Efficient source separation algorithms for acoustic fall detection using a microsoft kinect

Characterizing nursing time with patients using computer vision

BACKGROUND

Compared to other providers, nurses spend more time with patients, but the exact quantity and nature of those interactions remain largely unknown. The purpose of this study was to characterize the interactions of nurses at the bedside using continuous surveillance over a year long period.

METHODS

Nurses' time and activity at the bedside were characterized using a device that integrates the use of obfuscated computer vision in combination with a Bluetooth beacon on the nurses' identification badge to track nurses' activities at the bedside. The surveillance device (AUGi) was installed over 37 patient beds in two medical/surgical units in a major urban hospital. Forty-nine nurse users were tracked using the beacon. Data were collected 4/15/19-3/15/20. Statistics were performed to describe nurses' time and activity at the bedside.

RESULTS

A total of n = 408,588 interactions were analyzed over 670 shifts, with >1.5 times more interactions during day shifts (n = 247,273) compared to night shifts (n = 161,315); the mean interaction time was 3.34 s longer during nights than days (p < 0.0001). Each nurse had an average of 7.86 (standard deviation [SD] = 10.13) interactions per bed each shift and a mean total interaction time per bed of 9.39 min (SD = 14.16). On average, nurses covered 7.43 beds (SD = 4.03) per shift (day: mean = 7.80 beds/nurse/shift, SD = 3.87; night: mean = 7.07/nurse/shift, SD = 4.17). The mean time per hourly rounding (HR) was 69.5 s (SD = 98.07) and 50.1 s (SD = 56.58) for bedside shift report.

DISCUSSION

As far as we are aware, this is the first study to provide continuous surveillance of nurse activities at the bedside over a year long period, 24 h/day, 7 days/week. We detected that nurses spend less than 1 min giving report at the bedside, and this is only completed 20.7% of the time. Additionally, hourly rounding was completed only 52.9% of the time and nurses spent only 9 min total with each patient per shift. Further study is needed to detect whether there is an optimal timing or duration of interactions to improve patient outcomes.

CLINICAL RELEVANCE

Nursing time with the patient has been shown to improve patient outcomes but precise information about how much time nurses spend with patients has been heretofore unknown. By understanding minute-by-minute activities at the bedside over a full year, we provide a full picture of nursing activity; this can be used in the future to determine how these activities affect patient outcomes.

Comparing fall detection methods in people with multiple sclerosis: A prospective observational cohort study

Background Falls occur across the population but are more common, and have more negative sequelae, in people with multiple sclerosis (MS). Given the prevalence and impact of falls, accurate measures of fall frequency are needed. This study compares the sensitivity and false discovery rates of three methods of fall detection: the current gold standard, prospective paper fall calendars, real-time self-reporting and automated detection, the latter two from a novel body-worn device. Methods Falls in twenty-five people with MS were recorded for eight weeks with prospective fall calendars, real-time body-worn self-report, and an automated body-worn detector concurrently. Eligible individuals were adults with MS enrolled in a randomized controlled trial of a fall prevention intervention. Entry criteria were at least two falls or near-falls in the previous two months, Expanded Disability Status Scale ≤ 6.0, community dwelling, and no MS relapse in the previous month. The sensitivity (proportion of true falls detected) and false discovery rates (proportion of false reports generated) of the fall detection methods were compared. A true fall was a fall reported by at least two methods. A false report was a fall reported by only one method. The trial is registered on ClinicalTrials.gov (NCT02583386) and is closed. Results In the 1,276 person-days of fall counting with all three methods in use simultaneously there were 1344 unique fall events. Of these, 8.5% (114) were true falls and 91.5% (1230) were false reports. Fall calendars had the lowest sensitivity (0.614) and the lowest false discovery rate (0.067). The automated detector had the highest sensitivity (0.921) and the highest false discovery rate (0.919). All methods generated under one false report per day. There were no fall detection-related adverse events. Conclusion Fall calendars likely underestimate fall frequency by around 40%. The automated detector evaluated here misses very few falls but likely overestimates the number of falls by around one fall per day. Additional research is needed to produce an ideal fall detection and counting method for use in clinical and research applications. Funding United States Department of Veterans Affairs, Rehabilitations Research and Development Service.

