Steen J. Hsu

Abnormality detection for improving elder's daily life independent

Since the dramatic demographic change makes it inevitable that rapid aging of the population is an unprecedented phenomenon in Taiwan. A growing social problem is supporting older adults who want to live independently in their own homes. It needs a health assistance system to make them independent living up to a higher age. Recently, technological advancements have spurred various ideas and innovations to assist the elders living independently. In this paper, we proposed a homecare sensory system that uses RFID-based sensor networks to collect elders daily activities and conducts the data into Hidden Markov model (HMM) and Support Vector Machines (SVMs) to estimate whether the elders behavior is abnormal or not. Through detecting and distinguishing the abnormal behaviors of elders daily activities, the system provides assistance on elders independent living and improvement of aged quality of life.

Location-Aware Fall Detection System for Medical Care Quality Improvement

Abstract—Falls are one of the most common adverse events in hospitals and fall management remains a major challenge in the medical care quality. Falls in patients are associated with major health complications that can result in health decline and increased medical care cost. To deliver medical care in time, reliable location-aware fall detection is needed. In this paper, we propose a patient alert alarm for fall management. It is ZigBee-Based location awareness fall detection system that provides immediate position information to the caregivers as soon as the fall happened. Obviously, the integration of location awareness and fall detection technologies fulfills the requirements of delivering critical information to relative professions and improve the medical care quality.

Task Scheduling Based on Load Approximation in Cloud Computing Environment

Cloud computing is an emerging technology and gain attention in academic and business area. Resources are pooled and share with users on demand. In order to provide better performance of cloud environment, task scheduling is an important issue. On the other hand, in large scale cloud environment, cloud devices search and query influence the performance also. Therefore, in this paper we propose a AVL-tree based cloud computing environment to organized the cloud devices in O(logn) regarding the cloud device management operations complexity and task scheduling considering the real-time average system load of cloud computing environment.

An Improved Data Hiding Method Using Image Interpolation

Reversible data hiding method enables to recover the cover image after the secret data were extracted. Recently, a reversible data hiding method based on image interpolation has been proposed. It can contain high capacity and keep good visual quality. In this study, an improved method has proposed to refine the interpolation image based method. Experimental results revealed that the proposed method has the same capacity and promoted the quality of stego-image when compared with the origin method.

The design of UPnP-based home environment over peer-to-peer overlay network

With the population of network usage, it is possible to connect home appliances with each other. The basic demand is to connect home appliances easily with less user intervening. UPnP is the most popular technology for realizing digital home. UPnP provide service and device discovery by sending broadcast message periodically. Therefore, this causes efficiency problem arising from many broadcast messages. In addition, home network is an emergence environment which might comprise several network architecture including Ethernet, wireless network, or power-line etc.. Therefore, traditional broadcast will not work well in the complex network architecture. In order to solve the above problems, we propose a middleware implementation to realize UPnP operations over peer-to-peer overlay network.

Data Collection and Analysis Using Wearable Sensors for Monitoring Knee Range of Motion after Total Knee Arthroplasty

Total knee arthroplasty (TKA) is the most common treatment for degenerative osteoarthritis of that articulation. However, either in rehabilitation clinics or in hospital wards, the knee range of motion (ROM) can currently only be assessed using a goniometer. In order to provide continuous and objective measurements of knee ROM, we propose the use of wearable inertial sensors to record the knee ROM during the recovery progress. Digitalized and objective data can assist the surgeons to control the recovery status and flexibly adjust rehabilitation programs during the early acute inpatient stage. The more knee flexion ROM regained during the early inpatient period, the better the long-term knee recovery will be and the sooner early discharge can be achieved. The results of this work show that the proposed wearable sensor approach can provide an alternative for continuous monitoring and objective assessment of knee ROM recovery progress for TKA patients compared to the traditional goniometer measurements.

