Mohammad Pourhomayoun

Accurate Localization of In-Body Medical Implants Based on Spatial Sparsity

Wearable and implantable wireless communication devices have in recent years gained increasing attention for medical diagnostics and therapeutics. In particular, wireless capsule endoscopy has become a popular method to visualize and diagnose the human gastrointestinal tract. Estimating the exact position of the capsule when each image is taken is a very critical issue in capsule endoscopy. Several approaches have been developed by researchers to estimate the capsule location. However, some unique challenges exist for in-body localization, such as the severe multipath issue caused by the boundaries of different organs, inconsistency of signal propagation velocity and path loss parameters inside the human body, and the regulatory restrictions on using high-bandwidth or high-power signals. In this paper, we propose a novel localization method based on spatial sparsity. We directly estimate the location of the capsule without going through the usual intermediate stage of first estimating time-of-arrival or received-signal strength, and then a second stage of estimating the location. We demonstrate the accuracy of the proposed method through extensive Monte Carlo simulations for radio frequency emission signals within the required power and bandwidth range. The results show that the proposed method is effective and accurate, even in massive multipath conditions.

A novel method for medical implant in-body localization

Wireless communication medical implants are gaining an important role in healthcare systems by controlling and transmitting the vital information of the patients. Recently, Wireless Capsule Endoscopy (WCE) has become a popular method to visualize and diagnose the human gastrointestinal (GI) tract. Estimating the exact location of the capsule when each image is taken is a very critical issue in capsule endoscopy. Most of the common capsule localization methods are based on estimating one or more location-dependent signal parameters like TOA or RSS. However, some unique challenges exist for in-body localization due to the complex nature within the human body. In this paper, we propose a novel one-stage localization method based on spatial sparsity in 3D space. In this method, we directly estimate the location of the capsule (as the emitter) without going through the intermediate stage of TOA or signal strength estimation. We evaluate the performance of the proposed method using Monte Carlo simulation with an RF signal following the allowable power and bandwidth ranges according to the standards. The results show that the proposed method is very effective and accurate even in massive multipath and shadowing conditions.

Determining the Single Best Axis for Exercise Repetition Recognition and Counting on SmartWatches

Due to the exploding costs of chronic diseasesstemming from physical inactivity, wearable sensor systems toenable remote, continuous monitoring of individuals has increasedin popularity. Many research and commercial systems exist inorder to track the activity levels of users from general dailymotion to detailed movements. This work examines this problemfrom the space of smartwatches, using the Samsung GalaxyGear, a commercial device containing an accelerometer and agyroscope, to be used in recognizing physical activity. This workalso shows the sensors and features necessary to enable suchsmartwatches to accurately count, in real-time, the repetitions offree-weight and body-weight exercises. The goal of this work isto try and select only the best single axis for each activity byextracting only the most informative activity-specific features, inorder to minimize computational load and power consumptionin repetition counting. The five activities are incorporated in aworkout routine, and knowing this information, a random forestclassifier is built with average area under the curve (AUC) of: 974, with average accuracy of 93%, in cross validation to identify eachrepetition of a given exercise using all available sensors and AUCof: 950 with accuracy of 89.9% using the single best axis foreach activity alone. Adding a gyroscope with the accelerometerincreased the average AUC from: 968 to: 974, increasing theaccuracy of specific movements as much as 2%. Results show that, while a combination of accelerometer and gyroscope provide thestrongest classification results, often times features extracted froma single, best axis are enough to accurately identify movementsfor a personal training routine, where that axis is often, but notalways, an accelerometer axis.

Recognition of Nutrition Intake Using Time-Frequency Decomposition in a Wearable Necklace Using a Piezoelectric Sensor

Food intake levels, hydration, ingestion rate, and dietary choices are all factors known to impact the risk of obesity. This paper presents a novel wearable system in the form of a necklace, which aggregates data from an embedded piezoelectric sensor capable of detecting skin motion in the lower trachea during ingestion. The skin motion produces an output voltage with varying frequencies over time. As a result, we propose an algorithm based on time-frequency decomposition, spectrogram analysis of piezoelectric sensor signals, to accurately distinguish between food types, such as liquid and solid, hot and cold drinks, and hard and soft foods. The necklace transmits data to a smartphone, which performs the processing of the signals, classifies the food type, and provides visual feedback to the user to assist the user in monitoring their eating habits over time. We compare our spectrogram analysis with other time-frequency features, such as matching pursuit and wavelets. Experimental results demonstrate promise in using time-frequency features, with high accuracy of distinguishing between food categories using spectrogram analysis and extracting key features representative of the unique swallow patterns of various foods.

