Table 1.

Main evidence from systematic review.

First author, year	Study design	Sample size and characteristics	Robot	Mental health or well-being outcome	Intervention	Main findings
Bemelmans, 2015 [41]	Pre-post (single session)	71 nursing home residents with dementia (age range not reported)	Paro	IPPAa score and mood via COOP/WONCAb chart recorded by care provider	Quasi-experimental time series study: 15-minute interaction with Paro while experiencing unrest or negative mood	Significant positive effect on mood and IPPA score
Galvão Gomes da Silva, 2018 [42]	Pre-post (single session)	20 psychology students (aged 18 to 62 years, majority under 25 years)	NAO	24 open-ended self-reported items in questionnaire assessing motivation for exercise (author created)	Two lab sessions of motivational interviewing for exercise with NAO (1-week interval between sessions)	Positive appraisals of robot as nonjudgmental Increased “change talk” in participants Increased motivation to exercise
Kurashige, 2017 [43]	Pre-post (single session)	12 male students aged 21 to 23 years (mean age not reported)	CRECAc	Author-created self-report items (15) on conversational flow, perceived trust/reliability in CRECA, and stress reduction	Motivational interviewing session with nodding or not nodding CRECA around stress management	Positive appraisal of dialogue with nodding CRECA Self-reported reduction in anxiety
Lane, 2016 [44]	Pre-post	106 VAd community living center elderly patients (aged 58 to 97 years, mean age 80 years)	Paro	Care staff observed behaviors and mood on researcher-created tracking sheet across 3 time periods (baseline, Paro treatment, posttreatment)	Veteran was actively presented with or observed to be actively using Paro for a minimum of 5 minutes	Observed that Paro reduced negative behavior and mood states Observed that Paro induced increases in indicators of positive mood states
Loi, 2017 [45]	Pre-post	45-bed unit for younger adults with neuropsychiatric conditions, (residents < 65 years, mean age not reported)	Betty	Staff completed a pre- and post-SARse questionnaire regarding patient well-being, enjoyment, and quality of life (items based on technology acceptance model)	Betty was present at the facility for 12 weeks; engaged with residents via conversations, music, relaxation exercises, and games	Staff reported that Betty was helpful to patients by being comforting, relaxing, and improving the well-being of residents
Moyle, 2018 [46]	Cluster randomized RCTf	Residents with dementia in a long-term care facility (mean age not reported)	Paro	Staff reported patient comfort and well-being (qualitative interview)	Three 15 minute interactions were observed between Paro and elderly residents within 3 treatment groups: Paro, plush toy, or usual care	Staff indicated there were benefits to using Paro as a companion to elderly patients, although Paro did not comfort all residents
Šabanović, 2013 [47]	Pre-post	10 nursing home residents with dementia (ages not reported)	Paro	Researcher videotaped and coded interactions based on positive engagement with others	Residents interacted with Paro over 7 weekly sessions	Observed an increase in prosocial interaction between residents
Sefidgar, 2016 [48]	Pre-post (single session, within-subject design)	38 women aged 19 to 45 years, mean age 23.8 years	Haptic Creature	Self-reports on the SAMg; STAI-6h	Interaction with Haptic Creature on lap, compared with nonmoving stuffed animal replica	Biometric indicators of relaxation related to heart and respiration rates significantly decreased relative to stroking a nonbreathing replica Participants reported feeling calmer and happier
Valentí Soler, 2015 [49]	Pre-post	211 nursing home patients with dementia, 37 at day care facility (total n=248; age range 58 to 100 years, mean age 84.7 years	Paro, NAO	Staff reported on the Apathy Inventory and QUALIDi scale	Comparing interactions with Paro, NAO, and live dog over 3 months	Apathy and irritability improved for NAO and Paro groups Quality of life improved for Paro group
Wada, 2010 [50]	Pre-post	2 elderly individuals and 1 caregiver, age not reported	Paro	Researcher observed emotional responses and behaviors (ie, smiling)	Caregivers engaged in a manual-assisted 30-minute interaction between residents and Paro (4 sessions)	Observed-recorded increase in positive behaviors in 1 participant (smiling, laughing), no significant change in other participant
Wada, 2012 [51]	Pre-post	12 elderly participants (mean age 86.8 years) and 9 caregivers (mean age 28.1 years)	Paro	Observation sheet recording participant behaviors and emotional reactions (researcher-recorded)	Manual-assisted interaction with Paro; observed before caregiver used manual and after caregiver used manual	The manualized Paro interaction increased contentment and positive social interactions
Wada, 2014 [52]	Pre-post	64 elderly individuals in 7 elder-care facilities (mean age 86.5 years)	Paro	Observation sheet recording perceived participant behaviors and mood (anxiety, depression, aggression)	Manual-assisted interaction with Paro over 5 months	Following Paro interactions, caregivers observed decreases in perceived anxiety, depression, or aggression in 25 residents (39%)

^aIPPA: Individually Prioritized Problems Assessment.

^bCOOP/WONCA: Primary Care Cooperative Information Project/World Organization of Colleges, Academies, and Academic Associations of General Practitioners/Family Physicians.

^cCRECA: Contextual Respectful Counseling Agent.

^dVA: US Department of Veterans Affairs.

^eSAR: socially assistive robot.

^fRCT: randomized controlled trial.

^gSAM: Self-Assessment Manikin.

^hSTAI-6: State-Trait Anxiety Inventory.

ⁱQUALID: Quality of Life in Late Stage Dementia.