We develop rigorous, evidence-based AI systems at the boundary of psychophysiology, signal processing, and organizational psychology. Our work is grounded in large-scale empirical research and guided by ethical principles from design to deployment.
| Title | USPTO | SIC Colombia | Field |
|---|---|---|---|
|
Non-Invasive Remote System and Method to Determine the Probability of Deceit Based on AI
|
US 12,023,160 B1 ↗ | Res. 09864 Feb 2026 | rPPG · Physiological AI |
|
Physiological Signal Processing System Based on Artificial Intelligence
|
US 11,607,160 B2 ↗ | Res. 44244 Jun 2019 | Signal Processing · AI |
Our lab pursues three interconnected research directions, all centered on understanding human behavior through physiological signals and machine learning, at scales that were previously not possible.
We use remote photoplethysmography (rPPG) to extract cardiovascular signals from standard RGB video, with no contact sensors or specialized equipment. Our work addresses inter-individual rPPG bias through ipsative normalization, enabling cross-subject generalization in the classification of physiological states under structured assessment protocols.
This line produced two USPTO patents (US 11,607,160 B2 · US 12,023,160 B1) and two Colombian grants, alongside ongoing peer-review submissions.
We challenge the traditional unidimensional arousal model of the autonomic nervous system. Using multi-channel psychophysiological recordings at population scale (electrodermal activity, thoracic and abdominal respiration, and blood pressure), we examine whether the autonomic response to acute psychological stress is better described as a structured, multi-dimensional space.
Population-scale empirical study. Results available upon request for academic collaboration.
We investigate the mechanisms through which personality traits predict workplace misconduct in large-scale field samples. A central focus is understanding whether the Dark Triad's predictive power for counterproductive behaviors operates through its overlap with self-regulatory deficits rather than through its "dark" content per se.
Large-scale field study. Results available upon request for academic collaboration.
We believe that AI systems applied to human assessment carry significant responsibility. Every methodological decision in our lab, from study design to model deployment, is made with that responsibility in mind.
Our work adheres to rigorous standards: pre-registered protocols where applicable, transparent reporting of limitations, fairness evaluation across demographic subgroups, and individual-level explainability as a design requirement.
Our research is conducted in collaboration with leading universities and research institutions across Colombia and internationally.
Our research translates directly into Predicto AIO, a predictive screening system that analyzes eight dimensions of occupational risk using ML models trained on large-scale field data. Every dimension is grounded in peer-reviewed behavioral science taxonomies, includes individual-level explainability, and is evaluated for fairness prior to deployment.
Predicto AIO is distributed through a network of specialized partners in human resources, corporate security, and organizational risk assessment.
We welcome collaborations with researchers, institutions, and organizations. Whether for academic partnership, peer review, or applied research, we're open to conversation.
acuestas@predicto.systems