Stroke is the leading cause of long-term disability. Stroke patients can recover faster with personalized therapy treatments. This requires both clinical assessments and in-home assessments of daily activities. In this paper, we propose a daily activity recognition and assessment system for stroke patients. Our system is able to classify daily activities in real home environments and quantitatively evaluate upper body motions while preserving privacy by utilizing depth videos. Specifically, our system collects the depth videos and skeletal joint data of daily activities using a VicoVR sensor. It then recognizes and localizes clinically relevant actions from continuous untrimmed depth videos using a customized convolutional de-convolutional network. In addition, it assesses the extent of reach and speed metrics of both hands using skeletal joint data. The system has been tested on simulated cooking videos and real-life cooking videos in various kitchens with different room layouts and light conditions. The action recognition accuracies for simulated and real-life videos can reach 90.9% and 87.5%, respectively. With the valuable assessment feedback of our system, therapists can make better personalized treatments for stroke patients.
Interactive technologies are beneficial to stroke recovery as rehabilitation interventions; however, they lack evidence for use as assessment tools. Mystic Isle is a multi-planar full-body rehabilitation game developed using the Microsoft Kinect® V2. It aims to help stroke patients improve their motor function and daily activity performance and to assess the motions of the players. It is important that the assessment results generated from Mystic Isle are accurate. We evaluated the spatial accuracy and measurement validity of a Kinect-based game Mystic Isle in comparison to a gold-standard optical motion capture system, the Vicon system. Thirty participants completed six trials in sitting and standing. Game data from the Kinect sensor and the Vicon system were recorded simultaneously, then filtered and sample rate synchronized. The spatial accuracy was evaluated using Pearson’s r correlation coefficient, signal to noise ratio (SNR) and 3D distance difference. Each arm-joint signal had an average correlation coefficient above 0.9 and a SNR above 5. The hip joints data had less stability and a large variation in SNR. Also, the mean 3D distance difference of joints were less than 10 centimeters. For measurement validity, the accuracy was evaluated using mean and standard error of the difference, percentage error, Pearson’s r correlation coefficient and intra-class correlation (ICC). Average errors of maximum hand extent of reach were less than 5% and the average errors of mean and maximum velocities were about 10% and less than 5%, respectively. We have demonstrated that Mystic Isle provides accurate measurement and assessment of movement relative to the Vicon system.
Technology is constantly pioneering new paths in health care by creating new outlets for medical professionals to care for their patients. With the development of wearable sensors and enhanced forms of wireless communication like Bluetooth low energy (BLE) communication, health data can now be collected wirelessly without compromising the independence of patients. Interactive voice interfaces such as the Amazon Echo provide an easy way for individuals to access a variety of data by using voice commands, making interacting with technology easier for those of limited technological experience. Wearable sensors, such as the Angel Sensor, combined with voice interactive devices, such as the Amazon Echo, have allowed for the development of applications that provide easy user interaction. Here, we propose a smart application that monitors health status by combining the data collection capabilities of the Angel sensor and the voice interfacing capabilities of the Amazon Echo. We also present test results of the Amazon Echo speech recognition on different populations.
Bathing makes an important part of daily activities of people under all ages. However, existing shower systems designed for healthy adults can be a challenge for elder adults with certain limitations or people in wheel chairs. We propose an assistive adjustable smart shower system, which is capable to detect the user’s abilities and disabilities, therefore provide the necessary aids automatically. The shower system considers three user classes: normal healthy user, slow walking user and a user in the wheel chair. An Angel Sensor is utilized to collect user’s health data. A computer then processes the data and classifies the user. The shower system provides the necessary assistance based on the classification result. The average classification accuracy is 95. 2%.
Implemented a movie recommendation system using Collaborative Filtering algorithms. To predict the ratings for movies accurately, two models were investigated. An Alternating Least Squares (ALS) model was implemented using Spark APIs and an autoencoder network was implemented on Tensorflow platform. Implemented models were evaluated using the MovieLens 1M dataset. 70% of the data were randomly selected for training and the rest of the data was used for testing. The mean square errors of ALS model and autoencoder network on test set were 0.877 and 0.989, respectively.
Our laboratory has implemented a personnel detection system, Smart Carpet, which can detect the walking of older adults. The existing system consists of traditional individual components and chips, leading to a large system size, which is inconvenient. To reduce the number of chips and the size of the system, we have implemented a fully functional hardware system for the Smart Carpet using Programmable System on Chip (PSoC). Using PSoC's functional units, we meet the requirement of smart carpet functionality. Using software, the program can be loaded to PSoC chip to control each component on the chip using in the system. The system implemented by a PSoC has been successfully tested with the PSoC development board. Based on the functional units we need, we have designed the necessary peripherals circuitry for the chip, which I have placed onto a Printed Circuit Board (PCB). The I2C communication with PSoC has been applied to achieve multiple boards working cooperatively. The new design can handle 80 sensors. It enjoys smaller size and scalable with carpet size. The reduction in the total number of IC chips is from 17 to 1 and components from 98 to 56 and the reduction of the system size from about 26.3 square inches to 8.7 square inches.