Spatial sparsity based emitter localization

In classical TDOA/FDOA emitter location methods, pairs of sensors share the received data to compute the CAF and extract the ML estimates of TDOA/FDOA. The TDOA/FDOA estimates are then transmitted to a common site where they are used to estimate the emitter location. However, the two-stage method is not necessarily optimal because in the first stage of these methods, the TDOA and FDOA are estimated by ignoring the fact that all measurements should be consistent with a single emitter location. In this paper, we derive a one-stage localization method based on spatial sparsity of the grid plane. In this method, we directly estimate the location of the emitter without going through the intermediate stage of TDOA/FDOA estimation. The Monte Carlo simulation results show that the proposed method has better performance compared to two-stage classic method and also to another available one-stage method named Direct Position Determination (DPD). We will show that the proposed method is also a very effective and beneficial solution to deal with multipath scenarios.

Distributed computation for direct position determination emitter location

Classical geolocation based on time-difference-of-arrival (TDOA) and frequency-difference-of-arrival (FDOA) uses a two-stage estimation approach. The single-stage approach direct position determination (DPD) has been proposed to improve accuracy. However, unlike the classical two-stage method, the proposed DPD method does all processing at a single node. That is not desirable when computational capabilities are limited and makes the approach nonrobust to loss of the central sensor. We develop and assess several DPD variants that address these issues.

Sensor network distributed computation for Direct Position Determination

Classical localization systems based on TDOA/FDOA use a two-stage estimation approach. In the first stage, pairs of sensors share data to estimate TDOA/FDOA. Then, the extracted TDOA/FDOA measurements are used to estimate the emitter location. In some recently published methods, an optimal single-stage approach named Direct Position Determination (DPD) has been proposed to improve the position estimation accuracy. However, unlike the classical two-stage method where the TDOA/FDOA estimation can be distributed across all sensors, DPD processes all the received signals together at a single sensor node. However, when sensors have limited computational capabilities it is desirable to distribute the computation across all sensors. Furthermore, concentrating all the processing into a single node makes the location system less robust to the loss of sensors. In this paper, we develop a distributed localization method with the goal of reducing the computational load on each sensor and increasing the reliability of the system.

Context-Aware Data Processing to Enhance Quality of Measurements in Wireless Health Systems: An Application to MET Calculation of Exergaming Actions

Wireless health systems enable remote and continuous monitoring of individuals, with applications in elderly care support, chronic disease management, and preventive care. The underlying sensing platform provides constructs that consider the quality of information driven from the system and ensure the reliability/validity of the outcomes to support the decision-making processes. In this paper, we present an approach to integrate contextual information within the data processing flow in order to improve the quality of measurements. We focus on a pilot application that uses wearable motion sensors to calculate metabolic equivalent of task (MET) of exergaming movements. Exergames need to show energy expenditure values, often using accelerometer approximations applied to general activities. We focus on two contextual factors, namely “activity type” and “sensor location,” and demonstrate how these factors can be used to enhance the measured values, since allocating larger weights to more informative sensors can improve the final measurements. Further, designing regression models for each activity provides better results than any generalized model. Indeed, the averaged R2 value for the movements using simple sensor location improve from a general 0.71 to as high as 0.84 for an individual activity type. The different methods present a range of R2 value averages across activity type from 0.64 for sensor location to 0.89 for multidimensional regression, with an average game play MET value of 7.93. Finally, in a leaveone-subject-out cross validation, a mean absolute error of 2.231 METs is found when predicting the activity levels using the best models.

Multiple model analytics for adverse event prediction in remote health monitoring systems

Remote health monitoring systems (RHMS) are gaining an important role in healthcare by collecting and transmitting patient vital information and providing data analysis and medical adverse event prediction (e.g. hospital readmission prediction). Reduction in the readmission rate is typically achieved by early prediction of the readmission based on the data collected from RHMS, and then applying early intervention to prevent the readmission. Given the diversity of patient populations and the continuous nature of patient monitoring, a single static predictive model is insufficient for accurately predicting adverse events. To address this issue, we propose a multiple prediction modeling technique that includes a set of accurate prediction models rather than one single universal predictor. In this paper, we propose a novel analytics framework based on the physiological data collected from RHMS, advanced clustering algorithms and multiple-model-classification. We tested our proposed method on a subset of data collected through a remote health monitoring system from 600 Heart Failure patients. Our proposed method provides significant improvements in prediction accuracy and performance over single predictive models.

Using electronic health records to predict severity of condition for congestive heart failure patients

We propose a novel way to design an analytics engine based exclusively on electronic health records (EHR). We focus our efforts on Congestive Heart Failure (CHF) patients, although our approach could be extended to other chronic conditions. Our goal is to construct statistical models that predict a CHF patients length of stay and by extension the severity of his/her condition. We show that it is possible to predict length of hospital stay based on physiological data collected from the first day of hospitalization. Using 10-fold cross validation we achieve accurate predictions with a root mean square error of 3.3 days for hospital stays that are less than 15 days in duration. We also propose a clustering of patients that organizes them to risk groups according to their estimated severity of condition.