Conceptualizing Student Problem Dynamics Using Idiographic and Data-Driven Networks: A Feasibility Trial

Abstract Purpose Nomothetic models of mood disorders often neglect individual heterogeneity. Idiographic network models could offer an alternative, but it is unclear whether subjective prior networks, data-driven networks, or networks that integrate both subjective information and empirical data should inform psychotherapy. Methods We investigated the face validity of different networks, the utility, feasibility, and non-maleficence of implementing an ecological momentary assessment (EMA) protocol for network-informed treatment among 63 clinical psychology students. During the subjective prior network conceptualization , students explored personal issues, for which they generated EMA items and estimated the strength of the relationships between items. This information was later used to construct subjective prior networks serving as a prior for the PREMISE networks that integrate priors and longitudinal EMA data using Bayesian inference. In the EMA period , these items were prompted on their smartphones (7x/day for 15 days) to estimate data-driven networks. Finally, during the network feedback , trained therapists presented five networks ( temporal–unregularized, contemporaneous–unregularized , subjective prior , contemporaneous–Bayesian , contemporaneous–PREMISE) and subjects assessed their face validity. EMA response rates served as a feasibility-proxy and the Depression-Anxiety-Stress-Scale before and after EMA assessed mood non-maleficence. Results The temporal–unregularized network had a significantly worse face validity than other networks ( contemporaneous–unregularized , subjective prior , contemporaneous–PREMISE ). The subjective prior network descriptively showed the highest face validity. Its conceptualization was also perceived significantly more useful than the EMA period and the network feedback. The subjects’ response rates were lower compared to previous studies. EMA did not affect mood significantly. Conclusions The subjective prior network was perceived as descriptively most valid and its conceptualization most useful, suggesting it may serve as a subjectively valid fallback when data is limited. Due to its similar face validity and non-maleficence, the PREMISE method should be investigated in future clinical studies using EMA protocols with less prompts per day, but more assessment days. Lesson learned Thorough subjective prior network conceptualizations could enhance adherence to EMA and to network-informed therapy interventions.