• Glimpse captures raw behavioral data (reaction times, decision patterns).
  • It extracts ~250 nuanced features that reflect subtle cognitive behaviors.
  • Glimpse trains predictive models using a combination of proprietary machine learning algorithms and established techniques such as PCA, Elastic-Net, Support-Vector-Machine, and Gaussian-Process-Classifier.
  • Models are validated using cross-validation on unseen data, ensuring generalizability.

This approach skips manual summarization into predefined traits, allowing better predictive accuracy by incorporating difficult-to-explain but powerful behavioral signals.