Automated Detection of Real-World Falls: Modeled From People With Multiple Sclerosis

Falls are a major health problem with one in three people over the age of 65 falling each year, oftentimes causing hip fractures, disability, reduced mobility, hospitalization and death. A major limitation in fall detection algorithm development is an absence of real-world falls data. Fall detection algorithms are typically trained on simulated fall data that contain a well-balanced number of examples of falls and activities of daily living. However, real-world falls occur infrequently, making them difficult to capture and causing severe data imbalance. People with multiple sclerosis (MS) fall frequently, and their risk of falling increases with disease progression. Because of their high fall incidence, people with MS provide an ideal model for studying falls. This paper describes the development of a context-aware fall detection system based on inertial sensors and time of flight sensors that is robust to imbalance, which is trained and evaluated on real-world falls in people with MS. The algorithm uses an auto-encoder that detects fall candidates using reconstruction error of accelerometer signals followed by a hyper-ensemble of balanced random forests trained using both acceleration and movement features. On a clinical dataset obtained from 25 people with MS monitored over eight weeks during free-living conditions, 54 falls were observed and our system achieved a sensitivity of 92.14%, and false-positive rate of 0.65 false alarms per day.

Innovative Head-Mounted System Based on Inertial Sensors and Magnetometer for Detecting Falling Movements

This work presents a fall detection system that is worn on the head, where the acceleration and posture are stable such that everyday movement can be identified without disturbing the wearer. Falling movements are recognized by comparing the acceleration and orientation of a wearer's head using prespecified thresholds. The proposed system consists of a triaxial accelerometer, gyroscope, and magnetometer; as such, a Madgwick's filter is adopted to improve the accuracy of the estimation of orientation. Moreover, with its integrated Wi-Fi module, the proposed system can notify an emergency contact in a timely manner to provide help for the falling person. Based on experimental results concerning falling movements and activities of daily living, the proposed system achieved a sensitivity of 96.67% in fall detection, with a specificity of 98.27%, and, therefore, is suitable for detecting falling movements in daily life.

Recovering Hidden Diagonal Structures via Non-Negative Matrix Factorization with Multiple Constraints

Revealing data with intrinsically diagonal block structures is particularly useful for analyzing groups of highly correlated variables. Earlier researches based on non-negative matrix factorization (NMF) have been shown to be effective in representing such data by decomposing the observed data into two factors, where one factor is considered to be the feature and the other the expansion loading from a linear algebra perspective. If the data are sampled from multiple independent subspaces, the loading factor would possess a diagonal structure under an ideal matrix decomposition. However, the standard NMF method and its variants have not been reported to exploit this type of data via direct estimation. To address this issue, a non-negative matrix factorization with multiple constraints model is proposed in this paper. The constraints include an sparsity norm on the feature matrix and a total variational norm on each column of the loading matrix. The proposed model is shown to be capable of efficiently recovering diagonal block structures hidden in observed samples. An efficient numerical algorithm using the alternating direction method of multipliers model is proposed for optimizing the new model. Compared with several benchmark models, the proposed method performs robustly and effectively for simulated and real biological data.

Testing non-wearable fall detection methods in the homes of older adults

In this paper, we describe two longitudinal studies in which fall detection sensor technology was tested in the homes of older adults. The first study tested Doppler radar, a two-webcam system, and a depth camera system in ten apartments for two years. This continuous data collection allowed us to investigate the real-world setting of target users and compare the advantages and limitations of each sensor modality. Based on this study, the depth camera was chosen for a current ongoing study in which depth camera systems have been installed in 94 additional older adult apartments. We include a discussion of the different sensor systems, the pros and cons of each, and results of the fall detection and false alarms in the older adult homes.

