English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Objective: The artificial pancreas (AP) has shown promising potential inachieving closed-loop glucose control for individuals with type 1 diabetesmellitus (T1DM). However, designing an effective control policy for the APremains challenging due to the complex physiological processes, delayed insulinresponse, and inaccurate glucose measurements. While model predictive control(MPC) offers safety and stability through the dynamic model and safetyconstraints, it lacks individualization and is adversely affected byunannounced meals. Conversely, deep reinforcement learning (DRL) providespersonalized and adaptive strategies but faces challenges with distributionshifts and substantial data requirements. Methods: We propose a hybrid controlpolicy for the artificial pancreas (HyCPAP) to address the above challenges.HyCPAP combines an MPC policy with an ensemble DRL policy, leveraging thestrengths of both policies while compensating for their respective limitations.To facilitate faster deployment of AP systems in real-world settings, wefurther incorporate meta-learning techniques into HyCPAP, leveraging previousexperience and patient-shared knowledge to enable fast adaptation to newpatients with limited available data. Results: We conduct extensive experimentsusing the FDA-accepted UVA/Padova T1DM simulator across three scenarios. Ourapproaches achieve the highest percentage of time spent in the desiredeuglycemic range and the lowest occurrences of hypoglycemia. Conclusion: Theresults clearly demonstrate the superiority of our methods for closed-loopglucose management in individuals with T1DM. Significance: The study presentsnovel control policies for AP systems, affirming the great potential ofproposed methods for efficient closed-loop glucose control.

Hybrid Control Policy for Artificial Pancreas via Ensemble Deep Reinforcement Learning