Teddy Lazebnik

Q-learning policy trained on genetic-algorithm targets lets simulated locust-shaped agents with only visual input flock (CM score 0.7) and avoid collisions (CA 0.74 vs 0.17 for a flocking-only policy).

The paper presents a spatio-temporal PKPD model for nanoparticle-based drugs, showing improved fidelity over existing models in mouse experiments with 13 NPs.

This study introduces a graph-based PKPD model for nanoparticles to predict treatment outcomes, achieving high fitting accuracy in in vivo experiments.

The paper introduces a mathematical model to assess multi-strain pandemics' impact on mortality, reproduction number, and infection peaks, showing logarithmic growth in infected numbers and equivalence to single-strain reproduction numbers.

This paper develops a mathematical model of BCG immunotherapy for type 1 diabetes, showing it can delay disease effects and provide long-term benefits but does not eliminate the disease.