Megan Holtzman

Relative Thresholding With Under-Mattress Pressure Sensors to Detect Central Apnea

Unobtrusive pressure sensors can be used for biological monitoring and long-term health assessment in smart homes. The challenge in detecting events from smart home data is that people have different mattresses, unlike in hospitals where bedding is standardized. This paper proposes to model central apneas using an under-mattress pressure sensor as a measuring instrument. The model uses three parameters, namely, a relative threshold and two time lengths, applied to a moving variance signal. The use of a relative threshold allows apneas to be detected under a variety of different conditions and improves results compared to hard-coded thresholds. The algorithm developed herein was applied to simulated apneas collected from pressure sensors placed under nine different mattresses. The parameters determined from the training set were applied to the test set and produced classification results of 0.78 positive predictive value (PPV) if the bed occupants position is known and 0.75 PPV if the position is unknown. The use of the relative threshold approach overcomes the variability in mattress types found in smart homes.

Validation of pressure sensors for physiological monitoring in home environments

Previous research using pressure sensors to monitor physiological signs, such as breathing and heart rate, has often been in controlled laboratory or institutional settings. We are interested in validating the use of such sensors in diverse home environments, where unobtrusiveness is achieved by embedding pressure sensors inside or below furniture.We have quantitatively examined outputs from sensors placed below a variety of mattress types and compared the response to outputs from sensors placed on top of the mattress. We show that embedded pressure sensors are valid for use in home environments and could be reliably used to collect physiological data. However, resultant low signal levels may require more sophisticated signal extraction algorithms.

Breathing Signal Fusion in Pressure Sensor Arrays

Pressure sensors can be used to unobtrusively obtain breathing signals from a person in bed. Obtaining a single representation of the breathing signal from an array of such sensors requires data-level fusion. We propose a decision directed adaptive linear estimator to perform this fusion online. The proposed method was compared with three other online fusion methods and two offline methods using one hundred data records collected from five healthy participants. The decision directed adaptive linear estimator had signal to noise ratios comparable to the offline correlation method that it was adapted from and better mutual information results. In the presence of movement noise and for low amplitude signals, the proposed method also provides good fusion performance.

Force Estimation with a Non-Uniform Pressure Sensor Array

Embedding pressure sensors into household fixtures enables unobtrusive occupant health and safety monitoring at home. To monitor bathroom grab bar use, the force applied to a grab bar is desired from the output of three embedded pressure sensors. We examine the measurement of applied force in a non-uniform pressure sensor array, where forces are distributed with spatial nonlinearity to the pressure sensors. Two methods that ignore the spatial nonlinearities are compared to two methods that incorporate them. These include a polynomial response curve, a theoretical model, a lookup table, and an artificial neural network. When many calibration points can be taken and location estimates are accurate, the location-based lookup table presented the lowest error. However, when calibration time is limited, the theoretical model performs best, while an artificial neural network is preferred when location inputs are inaccurate.

Contact Location Estimation from a Nonlinear Array of Pressure Sensors

In physical medicine and rehabilitation, it is important to be able to collect information regarding a patients behavior and range of mobility throughout their daily activities. Grab bars are used widely in the homes of individuals with mobility impairments so their usage while performing physical tasks can provide valuable information as to the individuals physical status. This paper explores the extraction of location information for forces applied to a grab bar embedded with a nonlinear pressure sensor array of low spatial resolution. It first describes the instrumentation of the grab bar and the calibration procedure. It then investigates three methods of estimating the contact location; a simple centroid, a percentage-based lookup table and an artificial neural network. Results of the three methods are reported based on data collected from different forces and contact locations applied along the bar. The artificial neural network proves to be the most successful method of estimating the points of contact, by most accurately modeling the nonlinearities in the system.

Simulated central apnea detection using the pressure variance

This paper presents use of an unobtrusive pressure sensor array for simulated central apnea detection. Data was collected from seven volunteers who performed a series of regular breathing and breath holding exercises to simulate central apneas. Results of the feature extraction from the breathing signals show that breathing events may be differentiated with epoch based variance calculations. Two approaches were considered: the single sensor approach and the multisensor vote approach. The multisensor vote approach can decrease false positives and increase the value of Matthew’s Correlation Coefficient. The effect of lying position on correct classification was investigated by modifying the multisensor vote approach to reduce false positives segments caused by the balistocardiogram signal and as such increase sensitivity while maintaining a low false positive rate. Intersubject classification results had low variability in both approaches.

Highly survivable bed pressure mat remote patient monitoring system for mHealth

The high speed mobile networks like 4G and beyond are making a ubiquitous remote patient monitoring (RPM) system using multiple sensors and wireless sensor networks a realistic possibility. The high speed wireless RPM system will be an integral part of the mobile health (mHealth) paradigm reducing cost and providing better service to the patients. While the high speed wireless RPM system will allow clinicians to monitor various chronic and acute medical conditions, the reliability of such system will depend on the network Quality of Service (QoS). The RPM system needs to be resilient to temporary reduced network QoS. This paper presents a highly survivable bed pressure mat RPM system design using an adaptive information content management methodology for the monitored sensor data. The proposed design improves the resiliency of the RPM system under adverse network conditions like congestion and/or temporary loss of connectivity. It also shows how the proposed RPM system can reduce the information rate and correspondingly reduce the data transfer rate by a factor of 5.5 and 144 to address temporary network congestion. The RPM system data rate reduction results in a lower specificity and sensitivity for the features being monitored but increases the survivability of the system from 1 second to 2.4 minutes making it highly robust.

Analysis of commode grab bar usage for the monitoring of older adults in the smart home environment

The occurrence of falls inside the home is a common yet potentially hazardous issue for adults as they age. Even with the installation of physical aids such as grab bars, weight transfers on and off a toilet or bathtub can become increasingly difficult as a person’s level of physical mobility and sense of balance deteriorate. Detecting this deterioration becomes an important goal in fall prevention within a smart home. This paper develops an unobtrusive method of analyzing the usage of toilet grab bars using pressure sensors embedded into the arm rests of a commode. Clinical parameters are successfully extracted automatically from a series of stand-to-sit (StSi) and sit-to-stand (SiSt) transfers performed by a trial group of young and older adults. A preliminary comparison of the parameters indicates differences between the two groups, and aligns well with published characteristics obtained using accelerometers worn on the body. The unobtrusive nature of this method provides a useful tool to be incorporated into a system of continuous monitoring of older adults within the smart home environment.

Breathing sensor selection during movement

A pressure sensor array placed below a mattress can be used to estimate the breathing effort signal unobtrusively. When multiple breathing effort sensor outputs are available, there is sometimes a need to choose the sensor with the best approximation of the actual breathing effort. Previous work with pressure sensor arrays placed on top of or under mattresses used for respiration rate and breathing signal estimation have used either the amplitude or the power spectrum to choose the most representative sensor. These methods are both useful when the subject is still; however, pressure sensor signals also contain movement. We propose and test a spectral ratio method for selection in the presence of movement. The spectral ratio method is good at finding strong breathing signals and at discriminating movement signals from strong breathing signals. This method provides a mean correlation to respiration bands that is 4% higher than the next best method during small movements and 14% higher during larger movements.

Effect of windowing on central apnea detection

The detection of central apneas using an unobtrusive pressure sensor array installed in the beds of smart homes could allow comfortable diagnosis of sleep disturbances. To improve central apnea detection, two methods of improving the results of apneas classified by a previously developed method are presented: moving average windowing and window elimination. The first improved classifier sensitivity, while the second improved the specificity with better duration estimation. However, it was slightly more likely to miss apnea segments altogether.