Abstract:Rapid advances in the smartphone, wearable sensing, and wireless communication provide an unprecedented opportunity to develop mobile systems for smart health management. Mobile cardiac sensing collects health-related data from individuals and enables the extraction of information pertinent to cardiac conditions. However, wireless sensors in ambulatory care settings operate on batteries. All-time sensing and monitoring will result in fast depletion of the battery in the mobile system. There is an urgent need to develop optimal sensing schemes that will reduce energy consumption while satisfying the requirements in the detection of cardiac events. In this article, we develop a constrained Markov decision process (CMDP) framework to optimize mobile electrocardiography (ECG) sensing under the constraint of the energy budget. We first characterize the cardiac states from ECG signals using the heterogeneous recurrence analysis. Second, we model the stochastic dynamics in cardiac processes as a continuous-time Markov chain (CTMC). Third, we optimize the ECG sensing through a CMDP framework under the constraint of energy budget. Finally, we validate and evaluate the performance of our CMDP policy in both simulation and real-world case studies. Experimental results demonstrate that the proposed CMDP policy significantly outperforms the traditional uniform and mean-time-to-event (MTTE) policies. Specifically, the error of state estimation is reduced by 34.0% in the real-world case study for energy-constrained sensing of cardiac events. Note to Practitioners—This article is motivated by the markedly increasing applications of mobile health (mHealth) in cardiac care. MHealth systems enable the real-time monitoring, tracking, and transmitting of heart-health information but suffer from the problem with limited battery life. This article presents a novel energy-efficient framework for mobile ECG sensing by integrating ECG s-gnal analysis with sensing-policy optimization under the constraint of the energy budget. Experimental results demonstrate the effectiveness of the proposed framework to realize energy-efficient mobile sensing for cardiac events detection.

Fast Adaptation of Activity Sensing Policies in Mobile Devices

Stochastic Mobile Energy Replenishment and Adaptive Sensor Activation for Perpetual Wireless Rechargeable Sensor Networks

Stochastic Steepest-Descent Optimization Of Multiple-Objective Mobile Sensor Coverage

Delay-Aware Stochastic Resource Management for Mobile Edge Computing Systems Via Constrained Reinforcement Learning

Energy Optimization for Federated Learning on Consumer Mobile Devices with Asynchronous SGD and Application Co-Execution

Constrained Markov Decision Process Modeling for Optimal Sensing of Cardiac Events in Mobile Health

Optimal Sensing Policies for Smartphones in Hybrid Networks: A POMDP Approach

Utility Optimization for Multi-User Task Offloading in Mobile Ad Hoc Cloud: A Stochastic Game Approach.

Off-Policy Estimation of Long-Term Average Outcomes with Applications to Mobile Health

Adaptive Optimization of Time-out Policy for Dynamic Power Management Based on SMCP

Personalized Policy Learning Using Longitudinal Mobile Health Data

Estimating Optimal Infinite Horizon Dynamic Treatment Regimes via pT-Learning

A Non-stochastic Learning Approach to Energy Efficient Mobility Management

Adaptive Dynamic Programming for Energy-Efficient Base Station Cell Switching

AD-Sense: activity-driven sensing for mobile devices.

An Energy-Aware Online Learning Framework for Resource Management in Heterogeneous Platforms

Performance Optimization in Mobile-Edge Computing Via Deep Reinforcement Learning

Decentralized Computation Offloading for Multi-User Mobile Edge Computing: A Deep Reinforcement Learning Approach

CODA: A COst-efficient Test-time Domain Adaptation Mechanism for HAR

Deep Reinforcement Learning Based Task Offloading and Resource Allocation in Small Cell MEC

Adaptive Configuration of Service-Based Smart Sensors in Edge Networks