Frenetic Cat-inspired Particle Optimization: a Markov state-switching hybrid swarm optimizer with application to cardiac digital twinning
A new 'frenetic cat' swarm algorithm calibrates digital heart twins in ~40 iterations, beating established methods.
A team of researchers including Jorge Sánchez has introduced Frenetic Cat-inspired Particle Optimization (FCPO), a novel hybrid swarm intelligence algorithm designed for expensive black-box optimization problems. The method uniquely couples particle swarm dynamics with an explicit-state Markov switching controller to intelligently schedule exploration and refinement operations on the fly. It integrates several advanced techniques: state-conditioned bounded motion, an elite-difference global jump operator to escape local optima, eigen-space guided local refinement, and linear population size reduction to control computational costs.
Benchmarked against established optimizers like PSO, CMA-ES, and L-SHADE on standard test functions, FCPO demonstrated superior speed, achieving the lowest mean runtime across ten cases. It was approximately 2.3 times faster than the powerful CMA-ES algorithm. The real-world impact was validated in a cardiac digital twinning task, where FCPO successfully calibrated a model of ventricular electrical activity to match target electrocardiogram (ECG) readings with high fidelity (RMSE < 0.1 mV) in roughly 40 iterations, producing physiologically plausible results. This performance, combined with robust convergence across different starting points, positions FCPO as a practical and efficient tool for complex inverse problems in medicine and engineering where each simulation is computationally costly.
- FCPO is a hybrid swarm optimizer using a Markov controller to switch between exploration and refinement states dynamically.
- It achieved an average runtime of 0.183 seconds in benchmarks, running 2.3x faster than the CMA-ES algorithm.
- Successfully calibrated a cardiac digital twin model to match an ECG signal within ~40 iterations, demonstrating practical utility for expensive simulations.
Why It Matters
This enables faster, more efficient calibration of complex simulation models like digital hearts, accelerating medical research and personalized healthcare.