A Combined One-Class SVM and Template-Matching Approach for User-Aided Human Fall Detection by Means of Floor Acoustic Features

The primary cause of injury-related death for the elders is represented by falls. The scientific community devoted them particular attention, since injuries can be limited by an early detection of the event. The solution proposed in this paper is based on a combined One-Class SVM (OCSVM) and template-matching classifier that discriminate human falls from nonfalls in a semisupervised framework. Acoustic signals are captured by means of a Floor Acoustic Sensor; then Mel-Frequency Cepstral Coefficients and Gaussian Mean Supervectors (GMSs) are extracted for the fall/nonfall discrimination. Here we propose a single-sensor two-stage user-aided approach: in the first stage, the OCSVM detects abnormal acoustic events. In the second, the template-matching classifier produces the final decision exploiting a set of template GMSs related to the events marked as false positives by the user. The performance of the algorithm has been evaluated on a corpus containing human falls and nonfall sounds. Compared to the OCSVM only approach, the proposed algorithm improves the performance by 10.14% in clean conditions and 4.84% in noisy conditions. Compared to Popescu and Mahnot (2009) the performance improvement is 19.96% in clean conditions and 8.08% in noisy conditions.

Wearable technology and ECG processing for fall risk assessment, prevention and detection

Falls represent one of the most common causes of injury-related morbidity and mortality in later life. Subjects with cardiovascular disorders (e.g., related to autonomic dysfunctions and postural hypotension) are at higher risk of falling. Autonomic dysfunctions increasing the risk of falling in the short and mid-term could be assessed by Heart Rate Variability (HRV) extracted by electrocardiograph (ECG). We developed three trials for assessing the usefulness of ECG monitoring using wearable devices for: risk assessment of falling in the next few weeks; prevention of imminent falls due to standing hypotension; and fall detection. Statistical and data-mining methods are adopted to develop classification and regression models, validated with the cross-validation approach. The first classifier based on HRV features enabled to identify future fallers among hypertensive patients with an accuracy of 72% (sensitivity: 51.1%, specificity: 80.2%). The regression model to predict falls due to orthostatic dropdown from HRV recorded before standing achieved an overall accuracy of 80% (sensitivity: 92%, specificity: 90%). Finally, the classifier to detect simulated falls using ECG achieved an accuracy of 77.3% (sensitivity: 81.8%, specificity: 72.7%). The evidence from these three studies showed that ECG monitoring and processing could achieve satisfactory performances compared to other system for risk assessment, fall prevention and detection. This is interesting as differently from other technologies actually employed to prevent falls, ECG is recommended for many other pathologies of later life and is more accepted by senior citizens.

Fall Prediction in Hypertensive Patients via Short-Term HRV Analysis

Falls are a major problem of later life having severe consequences on quality of life and a significant burden in occidental countries. Many technological solutions have been proposed to assess the risk or to predict falls and the majority is based on accelerometers and gyroscopes. However, very little was done for identifying first time fallers, which are very difficult to recognize. This paper presents a metamodel predicting falls using short term Heart Rate Variability (HRV) analysis acquired at the baseline. About 170 hypertensive patients (age: 72 ± 8 years, 56 female) were investigated, of which 34 fell once in the 3 months after the baseline assessment. This study is focused on hypertensive patients, which were considered as convenient pragmatic sample, as they undergo regular outpatient visits, during which short term Electrocardiogram (ECG) can be easily recorded without significant increase of healthcare costs. For each subject, 11 consecutive excerpts of 5 min each (55 min) were extracted from ECGs recorded between 10:30 and 12:30 and analysed. Linear and nonlinear HRV features were extracted and averaged among the 11 excerpts, which were, then, considered for the statistical and data mining analysis. The best predictive metamodel was based on Multinomial Naïve Bayes, which enabled to predict first-time fallers with sensitivity, specificity, and accuracy rates of 72%, 61%, and 68%, respectively.

Short term Heart Rate Variability to predict blood pressure drops due to standing: a pilot study

Background

Standing from a bed or chair may cause a significant lowering of blood pressure (ΔBP), which may have severe consequences such as, for example, falls in older subjects. The goal of this study was to develop a mathematical model to predict the ΔBP due to standing in healthy subjects, based on their Heart Rate Variability, recorded in the 5 minutes before standing.

