Optimal Sensing for Patient Health Monitoring

Miller, Delbert C.; Zhou, Zhengyuan; Bambos, Nicholas; Ben‐Gal, Irad

doi:10.1109/icc.2018.8422884

Cited by 5 publications

(9 citation statements)

References 19 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…For example, given = 6, the decision time-frame will consist of 6 epochs (30 minutes). We may define the set of actions as = {(4), (2,6), (1, 3, 5)}, where each tuple contains the index of epochs for which the sensor will carry out a measurement. The first tuple can be expressed as setting the sensor to a low frequency setting, whereas the last may be expressed as a high frequency setting.…”

Section: Sensors and Actionsmentioning

confidence: 99%

“…To compare between policies, we will examine the Total Cost, the power cost and the health state classification accuracy (calculated using the most likely health state, according to the belief state vector). We set the following parameters: = 6 , = 100 , = 50 and = {(4), (2,6), (1,3,5)}, where the last one is the set of possible actions (the epoch indices in which the sensor is activated).…”

Section: Model Parameter Extractionmentioning

confidence: 99%

“…The sensing process is managed by a computing unit, which selects a subset of sensors to be activated at a given time, analyses the obtained measurements, and transmits the information and initial results to a central unit (server) for further analysis and decision making. WBANs operation allows, in particular, estimating the physical and health condition of a patient and monitoring the evolution of these heath conditions over time and varying conditions [1], [2], [3], [7].…”

Section: Introductionmentioning

confidence: 99%

“…Recent research and practical developments in sensors and WBAN technology allow efficient implementations of WBANs in modern real-world applications [3], [4], [7]. One of the main factors that impacts the efficiency of such systems in the context of health monitoring is energy consumption as well as the associated trade-off between the amount of required energy to operate the system and the accuracy of health state estimation based on the collected information [1], [2]. This paper proposes a model and algorithms that account for both factors: the consumption of energy by the system and the level of accuracy that it provides.…”

Section: Introductionmentioning

confidence: 99%

“…Partially observable Markov decision processes (POMDP) have been used to model the behavior and information transitions in WBANs [2], [5], [7], [11]. A POMDP is defined by a set of states, a set of actions, the conditional transitions between states, a cost function, and a set of observations [6].…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

Wireless Body Area Network Control Policies for Energy-Efficient Health Monitoring

David¹,

Geller²,

Bistritz³

et al. 2021

Preprint

Self Cite

View full text Add to dashboard Cite

Wireless body area networks (WBANs) have strong potential in the field of health monitoring. However, the energy consumption required for accurate monitoring limits the time between battery charges of the wearable sensors, which is a key performance factor (and can be critical in the case of implantable devices). In this paper, we study the inherent trade-off between the power consumption of the sensors and the probability of misclassifying a patient’s health state. We formulate this trade-off as a dynamic problem, in which at each step we can choose to activate a subset of sensors that provide noisy measurements of the patient’s health state. We assume that the (unknown) health state follows a Markov chain, so our problem is formulated as a partially observable Markov decision problem (POMDP). We show that all the past measurements can be summarized as a belief state on the true health state of the patient, which allows tackling the POMDP problem as an MDP on the belief state. We then empirically study the performance of a greedy one-step look-ahead policy compared to the optimal policy obtained by solving the dynamic program. For that purpose, we use an open-source Continuous Glucose Monitoring (CGM) data set of 232 patients over six months and extract the transition matrix and sensor accuracies from the data. We find that the greedy policy saves ~50% of the energy costs while reducing the misclassification costs by less than 2% compared to the most accurate policy possible that always activates all sensors. Our sensitivity analysis reveals that the greedy policy remains nearly optimal across different cost parameters and a varying number of sensors. The results also have practical importance, because while the optimal policy is too complicated, a greedy one-step look-ahead policy can be easily implemented in WBAN systems.

show abstract

Section: Sensors and Actionsmentioning

confidence: 99%

Section: Model Parameter Extractionmentioning

confidence: 99%