Table 2.

Basic study characteristics and conclusions.

Study	Study design	Intervention	Participants	DTA scale	Engagement measure	Conclusions
Berry et al. (48)	Qualitative study	Actissist, a CBT informed app for people who have experienced a first episode of psychosis.	Stage 1: 9 Actissist users; Stage 2: 14 Actissist users and 10 mental health staff.	mARM	None	Developed mARM to measure DTA in the context of smartphone apps.
Goldberg et al. (49)	Study 1: cross sectional study and; Study 2: randomized controlled trail.	Study 1: Smartphone-based meditation apps in the market, such as Calm and Headspace. Study 2: Smartphone based Medication app—Healthy Minds Program (HMP).	Participants were in general population in both of the studies. N = 290 in study 1 and N = 314 in study 2.	DWAI	App Utilization. Study 1: self-report using frequency (daily, weekly, monthly, several times a year, or never); Study 2: Objective usage data gathered from the app.	DWAI correlates with frequency of app use (r = 0.42) in study 1, and correlates with HMP usage in study 2 (rs = 0.17–0.22). Early DWAI (week 1 and 2) didn't predict post treatment distress, but DWAI in weeks 3 and 4 associated with the clinical outcomes (βs= -.17 and -.13).
Clarke et al. (14)	Secondary Analysis of a Randomized Controlled Trial	Fully automated apps—myCompass	Participants were people with mild-to-moderate depression, anxiety, and/or stress symptoms. N = 90.	ARM	Number of program interactions (i.e., logins); number of modules completed; frequency of self-monitoring.	The scores of ARM did not correlate with clinical outcomes. TA subscales composite score was significantly positively correlated with engagement (r = 0.32–0.38).
Prochaska et al. (50)	Randomized controlled trial	CBT based Chatbot app (Woebot) with tracking and notification functions.	Participants were 8–65 years old and screened positive in substance misuse (scoring>1 on the CAGE-AID). N = 180	WAI short form revised	Usage data metrics: days used, in-app text messages, and completed modules.	Greater frequency of use (total numbers of in-app text) was weakly associated with a reduction in substance use occasions (r = 0.23).
Darcy et al. (51)	Cross-sectional, Retrospective Observational Study	CBT based Chatbot app (Woebot) with tracking and notification functions.	Participants were Woebot users in general population. N = 36,070	WAI short form revised	None	The mean of bond sub-score is 3.84 which is comparable to face-to-face therapy. Thus, there is a possibility that users can build bond with apps.
Hillier (52)	Qualitative study	All types of unguided technology based interventions, including fully automated apps.	Participants were people with variety of clinical issues, including depression, anxiety, and bipolar disorder. N = 13	None	None	Participants generally rejects the ideas of having bonds or relationships with technology based interventions.