| Startup failure rates exceed 90% within five years of their establishment, with founder characteristics playing a critical role alongside financial and market factors. Traditional assessment methods rely on self-reported surveys prone to social desirability bias and lack real-time behavioral validation. This paper presents a novel AI framework that combines natural language processing and machine learning models with game-based personality assessment to predict startup sustainability. The system integrates OCEAN Five personality profiles with textual pitch data and operational metrics using transformer-based NLP and ensemble learning models. We validate our approach on 44 startups, demonstrating that behavioral game data improves prediction accuracy by 18% compared to survey-based methods alone. Our multi-modal framework achieves 87% F1-score in sustainability classification. This work bridges entrepreneurship research, serious games, and AI-driven decision support, offering a proof of concept tool for investors, accelerators, and policymakers to reduce investment risk and support founder development. |
*** Title, author list and abstract as submitted during Camera-Ready version delivery. Small changes that may have occurred during processing by Springer may not appear in this window.