Table 1.

Summary of research design: data sources, analysis, and outcomes of each phase.

Aim	Data collection procedures	Data analysis procedures	Product or outcome
(1) Mediating effects nonverbal communication	Quantitative: MPathic game scores, Kinect nonverbal data, OSCEa communication score, RIASb and FACSc coding of video Qualitative: video recording of MPathic-VR interactions, student reflections	Dyadic data analysis, path modeling Thematic analysis of video data, followed by data transformation to quantify it	Understand what elements of the nonverbal behavior construct are most important in the human-computer interaction
(2) Develop model of nonverbal communication	Interviews with oncology providers about intentional nonverbal behaviors (eg, mirroring, silence, and nodding)	Grounded theory qualitative analysis: open coding and relating components to develop the model	Model of nonverbal communication; function to include in virtual human
(3) Add and validate a new automated nonverbal communication behavior assessment in MPathic-VR	Qualitative findings inform development of assessment along with FACS and RIAS models; Providers use MPathic-VR and assess if the system detects nonverbal behavior	Mixed methods to build assessment: codes → variables, quotes → item language, interrater reliability analysis; code interactions, dyadic data analysis	Updated virtual human with new automated nonverbal communication behavior assessment; initial quality control data

^aOSCE: objective structured clinical examination.

^bRIAS: Roter Interaction Analysis System.

^cFACS: Facial Action Coding System.