Marika Rizzo
Neural Optimal Control of a Mathematical Model for Skeletal Muscle Regeneration in mdx Mice.
Rel. Lamberto Rondoni, Davide Carbone. Politecnico di Torino, Corso di laurea magistrale in Ingegneria Biomedica, 2026
|
|
PDF (Tesi_di_laurea)
- Tesi
Accesso limitato a: Solo utenti staff fino al 28 Settembre 2027 (data di embargo). Licenza: Creative Commons Attribution Non-commercial No Derivatives. Download (10MB) |
Abstract
Duchenne muscular dystrophy (DMD) is a severe genetic disease characterized by progressive muscle degeneration and impaired tissue regeneration. This thesis proposes a neural network framework for designing optimal therapeutic strategies aimed at modulating muscle cell populations in dystrophic tissues, with particular attention to the mdx mouse model. The biological dynamics are formulated as a nonlinear ordinary differential equation (ODE) model, which captures the complex interactions between healthy, damaged, and regenerating muscle fibers, along with key inflammatory dynamics. To address the optimal treatment problem, we first focus on a machine learning framework based on Neural Ordinary Differential Equations (Neural ODEs). In this approach, a neural network is directly integrated into the state equations to parameterize the control input, optimizing a cost functional that balances the recovery of normal fibers with the cost of drug administration.
This approach is subsequently compared with a second computational framework based on the Neural Hamilton-Jacobi-Bellman (HJB) formulation
Relatori
Anno Accademico
Tipo di pubblicazione
Numero di pagine
Corso di laurea
Classe di laurea
URI
![]() |
Modifica (riservato agli operatori) |