Activity Recognition and Activity Level Estimation for Context-Based Prompting System of Mild Cognitive Impairment Patients

The number of elderly people, who are unable to live independently and need assistance due to cognitive impairment, will rise rapidly in the aging society. To assist the independent living of these individuals and decrease the caregiver burden have become an important public health concern in the future. Mild Cognitive Impairment (MCI) is an intermediate state between normal cognitive function and dementia. The symptoms of MCI include difficulty remembering recent events or recently acquired information, depression and anxiety. MCI also increases the fall risk and affects patients’ social function and behavior. Sufficient physical activities can improve health of brain and reduce the risk of MCI. Since the context-aware computing technologies for assisting living have gained great popularity. We proposed a context-based activity prompting system to improve quality of life for MCI patients. The proposed system utilizes a smart phone as a sensor device to transmit sensing data to cloud server for activity recognition and activity level estimation, and uses context-based technique to provide activity prompting message to MCI patients. The activity prompting service supplies the activities self-management for MCI patients and helps them living independently. The system also provides real time fall detection mechanism to shorten the rescue time when accident happened. The experimental results have demonstrated that the proposed system achieves high accuracy on activity recognition and activity level estimation.

Design and Implementation of Teleconsultation System for Instant Treatment

Since the medical environment becomes more complicated nowadays, frequent exchange of information and cooperation among the professionals are the indispensable factors to enhance the quality of medical care. An efficient teleconsultation plays an important role for surgical emergency and medical decision making. The advances in the information communication technique during the past decade have already made the remote consulting feasible. However, the remote consulting must provide rapid response time, high quality medical images and flexible cooperation platform. Based on the combination of Computer-Supported Cooperative Work (CSCW) and Picture Archiving and Communication System (PACS), we propose an effective teleconsultation system to achieve the purpose of remote diagnosis support. The Internet virtual community concept is also adopted to provide a convenient and rapid connection manner for participants. Furthermore, the comparison results between proposed system and previous works demonstrate that it fulfills the requirements of remote teleconsultation system.

A confidence-based approach to hand movements recognition for cleaning tasks using dynamic time warping

According to the WHO report in 2013, the world population aging over 60 years is predicted to increase to 20 million in 2050. Aging comes about many challenges to elders due to their cognitive decline, chronic age-related diseases, as well as limitations in physical activity, vision, and hearing. Recent advances in wearable computing and mobile health technology create new opportunity for ambient assisted living system to help the person perform the activities safely and independently. The activity monitoring of daily living is the core technique of the ambient assisted living system. Several well-known approaches have utilized various sensors for activity recognition such as camera, RFID, infrared detector and inertial sensor. Since the activities are well characterized by the objects, location or hand gesture that are manipulated during their performance on activities of daily living. However, some applications included, e.g. the monitoring of specific tasks and/or movements in a rehabilitation scenario or the classification of dietary intake gestures for an automated nutrition monitoring system, where reliable activity recognition on a more fine-grained level is needed. To fulfill the requirement, we design a hierarchical window approach based on the dynamic time warping algorithm to achieve fine-grained activity recognition, where the template selection and threshold configuration is developed to cope with the ambiguity with similar features. Furthermore, a confidence estimation for the pattern matching is also proposed. The recognition procedure was successfully adapted to the investigated cleaning tasks. The overall performance in precision, recall, and F1-socre is 89.0%, 88.6%, and 88.1% respectively. The results of the experiment demonstrate that the proposed mechanism is reliable and fulfills the requirements of the ambient assisted living.

Time-dependent analysis of dosage delivery information for patient-controlled analgesia services

Pain relief always plays the essential part of perioperative care and an important role of medical quality improvement. Patient-controlled analgesia (PCA) is a method that allows a patient to self-administer small boluses of analgesic to relieve the subjective pain. PCA logs from the infusion pump consisted of a lot of text messages which record all events during the therapies. The dosage information can be extracted from PCA logs to provide easily understanding features. The analysis of dosage information with time has great help to figure out the variance of a patient’s pain relief condition. To explore the trend of pain relief requirement, we developed a PCA dosage information generator (PCA DIG) to extract meaningful messages from PCA logs during the first 48 hours of therapies. PCA dosage information including consumption, delivery, infusion rate, and the ratio between demand and delivery is presented with corresponding values in 4 successive time frames. Time-dependent statistical analysis demonstrated the trends of analgesia requirements decreased gradually along with time. These findings are compatible with clinical observations and further provide valuable information about the strategy to customize postoperative pain management.