Despite the popularity of home medical devices, serious safety concerns have been raised, because the use-errors of home medical devices have linked to a large number of fatal hazards. To resolve the problem, we introduce a cognitive assistive system to automatically monitor the use of home medical devices. Being able to accurately recognize user operations is one of the most important functionalities of the proposed system. However, even though various action recognition algorithms have been proposed in recent years, it is still unknown whether they are adequate for recognizing operations in using home medical devices. Since the lack of the corresponding database is the main reason causing the situation, at the first part of this paper, we present a database specially designed for studying the use of home medical devices. Then, we evaluate the performance of the existing approaches on the proposed database. Although using state-of-art approaches which have demonstrated near perfect performance in recognizing certain general human actions, we observe significant performance drop when applying it to recognize device operations. We conclude that the tiny action involved in using devices is one of the most important reasons leading to the performance decrease. To accurately recognize tiny actions, it's critical to focus on where the target action happens, namely the region of interest(ROI) and have more elaborate action modeling based on the ROI. Therefore, in the second part of this paper, we introduce a simple but effective approach to estimating ROI for recognizing tiny actions. The key idea of this method is to analyze the correlation between an action and the sub-regions of a frame. The estimated ROI is then used as a filter for building more accurate action representations. Experimental results show significant performance improvements over the baseline methods by using the estimated ROI for action recognition.

The healthcare system is in crisis due to challenges including escalating costs, the inconsistent provision of care, an aging population, and high burden of chronic disease related to health behaviors. Mitigating this crisis will require a major transformation of healthcare to be proactive, preventive, patient-centered, and evidence-based with a focus on improving quality-of-life. Information technology, networking, and biomedical engineering are likely to be essential in making this transformation possible with the help of advances, such as sensor technology, mobile computing, machine learning, etc. This paper has three themes: 1) motivation for a transformation of healthcare; 2) description of how information technology and engineering can support this transformation with the help of computational models; and 3) a technical overview of several research areas that illustrate the need for mathematical modeling approaches, ranging from sparse sampling to behavioral phenotyping and early detection. A key tenet of this paper concerns complementing prior work on patient-specific modeling and simulation by modeling neuropsychological, behavioral, and social phenomena. The resulting models, in combination with frequent or continuous measurements, are likely to be key components of health interventions to enhance health and wellbeing and the provision of healthcare.

The Smart Health and Wellbeing workshop is organized to develop a platform for authors to discuss fundamental principles, algorithms or applications of intelligent data acquisition, processing and analysis of healthcare data. We are particularly interested in information and knowledge management papers, in which the approaches are accompanied by an in-depth experimental evaluation with real world data. This paper provides an overview of the workshop and the accepted contributions.

The Smart Health and Wellbeing workshop is organized to develop a platform for authors to discuss fundamental principles, algorithms or applications of intelligent data acquisition, processing and analysis of healthcare data. We are particularly interested in information and knowledge management papers, in which the approaches are accompanied by an in-depth experimental evaluation with real world data. This paper provides an overview of the workshop and the accepted contributions.

We adapt and compare several tracking algorithms to find a good combination method for tracking residents in a nursing home over extended periods of time. Since the cameras are scattered throughout the facility, tracking across multiple cameras and rooms cannot be handled by a single method, but different approaches are appropriate for different transitions between cameras. Overall, combining mean shift, person detection, SIFT and the new MoSIFT interest point trackers, we are able to track residents with good accuracy across multiple cameras, views and rooms. This provides a basis for analysis of activity levels and behaviors of clinical interest.

To detect errors when subjects operate a home medical device, we observe them with multiple cameras. We then perform action recognition with a robust approach to recognize action information based on explicitly encoding motion information. This algorithm detects interest points and encodes not only their local appearance but also explicitly models local motion. Our goal is to recognize individual human actions in the operations of a home medical device to see if the patient has correctly performed the required actions in the prescribed sequence. Using a specific infusion pump as a test case, requiring 22 operation steps from 6 action classes, our best classifier selects high likelihood action estimates from 4 available cameras, to obtain an average class recognition rate of 69%.

ASSISTIVE TECHNOLOGIES ARE RELATIVELY novel tools for research and daily care in long-term care (LTC) facilities that are faced with the burgeoning of the older adult population and dwindling staffing resources. The degree to which stakeholders in LTC facilities are receptive to the use of these technologies is poorly understood. Eighteen semi-structured focus groups and one interview were conducted with relevant groups of stakeholders at seven LTC facilities in southwestern Pennsylvania. Common themes identified across all focus groups centered on concerns for privacy, autonomy, cost, and safety associated with implementation of novel technologies. The relative importance of each theme varied by stakeholder group as well as the perceived severity of cognitive and/or physical disability. Our findings suggest that ethical issues are critical to acceptance of novel technologies by their end users, and that stakeholder groups are interdependent and require shared communication about the acceptance of these emerging technologies.

