Reduction of Time on the Ground Related to Real-Time Video Detection of Falls in Memory Care Facilities: Observational Study.

BACKGROUND: Lying on the floor for a long period of time has been described as a critical determinant of prognosis following a fall. In addition to fall-related injuries due to the trauma itself, prolonged immobilization on the floor results in a wide range of comorbidities and may double the risk of death in elderly. Thus, reducing the length of Time On the Ground (TOG) in fallers seems crucial in vulnerable individuals with cognitive disorders who cannot get up independently. OBJECTIVE: This study aimed to examine the effect of a new technology called SafelyYou Guardian (SYG) on early post-fall care including reduction of Time Until staff Assistance (TUA) and TOG. METHODS: SYG uses continuous video monitoring, artificial intelligence, secure networks, and customized computer applications to detect and notify caregivers about falls in real time while providing immediate access to video footage of falls. The present observational study was conducted in 6 California memory care facilities where SYG was installed in bedrooms of consenting residents and families. Fall events were video recorded over 10 months. During the baseline installation period (November 2017 to December 2017), SYG video captures of falls were not provided on a regular basis to facility staff review. During a second period (January 2018 to April 2018), video captures were delivered to facility staff on a regular weekly basis. During the third period (May 2018 to August 2018), real-time notification (RTN) of any fall was provided to facility staff. Two digital markers (TUA, TOG) were automatically measured and compared between the baseline period (first 2 months) and the RTN period (last 4 months). The total number of falls including those happening outside of the bedroom (such as common areas and bathrooms) was separately reported by facility staff. RESULTS: A total of 436 falls were recorded in 66 participants suffering from Alzheimer disease or related dementias (mean age 87 years; minimum 65, maximum 104 years). Over 80% of the falls happened in bedrooms, with two-thirds occurring overnight (8 PM to 8 AM). While only 8.1% (22/272) of falls were scored as moderate or severe, fallers were not able to stand up alone in 97.6% (247/253) of the cases. Reductions of 28.3 (CI 19.6-37.1) minutes in TUA and 29.6 (CI 20.3-38.9) minutes in TOG were observed between the baseline and RTN periods. The proportion of fallers with TOG >1 hour fell from 31% (8/26; baseline) to zero events (RTN period). During the RTN period, 76.6% (108/141) of fallers received human staff assistance in less than 10 minutes, and 55.3% (78/141) of them spent less than 10 minutes on the ground. CONCLUSIONS: SYG technology is capable of reducing TOG and TUA while efficiently covering the area (bedroom) and time zone (nighttime) that are at highest risk. After 6 months of SYG monitoring, TOG was reduced by a factor of 3. The drastic reduction of TOG is likely to decrease secondary comorbid complications, improve post-fall prognosis, and reduce health care costs.

Fully Automated Echocardiogram Interpretation in Clinical Practice.

BACKGROUND: Automated cardiac image interpretation has the potential to transform clinical practice in multiple ways, including enabling serial assessment of cardiac function by nonexperts in primary care and rural settings. We hypothesized that advances in computer vision could enable building a fully automated, scalable analysis pipeline for echocardiogram interpretation, including (1) view identification, (2) image segmentation, (3) quantification of structure and function, and (4) disease detection. METHODS: Using 14 035 echocardiograms spanning a 10-year period, we trained and evaluated convolutional neural network models for multiple tasks, including automated identification of 23 viewpoints and segmentation of cardiac chambers across 5 common views. The segmentation output was used to quantify chamber volumes and left ventricular mass, determine ejection fraction, and facilitate automated determination of longitudinal strain through speckle tracking. Results were evaluated through comparison to manual segmentation and measurements from 8666 echocardiograms obtained during the routine clinical workflow. Finally, we developed models to detect 3 diseases: hypertrophic cardiomyopathy, cardiac amyloid, and pulmonary arterial hypertension. RESULTS: Convolutional neural networks accurately identified views (eg, 96% for parasternal long axis), including flagging partially obscured cardiac chambers, and enabled the segmentation of individual cardiac chambers. The resulting cardiac structure measurements agreed with study report values (eg, median absolute deviations of 15% to 17% of observed values for left ventricular mass, left ventricular diastolic volume, and left atrial volume). In terms of function, we computed automated ejection fraction and longitudinal strain measurements (within 2 cohorts), which agreed with commercial software-derived values (for ejection fraction, median absolute deviation=9.7% of observed, N=6407 studies; for strain, median absolute deviation=7.5%, n=419, and 9.0%, n=110) and demonstrated applicability to serial monitoring of patients with breast cancer for trastuzumab cardiotoxicity. Overall, we found automated measurements to be comparable or superior to manual measurements across 11 internal consistency metrics (eg, the correlation of left atrial and ventricular volumes). Finally, we trained convolutional neural networks to detect hypertrophic cardiomyopathy, cardiac amyloidosis, and pulmonary arterial hypertension with C statistics of 0.93, 0.87, and 0.85, respectively. CONCLUSIONS: Our pipeline lays the groundwork for using automated interpretation to support serial patient tracking and scalable analysis of millions of echocardiograms archived within healthcare systems.

