files of both synthesized and person’s voices.

6 Experiment

In order to investigate what and how influence two types of voices through Google home result in to the participant’s sleep quality, heart rate and mental situation, we conduct an experiment with 2 participants (1 male, 1 female) in 10 days, each experiment is 5 days after the pre-experiment about one month.

  • The participants

    • Female (69 years old), male (72 years old)

    • They are a married couple and they are living in their home.

  • Experimental terms

    • 10 days (Synthesized voice experiment: 5days, person’svoice: 5days)

    • The both first days of the experiments are same day of the week.

    • First week: The speeches by synthesized voice,

    • Second week: The speeches by person’s voice

  • The rules

    The system plays the voice through Google home 3 times a day in their home (mainly at their living room). The contents of the voice are different very day every time.

  • Evaluations

    • Subjective evaluation

      Questionnaires before and after the experiments

  • Objective evaluation

    Sleep qualities and Heart rates of the participants

7 Conclusion

We have proposed a caregiver support system for nursing care of older adults to decrease the burden of caring for caregivers. This system helps care recipients in the following two aspects. First, the system periodically checks the care recipient’s sleeping situation, gives suggestions or advice when they have difficulty sleeping, and controls the room environment, including illumination and air conditioning. Second, the system helps both care recipients and caregivers to be aware of their sleep quality and condition, by monitoring a device that is displayed on a web application. Thus, they can deepen mutual acceptance and understanding.

In future work, we will develop the proposed system. Specifically, the development will focus on the following three research issues: A. Learning mechanism of an individual’s sleep state, B. Conversation control to enable a care recipient to undertake behavior modifications leading to sleep improvement, and C. Information-sharing mechanism to foster mutual understanding and acceptance between care recipients, caregivers, and related people.

Further, we will conduct preliminary experiments at a normal home. Afterward, we will perform field experiments with older adults in their own home, with their permission.