Objective To examine the total and domain-specific prevalence of verbally and physically abusive, socially inappropriate, and care-resistive behaviors according to the Minimum Data Set (MDS) compared with research instruments in nursing home residents with severe dementia. Design, Setting, and Methods As part of a longitudinal observational study, MDS behavioral symptoms data were compared with corresponding items from the Ryden Aggression Scale and the Cohen-Mansfield Agitation Inventory for 15 nursing home residents with severe dementia. McNemar's test was used to compare the difference in the proportion of subjects who experienced any symptoms, as well as specific symptoms in several domains, according to the MDS and the research instruments. Additionally, temporal fluctuations in behavioral symptoms were descriptively and graphically summarized. Results The MDS significantly underestimated both the total proportion of subjects experiencing any behavioral symptoms (P = .016), as well as the proportion of subjects experiencing verbally abusive symptoms (P < .002), physically abusive symptoms (P = .008), or socially inappropriate behaviors (P = .016) compared with corresponding items from the research instruments. Moreover, these behaviors exhibited considerable temporal instability, suggesting that the systematic daily collection of measures of behavioral disturbances is imperative during the week in which the MDS assessment is to be completed. Discussion Albeit from a small study sample, our findings call into question the validity of the MDS behavioral symptom items as they are currently recorded, and suggest that a simple intervention of twice daily completion of a behavioral symptoms checklist containing the MDS items during the week of the assessment may significantly improve the accuracy of the recorded data.

Video surveillance is an alternative approach to staff or self-reporting that has the potential to detect and monitor aggressive behaviors more accurately. In this paper, we propose an automatic algorithm capable of recognizing aggressive behaviors from video records using local binary motion descriptors. The proposed algorithm may increase the accuracy for retrieving aggressive behaviors from video records, and thereby facilitates scientific inquiry into this low frequency but high impact phenomenon that eludes other measurement approaches.

Automated detection of aggressive behaviors captured in continuously recorded nursing home video can increase the accuracy of those reported by subjects and caregivers. We implement a detection algorithm based on the extraction of features as local binary motion descriptors, and apply a novel clustering algorithm to merge similar LBMD's into a video codebook, and build recognizers to classify aggressive from non-aggressive behaviors. This facilitates clinical investigation into this difficult to measure low frequency but high impact behavior.

This paper presents the application of computer vision and machine learning technologies to a clinical task of paramount importance, improving safety of older persons. We propose an intelligent monitoring system equipped with a camera network and an automatic elopement detection algorithm to reduce the risks of un-witnessed elopements from a dementia unit in order to avoid their potential catastrophic consequences. The camera network employs 23 cameras to record daily activities in our test bed, which includes 15 residents, 4 registered and licensed practical nurses and a number of certified nursing assistants. An elopement detector is then built by using computer vision algorithms and a machine learning algorithm to automatically detect elopements and alert caregivers. The experiments demonstrate that the proposed system leverages the advantages of monitoring from multiple cameras and is able to detect elopements with almost 100% accuracy.

A number of researchers have been building high-level semantic concept detectors such as outdoors, face, building, to help with semantic video retrieval. Our goal is to examine how many concepts would be needed, and how they should be selected and used. Simulating performance of video retrieval under different assumptions of concept detection accuracy, we find that good retrieval can be achieved even when detection accuracy is low, if sufficiently many concepts are combined. We also derive suggestions regarding the types of concepts that would be most helpful for a large concept lexicon. Since our user study finds that people cannot predict which concepts will help their query, we also suggest ways to find the best concepts to use. Ultimately, this paper concludes that "concept-based" video retrieval with fewer than 5000 concepts, detected with a minimal accuracy of 10% mean average precision is likely to provide high accuracy results in broadcast news retrieval.

This paper discusses the application of speech alignment, image processing, and language understanding technologies to build efficient interfaces into large digital oral history archives, as exemplified by a thousand hour HistoryMakers corpus. Browsing, querying, and navigation features are discussed.

Nearly 2.5 million Americans currently reside in nursing homes and assisted living facilities in the United States, accounting for approximately five percent of persons sixty-five and older. The aging of the “Baby Boomer” generation is expected to lead to an exponential growth in the need for some form of long-term care (LTC) for this segment of the population within the next twenty-five years. In light of these sobering demographic shifts, there is an urgency to address the profound concerns that exist about the quality-of-care (QoC) and quality-of-life (QoL) of this frailest segment of our population.

CareMedia is a collaborative effort that to date has captured more than 13,000 hours of video and audio recordings of life in the shared spaces of a nursing home dementia unit, by using 23 ceiling-mounted cameras, 24 hours a day for 25 days, ensuring an un-occluded view of every point in the recorded space. Computer machine learning techniques are being applied to the resulting 25 Terabytes of data, automatically processing the record for efficient use by analytical observers (e.g., social and behavioral scientists, geriatricians, and healthcare policy makers) to monitor and understand residents' well-being, and enhance their quality of life. This truly interdisciplinary effort bridges the psychological, social and behavioral sciences, and clinical medicine with multiple engineering and computer science disciplines to establish a clinical evidence base to guide rational therapeutics, an elusive goal ardently articulated by the Institute of Medicine. This paper discusses early foundation work being conducted with the data.

Nearly 2.5 million Americans currently reside in nursing homes and assisted living facilities in the United States, accounting for approximately 5% of persons 65 years and older.The aging of the “Baby Boomer” generation is expected to lead to an exponential growth in the need for some form of long-term care (LTC) for this segment of the population within the next 25 years. In light of these sobering demographic shifts, there is an urgency to address the profound concerns that exist about the quality-of-care (QoC) and quality-of-life (QoL) of this frailest segment of our population.

Digital imagery for significant cultural and historical materials is an emerging research field that bridges people, culture, and technologies. In this paper, we first discuss the great importance of this field. Then we focus on its four interrelated subareas: (1) creation and preservation, (2) retrieval, (3) presentation and usability, and (4) applications and use. We propose several mechanisms to encourage collaboration and argue that the field has high potential impact on our digital society. Finally, we make specific recommendations on what to pursue in this field.