Reduction in Fall Rate in Dementia Managed Care Through Video Incident Review: Pilot Study.

BACKGROUND: Falls of individuals with dementia are frequent, dangerous, and costly. Early detection and access to the history of a fall is crucial for efficient care and secondary prevention in cognitively impaired individuals. However, most falls remain unwitnessed events. Furthermore, understanding why and how a fall occurred is a challenge. Video capture and secure transmission of real-world falls thus stands as a promising assistive tool. OBJECTIVE: The objective of this study was to analyze how continuous video monitoring and review of falls of individuals with dementia can support better quality of care. METHODS: A pilot observational study (July-September 2016) was carried out in a Californian memory care facility. Falls were video-captured (24×7), thanks to 43 wall-mounted cameras (deployed in all common areas and in 10 out of 40 private bedrooms of consenting residents and families). Video review was provided to facility staff, thanks to a customized mobile device app. The outcome measures were the count of residents' falls happening in the video-covered areas, the acceptability of video recording, the analysis of video review, and video replay possibilities for care practice. RESULTS: Over 3 months, 16 falls were video-captured. A drop in fall rate was observed in the last month of the study. Acceptability was good. Video review enabled screening for the severity of falls and fall-related injuries. Video replay enabled identifying cognitive-behavioral deficiencies and environmental circumstances contributing to the fall. This allowed for secondary prevention in high-risk multi-faller individuals and for updated facility care policies regarding a safer living environment for all residents. CONCLUSIONS: Video monitoring offers high potential to support conventional care in memory care facilities.

Visual dexterity: In-hand reorientation of novel and complex object shapes.

In-hand object reorientation is necessary for performing many dexterous manipulation tasks, such as tool use in less structured environments, which remain beyond the reach of current robots. Prior works built reorientation systems assuming one or many of the following conditions: reorienting only specific objects with simple shapes, limited range of reorientation, slow or quasi-static manipulation, simulation-only results, the need for specialized and costly sensor suites, and other constraints that make the system infeasible for real-world deployment. We present a general object reorientation controller that does not make these assumptions. It uses readings from a single commodity depth camera to dynamically reorient complex and new object shapes by any rotation in real time, with the median reorientation time being close to 7 seconds. The controller was trained using reinforcement learning in simulation and evaluated in the real world on new object shapes not used for training, including the most challenging scenario of reorienting objects held in the air by a downward-facing hand that must counteract gravity during reorientation. Our hardware platform only used open-source components that cost less than 5000 dollars. Although we demonstrate the ability to overcome assumptions in prior work, there is ample scope for improving absolute performance. For instance, the challenging duck-shaped object not used for training was dropped in 56% of the trials. When it was not dropped, our controller reoriented the object within 0.4 radians (23°) 75% of the time.

Real-time video detection of falls in dementia care facility and reduced emergency care.

An artificial intelligence-enabled video fall detection system using visual science reduced emergency department visits by 80% in 6 communities over 3 months.