The effectiveness of artificial agents on loneliness reduction in older adults: a systematic review and meta-analysis

This meta-analysis reviewed studies comparing loneliness of older adults receiving artificial agent interventions with those not receiving them, to evaluate this treatment's efficacy in alleviating loneliness and other mental health problems among older people; and assess factors that moderate intervention effectiveness. Eligible studies were identified by a systematic search of PubMed, the Cochrane Central Register of Controlled Trials, EMBASE, CINAHL (EBSCO), PsycINFO (EBSCO), CNKI, Wanfang, Weipu, SinoMed, and ClinicalTrials.gov. Standardized mean differences with 95 % confidence intervals were extracted and pooled. Meta-regression and subgroup analysis determined the influence of trial characteristics. A total of 16 studies with 781 participants were included. Artificial agent treatment may significantly reduce loneliness in older adults, with acceptable heterogeneity (SMD= -0.69 [95 % CI, -0.89 to -0.40]; p=.000; I²=64.51 %). Artificial agents reduced loneliness more markedly for participants dwelling in non-community-based settings (p=.016). Social robots reduced loneliness more effectively than app/web-based agents (p=.033). Animal-shaped artificial agents had a significantly larger effect size than humanoid agents on loneliness reduction (p=.016). Longer session duration (p=.005), fewer weekly sessions (p=.000), and shorter session times (p=.005) produced greater loneliness reduction. Artificial agents had a significant effect on depression (SMD=-0.42, [95 % CI -0.72 to -0.12], P=.01), but no significant effects on quality of life or happiness (SMD=0.24, [95 % CI -1.0 to 0.58], P=.16 and SMD=0.63, [95 % CI -1.09 to 2.35], P=.47, respectively). Findings suggested artificial agents may alleviate loneliness in older adults. Treatment effectiveness may be moderated by participants' settings, session length, treatment duration, sessions per week, agent delivery platform, and agent type.