Re-engineering the disordered mind: clinical experimentation, dynamical systems, and AI for personalized psychiatry

This perspective proposes a neuropsychiatric model of psychological and psychiatric interventions by reframing treatment as a control engineering problem grounded in dynamical systems theory and artificial intelligence (AI). We argue that psychopathology arises from distortions in the geometry of underlying neurobehavioral low-dimensional cognitive-affective manifolds rather than from isolated biological dysfunctions, and we use a formal dynamical framework to show how clinical interventions can be modeled as control inputs that reshape the manifold itself to restore healthy dynamics. To operationalize this approach clinically, we propose a closed-loop, N-of-1 experimental paradigm in which dense longitudinal measurements and strategically designed perturbations are used to train individualized AI surrogate models of a person's manifold. This model supports the simulation of counterfactual interventions and guide the design of optimized, personalized treatments. Active perturbation reduces required sample size dramatically, enabling precise modeling from limited but richly sampled individual data. This engineering-inspired framework reconceptualizes clinical improvement as the restoration of regulatory capacity and resilient trajectories rather than the mere reduction of symptom counts. By integrating dynamical systems theory, AI-based surrogate modeling, and adaptive clinical experimentation, we outline a principled pathway toward personalized neuropsychiatry based on dynamical systems theory and AI.