A stochastic jump diffusion model for type 2 diabetes (T2D) patients is proposed to account for unknown meals during treatment. The model offers the chance to estimate parameters describing how often does the patient consume carbohydrates and how much is consumed. In addition, a strategy based on a Particle Markov chain Monte Carlo (PMCMC) method combined with parameter learning is proposed to estimate the stochastic parameters with continues glucose monitoring (CGM) data and injected insulin amounts only. The strategy was tested both for clinical and simulated data and was shown to be able to estimate all the stochastic parameters with various degrees of accuracy.
|2021 IEEE Conference on Control Technology and Applications (CCTA)
|09/08/2021 → 11/08/2021
|IEEE Conference on Control Technology and Applications (CCTA) - Proceedings