Association between social asymmetry and depression in older adults: A phone Call Detail Records analysis

Abstract

Analyzing social interactions on a passive and non-invasive way through the use of phone call detail records (CDRs) is now recognized as a promising approach in health monitoring. However, deeper investigations are required to confirm its relevance in social interaction modeling. Particularly, no clear consensus exists in the use of the direction parameter characterizing the directed nature of interactions in CDRs. In the present work, we specifically investigate, in a 26-older-adults population over 12 months, whether and how this parameter could be used in CDRs analysis. We then evaluate its added-value for depression assessment regarding the Geriatric Depression Scale score assessed within our population during the study. The results show the existence of three clusters of phone call activity named (1) proactive, (2) interactive, and (3) reactive. Then, we introduce the notion of asymmetry that synthesizes these activities. We find significant correlations between asymmetry and the depressive state assessed in the older individual. Particularly, (1) reactive users are more depressed than the others, and (2) not depressed older adults tend to be proactive. Taken together, the present findings suggest the phone's potential to be used as a social sensor containing relevant health-related insights when the direction parameter is considered.

Figure 1

Phone call activity behaviors in older population regarding the monthly number of phone calls and their direction. Bold lines represent outgoing phone calls whereas thin lines represent incoming ones. For each individual, the total average of his monthly number of phone calls is represented by a black dot, whereas the corresponding standard deviation is represented by an error bar. Here, we observe three distinct clusters in older adults: (1) those who phone more than they respond to their incoming phone calls (proactive users), (2) those who phone as much as they respond to their incoming phone calls (interactive users), and (3) those who phone less than they respond to their incoming phone calls (reactive users). Such behaviors emphasize the fact that, behind the phone ubiquitous status, different phone call activity use-cases exist in older adults.

Figure 2

Bland-Altman plot for outgoing phone calls and incoming phone calls. Individuals are represented by dots whose colors are assigned according to the individual’s general phone call activity behavior. On this figure, two points stand out: (1) the general bias of difference between the monthly number of outgoing phone calls and the monthly number of outgoing phone calls is low for the overall population, but (2) when considering each clusters of phone call activity, we observe that only interactive users seem to have such a low bias, whereas there is a sharp contrast between Proactive users and Reactive phone users. In short, there is a low agreement between outgoing phone calls and incoming phone calls depending on the considered cluster of phone call activity. Taken together, these observations emphasize the fact that different behaviors exist in phone call activity in older adults. In particular, depending on these behaviors, there is no evidence in neglecting the phone call activity direction when leading statistical analysis.

Figure 3

Asymmetry in phone call activity behaviors in older population regarding the outgoing and incoming monthly number of phone calls. For each cluster of phone call activity, the asymmetry coefficient and the information of direction values are calculated for each individual at each month. Their association is then represented by a dot displayed onto the corresponding panel. Here, the dot’s color is assigned according to the individual’s general phone call activity behavior. On this figure, we can observe that while proactive and reactive users are characterized by a wide range of, respectively, positive and negative values, interactive users are characterized by both values oscillating around zeros, but also by a few highly positive and negative ones. Thus, taken together, these observations highlight the existence of variations between proactive, reactive, and interactive users, as well as the asymmetry indicator’ ability in catching their characteristics.

Figure 4

Geriatric Depression Scale values’ distribution according to asymmetry coefficient values. Here, dots correspond to the association between the individual’s GDS value and the individual’s asymmetry coefficient value calculated from CDRs over the 4-week period before the date at which the older adult passed the GDS test. Their colors are assigned according to the individual’s general phone call activity behavior. On this figure, we observe that individuals with a negative asymmetry coefficient tend to obtain a high GDS score corresponding mostly to mild depression. In contrast, individuals obtaining a low GDS score tend to have a positive asymmetry coefficient.

Figure 5

Geriatric Depression Scale values’ distribution according to the skewness coefficient values. Dots correspond to the association between the individual’s GDS value and the individual’s skewness coefficient value calculated from CDRs over the 4-week period before the date at which the older adult passed the GDS test. Their colors are assigned according to the individual’s general phone call activity behavior. On this figure, we observe that, similarly to the asymmetry coefficient, individuals with a negative value of their skewness coefficient tend to obtain a high GDS score corresponding mostly to mild depression. In contrast, individuals obtaining a low GDS score tend to have a positive value of information of direction.