No abstract

Healthcare electronics count on the effectiveness of the on-patient signal preprocessing unit to moderate the wireless data transfer for better power efficiency. In order to reduce the system power in long-time ECG acquisition, this work describes an on-patient QRS detection processor for arrhythmia monitoring. It extracts the concerned ECG part, i.e., the RR-interval between the QRS complex for evaluating the heart rate variability. The processor is structured by a scale-3 quadratic spline wavelet transform followed by a maxima modulus recognition stage. The former is implemented via a symmetric FIR filter, whereas the latter includes a number of feature extraction steps: zero-crossing detection, peak (zero-derivative) detection, threshold adjustment and two finite state machines for executing the decision rules. Fabricated in 0.35-μm CMOS the 300-Hz processor draws only 0.83 μW, which is favorably comparable with the prior arts. In the system tests, the input data is placed via an on-chip 10-bit SAR analog-to-digital converter, while the output data is emitted via an off-the-shelf wireless transmitter (TI CC2500) that is configurable by the processor for different data transmission modes: 1) QRS detection result, 2) raw ECG data or 3) both. Validated with all recordings from the MIT-BIH arrhythmia database, 99.31% sensitivity and 99.70% predictivity are achieved. Mode 1 with solely the result of QRS detection exhibits 6× reduction of system power over modes 2 and 3.

No abstract

Abstract-In this paper, we propose an efficient algorithm to find an optimal control policy in a discrete-time hidden mode stochastic hybrid system, which is a special case of partially observable discrete-time stochastic hybrid systems in which only discrete states are hidden. Many human-centered systems can be modeled as such systems, in which the intent of the human operator is unknown and can be modeled as the hidden mode. In the literature, the optimal control problem of hidden mode stochastic hybrid system is known to have high computational complexity due to the continuous state space. In this paper, we will tackle this computational challenge by using local quadratic functions to approximate the optimal expected reward, which does not have a closed-form expression in general. We will show the efficacy of our proposed method, and the significant improvement in the computational time.

To solve the research–practice gap and take one step forward toward using big data with real-world evidence, the present study aims to adopt a novel method using machine learning to pool findings from meta-analyses and predict the change of countermovement jump. The data were collected through a total of 124 individual studies included in 16 recent meta-analyses. The performance of four selected machine learning algorithms including support vector machine, random forest (RF) ensemble, light gradient boosted machine, and the neural network using multi-layer perceptron was compared. The RF yielded the highest accuracy (mean absolute error: 0.071 cm; R2: 0.985). Based on the feature importance calculated by the RF regressor, the baseline CMJ (“Pre-CMJ”) was the most impactful predictor, followed by age (“Age”), the total number of training sessions received (“Total number of training_session”), controlled or non-controlled conditions (“Control (no training)”), whether the training program included squat, lunge, deadlift, or hip thrust exercises (“Squat_Lunge_Deadlift_Hipthrust_True”, “Squat_Lunge_Deadlift_Hipthrust_False”), or “Plyometric (mixed fast/slow SSC)”, and whether the athlete was from an Asian pacific region including Australia (“Race_Asian or Australian”). By using multiple simulated virtual cases, the successful predictions of the CMJ improvement are shown, whereas the perceived benefits and limitations of using machine learning in a meta-analysis are discussed.