Parkinson’s disease (PD) is a chronic, non-reversible neurodegenerative disorder, and freezing of gait (FOG) is one of the most disabling symptoms in PD as it is often the leading cause of falls and injuries that drastically reduces patients’ quality of life. In order to monitor continuously and objectively PD patients who suffer from FOG and enable the possibility of on-demand cueing assistance, a sensor-based FOG detection solution can help clinicians manage the disease and help patients overcome freezing episodes. Many recent studies have leveraged deep learning models to detect FOG using signals extracted from inertial measurement unit (IMU) devices. Usually, the latent features and patterns of FOG are discovered from either the time or frequency domain. In this study, we investigated the use of the time-frequency domain by applying the Continuous Wavelet Transform to signals from IMUs placed on the lower limbs of 63 PD patients who suffered from FOG. We built convolutional neural networks to detect the FOG occurrences, and employed the Bayesian Optimisation approach to obtain the hyper-parameters. The results showed that the proposed subject-independent model was able to achieve a geometric mean of 90.7% and a F1 score of 91.5%.

A dual-drive H-gantry is commonly used in many industrial processes to meet the requirement of high-precision Cartesian motion. Unlike the rigid-linked gantry stage, the flexure-linked counterpart allows a small degree of rotation of the crossarm to prevent possible damages. However, by this design, the chattering of control signals and inappropriate stiffness of the flexure may induce the resonant modes of the gantry. Hence, to maintain the precision tracking of the midpoint position and the orientation of the gantry, as well as to minimize the vibration on the end effector, we seek the most suitable flexure stiffness and controller parameters by formulating a constrained linear-quadratic optimization problem. Since such a mechatronic design problem is not solvable via standard linear-quadratic regulator formulas, we convert it to a constrained projection gradient-based optimization problem, which can be efficiently solved by direct computation of projection gradient and line search of optimal step length. A fast convergence of parameters is achieved after first several iterations. Through a series of comparative experiments, the effectiveness of the proposed method is validated.

Games are played by a wide variety of audiences. Different individuals will play with different gaming styles and employ different strategic approaches. This often involves interacting with nonplayer characters that are controlled by the game AI. From a developer's standpoint, it is important to design a game AI that is able to satisfy the variety of players that will interact with the game. Thus, an adaptive game AI that can scale the difficulty of the game according to the proficiency of the player has greater potential to customize a personalized and entertaining game experience compared to a static game AI. In particular, dynamic game difficulty scaling refers to the use of an adaptive game AI that performs game adaptations in real time during the game session. This paper presents two adaptive algorithms that use ideas from reinforcement learning and evolutionary computation to improve player satisfaction by scaling the difficulty of the game AI while the game is being played. The effects of varying the learning and mutation rates are examined and a general rule of thumb for the parameters is proposed. The proposed algorithms are demonstrated to be capable of matching its opponents in terms of mean scores and winning percentages. Both algorithms are able to generalize well to a variety of opponents.

Freezing of Gait (FoG) is a common motor related impairment among Parkinson's disease patients which substantially reduces their quality of life and puts them at risk of falls. These patients benefit from wearable FoG detection systems that provide timely biofeedback cues and hence help them regain control over their gait. Unfortunately, the systems proposed thus far are bulky and obtrusive when worn. The objective of this paper is to demonstrate the first integration of a FoG detection system into a single sensor node. To achieve such an integration, features with low computational load are selected and dedicated hardware is designed that limits area and memory utilization. Classification is achieved with a neural network that is capable of learning in real-time and thus allows the system to adapt to a patient during run-time. A small form factor FPGA implements the feature extraction and classification, while a custom PCB integrates the system into a single node. The system fits into a $4.5\times 3.5\times 1.5 cm^3$ housing case, weighs 32 g and achieves 95.6% sensitivity and 90.2% specificity when adapted to a patient. Biofeedback cues are provided either through auditory or somatosensory means and the system can remain operational for longer than 9 hours while providing cues. The proposed system is highly competitive in terms of classification performance and excels with respect to wearability and real-time patient adaptivity.

Most acute stroke patients with disabilities do not receive recommended rehabilitation following discharge to the community. Functional and social barriers are common reasons for non-adherence to post-discharge rehabilitation. Home rehabilitation is an alternative to centre-based rehabilitation but is costlier. Tele-rehabilitation is a possible solution, allowing for remote supervision of rehabilitation and eliminating access barriers. The objective of the Singapore Tele-technology Aided Rehabilitation in Stroke (STARS) trial is to determine if a novel tele-rehabilitation intervention for the first three months after stroke admission improves functional recovery compared to usual care. This is a single blind (evaluator blinded), parallel, two-arm randomised controlled trial study design involving 100 recent stroke patients. The inclusion criteria are age ≥40 years, having caregiver support and recent stroke defined as stroke diagnosis within 4 weeks. Consenting participants will be randomized with varying block size of 4 or 6 assuming a 1:1 treatment allocation with the participating centre as the stratification factor. The baseline assessment will be done within 4 weeks of stroke onset, followed by follow-up assessments at 3 and 6 months. The tele-rehabilitation intervention lasts for 3 months and includes exercise 5-days-a-week using an iPad-based system that allows recording of daily exercise with video and sensor data and weekly video-conferencing with tele-therapists after data review. Those allocated to the control group will receive usual care. The primary outcome measure is improvement in life task’s social activity participation at three months as measured by the disability component of the Jette Late Life Functional and Disability Instrument (LLFDI). Secondary outcome variables consist of gait speed (Timed 5-Meter Walk Test) and endurance (Two-Minute Walk test), performance of basic activities of daily living (Shah-modified Barthel Index), balance confidence (Activities-Specific Balance Confidence Scale), patient self-reported health-related quality-of-life [Euro-QOL (EQ-5D)], health service utilization (Singapore Stroke Study Health Service Utilization Form) and caregiver reported stress (Zarit Caregiver Burden Inventory). The goal of this trial is to provide evidence on the potential benefit and cost-effectiveness of this novel tele-rehabilitation programme which will guide health care decision-making and potentially improve performance of post-stroke community-based rehabilitation. This trial protocol was registered under ClinicalTrials.gov on 18 July 2013 as study title “The Singapore Tele-technology Aided Rehabilitation in Stroke (STARS) Study” (ID: The STARS Study, ClinicalTrials.gov Identifier: NCT01905917 ).