Abstract for: Modeling the I-Corps Program Theory: A System Dynamics Approach to Understanding Translational Scientist Development

The National Science Foundation's Innovation Corps (I-Corps) trains researchers to evaluate commercial potential through intensive customer discovery. While evaluations demonstrate impact, the mechanisms driving capability development remain poorly understood. We developed a system dynamics model explicating program theory for Case Western Reserve University I-Corps@NCATS, identifying feedback loop structures that drive translational scientist capability development during and after this entrepreneurial training program for biomedical researchers and clinicians. Using Stella Architect, we constructed a conceptual model representing eight capability stocks linked through three primary feedback loops. The model simulates a single innovation team over 30 months: 6-month pre-program period (concurrent CWRU Translational Fellows participation), 2-month I-Corps intervention, and 22-month post-program period. Four intervention points represent pedagogical mechanisms. Model structure draws from Kolb's experiential learning framework and empirical I-Corps literature. The model demonstrates three feedback mechanisms: R1 (Customer Discovery Engine), R2 (Capability Development Engine), and B1 (Assumption Correction). During pre-program, capabilities show modest baseline development (translational skills: 0.40→0.44). I-Corps intervention produces clear acceleration (skills: 0.44→0.55). Post-intervention, reinforcing loops drive sustained growth, with skills reaching 0.75, networks 0.40, and humility 0.50 by month 24, demonstrating brief interventions produce lasting effects through feedback dynamics. The model reveals I-Corps works through coordinated activation of three feedback loops, not individual skill acquisition. Long-term sustainability derives from self-reinforcing dynamics, explaining why brief interventions produce sustained effects. These findings suggest program improvements or adaptations should focus on strengthening feedback loop activation rather than lengthening program duration. AI used by corresponding author to translate Stella Architect equations into narrative passages.