Data Augmentation to Build High Performance DNN for In-bed Posture Classification

Abstract

E-textiles have come to be used instead of several types of common equipment, such as bed-sheets, in some cases. An application using body pressure data collected through such bed-sheet type sensors is the in-bed posture classification expected for pressure ulcer prevention. Since such body pressure data is a kind of low-resolution image, Deep Neural Network (DNN) based algorithms seem suitable. However, it is difficult to collect enough data to use for DNN in this domain because the number of sleep postures obtained from one experiment is small. For an example, the number of postures collected from 19 subjects with four hours of sleep each is only 224. To solve such a small data-size problem in DNN, data augmentation techniques have been proposed. However, random augmentations are not so suitable. Therefore, we investigated appropriate augmentation parameters for this domain. As a result, the combination of the up to ±20% and ±40% random shifts along short and long sides of a bed, the up to ±10 degree rotation, and non-use of other transformations showed the best performance. With the parameters, the built DNN showed 99.7% accuracy and 0.997 Weighted F 1-score for three posture classifications: supine, left and right lateral positions, and 97.1% accuracy and 0.970 Weighted F 1-score for four posture classifications: supine, prone, left and right lateral positions.

Joint Position Estimation for Body Pressure Images during Sleep: An Extension for CPM Using Body Area and Posture Estimation Mashups

“…Xu et al estimated six sleeping postures with 90.8% accuracy by a pressure sensor with 64 × 128 measurement points [10]. Enokibori et al estimated three sleeping postures with 99.7% accuracy by an optimized deep learning method with an augmented dataset [11].…”

“…Figure 5 shows an overview of the measurement. We used the same sheet-type pressure sensor as a previous study [11]. This sensor had 3,200 measurement points (40 × 80), and the sampling rate was set to about 6 Hz.…”

Design of Intelligent Mosquito Nets Based on Deep Learning Algorithms

“…Su et al [11] proposed a hydraulic sensor system underneath the mattress that can estimate the relative systolic blood pressure of a person, and they used this sensor system to monitor blood pressure based on two features, ballistocardiogram pulse strength (BPS) and ballistocardiogram pulse deviation (BPD). Some researchers used bed sensors to detect falls [12][13][14]. Mineharu et al [14] used pressure sensors to detect information of sleep position.…”

“…This approach automatically classifies the sleep position of people and detects the danger of falling in advance with nine types of sleep postures and the possibilities of falling. Enokibori et al [13] also considered care for the elderly living alone and presented a bed monitoring system that included a fall detection function. In their study, the infrared and pressure sensors were used to monitor the bed-going, out-of-bed behavior, and lying-down states of elderly people whose existing state or fall events were detected by applying the finite state machine (FSM) method.…”

A Wearable Sleep Position Tracking System Based on Dynamic State Transition Framework

“…A high-resolution pressure map using Force Sensing Resistor (FSR) sensors was used for classifying three sleep positions such as supine, left lying, and right lying [10]. Three positions except for a prone position were further classified into a total of six postures [11], or four main sleep positions were estimated with Deep Neural Network (DNN) using the pressure map [12]. For specific application use, infant postures (such as prone, supine and seated) were estimated by using a modular pressure-sensitive mat [13].…”