Fiber-optics IoT healthcare system based on deep reinforcement learning combinatorial constraint scheduling for hybrid telemedicine applications
Language English Country United States Media print-electronic
Document type Journal Article, Research Support, Non-U.S. Gov't
PubMed
38870728
DOI
10.1016/j.compbiomed.2024.108694
PII: S0010-4825(24)00779-0
Knihovny.cz E-resources
- Keywords
- Adaptive security deep q-learning network, Deep reinforcement learning, Fiber-optics, Internet of things, Telemedicine applications,
- MeSH
- Algorithms MeSH
- Deep Learning * MeSH
- Internet of Things * MeSH
- Humans MeSH
- Fiber Optic Technology MeSH
- Telemedicine * MeSH
- Check Tag
- Humans MeSH
- Publication type
- Journal Article MeSH
- Research Support, Non-U.S. Gov't MeSH
Telemedicine is an emerging development in the healthcare domain, where the Internet of Things (IoT) fiber optics technology assists telemedicine applications to improve overall digital healthcare performances for society. Telemedicine applications are bowel disease monitoring based on fiber optics laser endoscopy, gastrointestinal disease fiber optics lights, remote doctor-patient communication, and remote surgeries. However, many existing systems are not effective and their approaches based on deep reinforcement learning have not obtained optimal results. This paper presents the fiber optics IoT healthcare system based on deep reinforcement learning combinatorial constraint scheduling for hybrid telemedicine applications. In the proposed system, we propose the adaptive security deep q-learning network (ASDQN) algorithm methodology to execute all telemedicine applications under their given quality of services (deadline, latency, security, and resources) constraints. For the problem solution, we have exploited different fiber optics endoscopy datasets with images, video, and numeric data for telemedicine applications. The objective is to minimize the overall latency of telemedicine applications (e.g., local, communication, and edge nodes) and maximize the overall rewards during offloading and scheduling on different nodes. The simulation results show that ASDQN outperforms all telemedicine applications with their QoS and objectives compared to existing state action reward state (SARSA) and deep q-learning network (DQN) policy during execution and scheduling on different nodes.
References provided by Crossref.org