Methods

Heart Rate Variability was extracted from an electrocardiogram, recorded from 10 healthy subjects during the 5 minutes before standing. The blood pressure value was measured before and after rising. A mathematical model aiming to predict ΔBP based on Heart Rate Variability measurements was developed using a robust multi-linear regression and was validated with the leave-one-subject-out cross-validation technique.

Results

The model predicted correctly the ΔBP in 80% of experiments, with an error below the measurement error of sphygmomanometer digital devices (±4.5 mmHg), a false negative rate of 7.5% and a false positive rate of 10%. The magnitude of the ΔBP was associated with a depressed and less chaotic Heart Rate Variability pattern.

Conclusions

The present study showes that blood pressure lowering due to standing can be predicted by monitoring the Heart Rate Variability in the 5 minutes before standing.

Unified framework for triaxial accelerometer-based fall event detection and classification using cumulants and hierarchical decision tree classifier

In this Letter, the authors present a unified framework for fall event detection and classification using the cumulants extracted from the acceleration (ACC) signals acquired using a single waist-mounted triaxial accelerometer. The main objective of this Letter is to find suitable representative cumulants and classifiers in effectively detecting and classifying different types of fall and non-fall events. It was discovered that the first level of the proposed hierarchical decision tree algorithm implements fall detection using fifth-order cumulants and support vector machine (SVM) classifier. In the second level, the fall event classification algorithm uses the fifth-order cumulants and SVM. Finally, human activity classification is performed using the second-order cumulants and SVM. The detection and classification results are compared with those of the decision tree, naive Bayes, multilayer perceptron and SVM classifiers with different types of time-domain features including the second-, third-, fourth- and fifth-order cumulants and the signal magnitude vector and signal magnitude area. The experimental results demonstrate that the second- and fifth-order cumulant features and SVM classifier can achieve optimal detection and classification rates of above 95%, as well as the lowest false alarm rate of 1.03%.

Automatic classifier based on heart rate variability to identify fallers among hypertensive subjects

Accidental falls are a major problem of later life. Different technologies to predict falls have been investigated, but with limited success, mainly because of low specificity due to a high false positive rate. This Letter presents an automatic classifier based on heart rate variability (HRV) analysis with the goal to identify fallers automatically. HRV was used in this study as it is considered a good estimator of autonomic nervous system (ANS) states, which are responsible, among other things, for human balance control. Nominal 24 h electrocardiogram recordings from 168 cardiac patients (age 72 ± 8 years, 60 female), of which 47 were fallers, were investigated. Linear and nonlinear HRV properties were analysed in 30 min excerpts. Different data mining approaches were adopted and their performances were compared with a subject-based receiver operating characteristic analysis. The best performance was achieved by a hybrid algorithm, RUSBoost, integrated with feature selection method based on principal component analysis, which achieved satisfactory specificity and accuracy (80 and 72%, respectively), but low sensitivity (51%). These results suggested that ANS states causing falls could be reliably detected, but also that not all the falls were due to ANS states.

Automated Health Alerts Using In-Home Sensor Data for Embedded Health Assessment

We present an example of unobtrusive, continuous monitoring in the home for the purpose of assessing early health changes. Sensors embedded in the environment capture behavior and activity patterns. Changes in patterns are detected as potential signs of changing health. We first present results of a preliminary study investigating 22 features extracted from in-home sensor data. A 1-D alert algorithm was then implemented to generate health alerts to clinicians in a senior housing facility. Clinicians analyze each alert and provide a rating on the clinical relevance. These ratings are then used as ground truth for training and testing classifiers. Here, we present the methodology for four classification approaches that fuse multisensor data. Results are shown using embedded sensor data and health alert ratings collected on 21 seniors over nine months. The best results show similar performance for two techniques, where one approach uses only domain knowledge and the second uses supervised learning for training. Finally, we propose a health change detection model based on these results and clinical expertise. The system of in-home sensors and algorithms for automated health alerts provides a method for detecting health problems very early so that early treatment is possible. This method of passive in-home sensing alleviates compliance issues.