There is growing evidence that social isolation is implicated in an increased risk of poor health, all-cause mortality, depressive symptoms, cardiovascular disease, and cognitive decline (1–3). In recent years, social network sizes and contact with communities, neighborhoods, and families have decreased (4, 5). In particular, older adults frequently experience social isolation due to declining physical mobility, cognitive function, and economic resources or roles in society, as well as the death of their spouses. Globally, more than 25% of adults aged 65 and older experience social isolation (6), making it a significant public health concern in aging societies (7). Social isolation is a major modifiable risk factor for dementia, and social activity can increase cognitive reserve, improve motivation for healthy behaviors, and reduce stress and inflammation (8–10). Therefore, early identification and management of older adults at risk of social isolation is essential for public health. Two distinct terms refer to subjective experience and objective isolation from other social actors: loneliness refers to the perceived experience of being isolated, while social isolation denotes the actual absence of social contact. This study prefers the term “social isolation” over “loneliness.” Social isolation is characterized as a scarcity of social contact and communication with a broader community, neighborhood, and relatives. It is measured by the frequency of social engagement or social contact and social network size, including marital status or family structure, via questionnaires (4, 5, 11). However, the usefulness of subjective self-report questionnaires is limited with respect to consistency and reliability due to recall bias or misclassification among older adults (12, 13). Moreover, the social stigma of social isolation reduces the likelihood of reporting social isolation, resulting in decreased identification of isolated individuals. Developing an objective method to assess social isolation would help address it among older adults. Wristband sensors are cost-effective and non-invasive tools for objectively and continuously measuring lifestyle factors without recall bias in the community (14–17). We have developed a wristband sensor for quantifying conversation time and used it in a prospective cohort study on the association of lifestyle factors with cognitive function in older adults living in the community (16). The random forest regression analysis results show that objective conversation time is non-linearly associated with the global cognitive function. These results lead us to hypothesize that decreased conversation time may be related to cognitive impairment through social isolation in older adults. Several studies have examined the association of objectively measured conversation time using smartphones with loneliness in young adults or patients with schizophrenia spectrum disorders (18–20). To the best of our knowledge, few studies have measured conversation time using wearable sensors in older adults living in the community. Therefore, this study aims to determine whether objective conversation time is cross-sectionally associated with social behavior.

We developed a novel wearable sensor that enables the quantification of conversation time. To the best of our knowledge, this is the first study to examine the association between objectively measured conversation time and self-reported social behavior in older adults living in the community. This study provides valuable insight into the relationship between daily living activities and social isolation. Our results show that decreased conversation time is associated with less frequency of engagement in community activities and lessons or classes. We suggest that objective conversation time may be a valuable marker of social isolation. Moreover, the main advantages of this study include the objective measurement of conversations and the large-scale inclusion of community-dwelling older adults.

A key finding of this study was that objective conversation time is positively associated with the frequency of engagement in community activities and lessons or classes. Furthermore, this association remained significant even after controlling for age, sex, and years of education. However, no significant correlation was found between conversation time and the frequency of outings or contact with friends or relatives. These results suggest that conversation time is associated with engagement in community activities or classes rather than simple outings or social contact. Social isolation is defined as a lack of social contact with family and friends or involvement in social, religious, or other outdoor activities beyond social contact. Social isolation is generally assessed by self-report questionnaires regarding social network size, frequency of social contacts, and engagement in social activities (4, 5, 11). However, there is little consensus on these assessment methods. In this study, we used self-report questionnaires to assess social behavior, such as the frequency of outings, contact, and cognitive or social activity. The increasing number of older adults who live alone and are housebound has led to a higher proportion of them experiencing social isolation. Therefore, quantifying social isolation is essential better to understand the relationship between social isolation and health. Additionally, developing simple methods to quantify the degree of social activity is an important yet challenging task for researchers. Since wearable sensors can remotely collect data on daily life and behavior in real time, they may be used for early detection of social issues, including social isolation (21, 22). Several studies have assessed loneliness or social isolation by analyzing data on household behavior, physical activity, or sleep from smartphones or wearable sensors (23–28). These results show that eating behavior, mobility, and sleep quality are associated with social isolation (23, 24). On the other hand, the association of objectively measured physical activity with social isolation showed inconsistent findings (25–27). Moreover, making and receiving phone calls, social engagement on the web, computer usage, social interaction, smartphone usage, physical activity, or sleep duration are associated with loneliness (28). Therefore, objectively measuring various lifestyles using sensing techniques may be a surrogate parameter for social isolation. Conversation is an essential lifestyle factor for older adults to develop social relationships and preserve brain function, mental health, and quality of life (29, 30). Only three studies have measured conversation time using smartphone data collected in the ambient environment as a conversation and examined the association between conversation time and loneliness (18–20). Two studies found no significant association between conversation time and loneliness in young adults or patients with schizophrenia spectrum disorders (18, 19). In contrast, the remaining research showed a substantial association of conversation time with depression in students (20). Our results showed a significant association of objective conversation time with social behavior. We suggest that objectively measured conversation time using a wearable sensor could be a simple and non-invasive method to assess the social behavior of older adults living in the community. Social isolation is associated with an increased risk of poor health and leads to increased personal and societal costs by increasing the demands for healthcare resources, including the management of acute, chronic, and consumptive illnesses, the use of emergency medical care, and institutionalization (31–33). Therefore, early detection and management of social isolation are crucial for the prevention of various diseases, including dementia, and decreased need for hospital visits, hospitalization, and institutionalization. Social implementation of conversation time measuring technology has the potential to facilitate screening for social isolation among community-dwelling older adults. Additionally, this tool may reduce the demands for human resources, resulting in decreased time or cost burden on individuals, medical staff, or researchers in research or clinical settings.

Several limitations should be noted in this study. First, since the aim of this study is to clarify the association of lifestyle factors with cognitive function, established self-report questionnaires of social isolation, such as the Lubben Social Network Scale (34) and Social Disconnectedness Scale (35), were not included in this study design. Second, we did not consider the effect of family structure on conversation time. Nonetheless, there was no significant association between conversation time and superficial contact with friends or relatives. Third, the accuracy of conversation time detection was verified by comparing the wristband sensor data with video observational data, but television viewing or radio-listening sounds could have been detected as conversation. Fourth, it is important to collect only conversations with good reciprocity and alternation between social actors. However, our sensor technology cannot distinguish between one-way and reciprocal conversations. In this study, we considered that participation in both one-way or reciprocal conversation was necessary for social activity. Further technical developments are required to identify only reciprocal conversations. Fifth, the number of valid data form self-report questionnaire was different for each item due to missing data.

To our knowledge, this is the first study to report the association between objective conversation time and the frequency of engagement in community activities and lessons or classes. Objectively measured conversation time using a wearable sensor could be a valuable parameter for determining social isolation in large cohort studies.