Key result
A deep learning fuzzy immersion and invariance control approach effectively regulated glucose levels in simulated type-I diabetes patients, maintaining glucose in the desired range for >99% of the time.
Why the study?
Most studies do not consider the glucose-insulin metabolism to be uncertain when developing glucose regulation approaches for type-I diabetes patients.
A novel deep learning fuzzy immersion and invariance control approach effectively regulates glucose levels in simulated type-I diabetes models.
May support adaptive T1D glucose control; hypothesis-generating and requires prospective trials before any clinical consideration.
In this study, a novel approach is proposed for glucose regulation in type-I diabetes patients. Unlike most studies, the glucose-insulin metabolism is considered to be uncertain. A new approach on the basis of the Immersion and Invariance (I&I) theorem is presented to derive the adaptation rules for the unknown parameters. Also, a new deep learned type-II fuzzy logic system (T2FLS) is proposed to compensate the estimation errors and guarantee stability. The suggested T2FLS is tuned by the singular value decomposition (SVD) method and adaptive tuning rules that are extracted from stability investigation. To evaluate the performance, the modified Bergman model (BM) is applied. Besides the dynamic uncertainties, the meal effect on glucose level is also considered. The meal effect is defined as the effect of edibles. Similar to the patient activities, the edibles can also have a major impact on the glucose level. Furthermore, to assess the effect of patient informal activities and the effect of other illnesses, a high random perturbation is applied to glucose-insulin dynamics. The effectiveness of the suggested approach is demonstrated by comparing the simulation results with some other methods. Simulations show that the glucose level is well regulated by the suggested method after a short time. By examination on some patients with various diabetic condition, it is seen that the suggested approach is well effective, and the glucose level of patients lies in the desired range in more than 99% h.
No takes yet. Share an insight, caveat, or question.
Mosavi et al. (2022) studied Type-I diabetes. Deep learning fuzzy immersion and invariance control vs. Other methods was evaluated on Glucose level in the desired range. A deep learning fuzzy immersion and invariance control approach effectively regulated glucose levels in simulated type-I diabetes patients, maintaining glucose in the desired range for >99% of the time.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: