Bias Analysis in Healthcare Time Series (BAHT) Decision Support Systems from Meta Data.

One of the hindrances in the widespread acceptance of deep learning-based decision support systems in healthcare is bias. Bias in its many forms occurs in the datasets used to train and test deep learning models and is amplified when deployed in the real world, leading to challenges such as model drift. Recent advancements in the field of deep learning have led to the deployment of deployable automated healthcare diagnosis decision support systems at hospitals as well as tele-medicine through IoT devices. Research has been focused primarily on the development and improvement of these systems leaving a gap in the analysis of the fairness. The domain of FAccT ML (fairness, accountability, and transparency) accounts for the analysis of these deployable machine learning systems. In this work, we present a framework for bias analysis in healthcare time series (BAHT) signals such as electrocardiogram (ECG) and electroencephalogram (EEG). BAHT provides a graphical interpretive analysis of bias in the training, testing datasets in terms of protected variables, and analysis of bias amplification by the trained supervised learning model for time series healthcare decision support systems. We thoroughly investigate three prominent time series ECG and EEG healthcare datasets used for model training and research. We show the extensive presence of bias in the datasets leads to potentially biased or unfair machine-learning models. Our experiments also demonstrate the amplification of identified bias with an observed maximum of 66.66%. We investigate the effect of model drift due to unanalyzed bias in datasets and algorithms. Bias mitigation though prudent is a nascent area of research. We present experiments and analyze the most prevalently accepted bias mitigation strategies of under-sampling, oversampling, and the use of synthetic data for balancing the dataset through augmentation. It is important that healthcare models, datasets, and bias mitigation strategies should be properly analyzed for a fair unbiased delivery of service.

A novel mixed reality system to manage phantom pain in-home: results of a pilot clinical trial.

Introduction: Mirror therapy for phantom limb pain (PLP) is a well-accepted treatment method that allows participants to use a mirror to visually perceive the missing limb. Mixed reality options are now becoming increasingly available, but an in-home virtual mirror therapy option has yet to be adequately investigated. Methods: We had previously developed a mixed reality system for Managing Phantom Pain (Mr. MAPP) that registers the intact limb and mirrors it onto the amputated limb with the system's visual field, allowing the user to engage with interactive games targeting different large lower limb movements. Feasibility and pilot outcomes of treating patients with lower extremity PLP by using Mr. MAPP at home for 1 month were evaluated in this study. Pain intensity and interference were assessed using the McGill Pain Questionnaire, Brief Pain Inventory, and a daily exercise diary. Function was assessed using the Patient Specific Functional Scale (PSFS). The clinical trial registry number for this study is NCT04529083. Results: This pilot study showed that it was feasible for patients with PLP to use Mr. MAPP at home. Among pilot clinical outcomes, statistically significant differences were noted in mean current pain intensity [1.75 (SD = 0.46) to 1.125 (SD = 0.35) out of 5, P = .011] and PSFS goal scores [4.28 (SD = 2.27) to 6.22 (SD = 2.58) out of 10, P = .006], with other outcome measures showing non-significant trends towards improvement. Discussion: This pilot study revealed that in-home use of Mr. MAPP has potential to provide pain relief and improve function in patients with lower extremity PLP and is feasible. Each scale used provided unique perspective on the functional impact of PLP. Further expanded studies and investigation, including a fully powered clinical trial, with these scales are warranted. Clinical Trial Registration: https://www.clinicaltrials.gov/ct2/show/NCT04529083, Identifier: NCT04529083.

Personalized 3D exergames for in-home rehabilitation after stroke: a pilot study.

PURPOSE: To describe a novel 3-dimensional (3D) exergames system and the results of a clinical feasibility study of stroke survivors needing in-home rehabilitation. MATERIALS AND METHODS: The customisable Personalized In-home eXErgames for Rehabilitation (PIXER) system captures the user's image, generates a live model, and incorporates it into a virtual exergame. PIXER provides a recording system for home exercise programs (HEPs) by adapting virtual objects, customizes the exergame and creates a digital diary. Ten persons with stroke, performed HEPs with PIXER for 1 month, and without PIXER for 2 additional months. In-game performance data, measures of physical functioning (PF) including Stroke Impact Scale (SIS), Timed Up & Go (TUG) and Goal Attainment (GA) Scale obtained at baseline, 1- and 3 months were evaluated. RESULTS: Seventy percent of participants completed the 1-month timepoint, 50% completed all timepoints. In-game data: Number of repetitions completed; Anomalies reported; and Percentage of bubbles hit showed positive trends. Compared to baseline, all SIS physical functioning (PF) scores were higher at 1 month, TUG scores showed no overall improvement and GA scale scores were 77% at 3 months. CONCLUSION: It is feasible for community-dwelling patients to perform HEP after stroke using PIXER, a novel, exergames system, and potentially improve their function.IMPLICATIONS FOR REHABILITATIONHome Exercises performed using a novel, 3-dimensional, customizable Personalized In-home eXErgames for Rehabilitation (PIXER) system is feasible for community-dwelling patients after stroke.In-game performance data obtained in this clinical pilot study showed positive trends of improvement in several study participants.PIXER has potential to improve functional outcomes for community-dwelling adults with stroke.

Comparison of in-person and synchronous remote musculoskeletal exam using augmented reality and haptics: A pilot study.

INTRODUCTION: Utilization of telemedicine for health care delivery increased rapidly during the coronavirus disease 2019 (COVID-19) pandemic. However, physical examination during telehealth visits remains limited. A novel telerehabilitation system-The Augmented Reality-based Telerehabilitation System with Haptics (ARTESH)-shows promise for performing synchronous, remote musculoskeletal examination. OBJECTIVE: To assess the potential of ARTESH in remotely examining upper extremity passive range of motion (PROM) and maximum isometric strength (MIS). DESIGN: In this cross-sectional pilot study, we compared the in-person (reference standard) and remote evaluations (ARTESH) of participants' upper extremity PROM and MIS in 10 shoulder and arm movements. The evaluators were blinded to each other's results. SETTING: Participants underwent in-person evaluations at a Veterans Affairs hospital's outpatient Physical Medicine and Rehabilitation (PM&R) clinic, and underwent remote examination using ARTESH with the evaluator located at a research lab 30 miles away, connected via a high-speed network. PATIENTS: Fifteen participants with upper extremity pain and/or weakness. INTERVENTIONS: Not applicable. MAIN OUTCOME MEASURES: Inter-rater agreement between in-person and remote evaluations on 10 PROM and MIS movements and presence/absence of pain with movement was calculated. RESULTS: The highest inter-rater agreements were noted in shoulder abduction and protraction PROM (kappa (κ) = 0.44, confidence interval (CI): -0.1 to 1.0), and in elbow flexion, shoulder abduction, and shoulder protraction MIS (κ = 0.63, CI: 0 to 1.0). CONCLUSIONS: This pilot study suggests that synchronous tele-physical examination using the ARTESH system with augmented reality and haptics has the potential to provide enhanced value to existing telemedicine platforms. With the additional technological and procedural improvements and with an adequately powered study, the accuracy of ARTESH-enabled remote tele-physical examinations can be better evaluated.

Clinical feasibility and preliminary outcomes of a novel mixed reality system to manage phantom pain: a pilot study.

BACKGROUND: To assess the clinical feasibility of a virtual mirror therapy system in a pilot sample of patients with phantom pain. METHODS: Our Mixed reality system for Managing Phantom Pain (Mr. MAPP) mirrors the preserved limb to visualize the amputated limb virtually and perform exercises. Seven patients with limb loss and phantom pain agreed to participate and received the system for 1-month home use. Outcome measures were collected at baseline and 1 month. RESULTS: Four (of seven recruited) participants completed the study, which was temporarily suspended due to COVID-19 restrictions. At 1 month, in-game data showed a positive trend, but pain scores showed no clear trends. Functioning scores improved for 1 participant. CONCLUSIONS: Mr. MAPP is feasible and has the potential to improve pain and function in patients with phantom pain. TRIAL REGISTRATION: Clinical Trials Registration, NCT04529083.

CEFEs: A CNN Explainable Framework for ECG Signals.

In the healthcare domain, trust, confidence, and functional understanding are critical for decision support systems, therefore, presenting challenges in the prevalent use of black-box deep learning (DL) models. With recent advances in deep learning methods for classification tasks, there is an increased use of deep learning in healthcare decision support systems, such as detection and classification of abnormal Electrocardiogram (ECG) signals. Domain experts seek to understand the functional mechanism of black-box models with an emphasis on understanding how these models arrive at specific classification of patient medical data. In this paper, we focus on ECG data as the healthcare data signal to be analyzed. Since ECG is a one-dimensional time-series data, we target 1D-CNN (Convolutional Neural Networks) as the candidate DL model. Majority of existing interpretation and explanations research has been on 2D-CNN models in non-medical domain leaving a gap in terms of explanation of CNN models used on medical time-series data. Hence, we propose a modular framework, CNN Explanations Framework for ECG Signals (CEFEs), for interpretable explanations. Each module of CEFEs provides users with the functional understanding of the underlying CNN models in terms of data descriptive statistics, feature visualization, feature detection, and feature mapping. The modules evaluate a model's capacity while inherently accounting for correlation between learned features and raw signals which translates to correlation between model's capacity to classify and it's learned features. Explainable models such as CEFEs could be evaluated in different ways: training one deep learning architecture on different volumes/amounts of the same dataset, training different architectures on the same data set or a combination of different CNN architectures and datasets. In this paper, we choose to evaluate CEFEs extensively by training on different volumes of datasets with the same CNN architecture. The CEFEs' interpretations, in terms of quantifiable metrics, feature visualization, provide explanation as to the quality of the deep learning model where traditional performance metrics (such as precision, recall, accuracy, etc.) do not suffice.

Using Biometric Technology for Telehealth and Telerehabilitation.

This article discusses the use of physical and biometric sensors in telerehabilitation. It also discusses synchronous tele-physical assessment using haptics and augmented reality and asynchronous physical assessment using remote pose estimation. The article additionally focuses on computational models that have the potential to monitor and evaluate changes in kinematic and kinetic properties during telerehabilitation using biometric sensors such as electromyography and other wearable and noncontact sensors based on force and speed. And finally, the article discusses how virtual reality environments can be facilitated in telerehabilitation.

Usability of an Immersive Augmented Reality Based Telerehabilitation System with Haptics (ARTESH) for Synchronous Remote Musculoskeletal Examination.

This study describes the features and utility of a novel augmented reality based telemedicine system with haptics that allows the sense of touch and direct physical examination during a synchronous immersive telemedicine consultation and physical examination. The system employs novel engineering features: (a) a new force enhancement algorithm to improve force rendering and overcoming the "just-noticeable-difference" limitation; (b) an improved force compensation method to reduce the delay in force rendering; (c) use of the "haptic interface point" to reduce disparity between the visual and haptic data; and (d) implementation of efficient algorithms to process, compress, decompress, transmit and render 3-D tele-immersion data. A qualitative pilot study (n=20) evaluated the usability of the system. Users rated the system on a 26-question survey using a seven-point Likert scale, with percent agreement calculated from the total users who agreed with a given statement. Survey questions fell into three main categories: (1) ease and simplicity of use, (2) quality of experience, and (3) comparison to in-person evaluation. Average percent agreements between the telemedicine and in-person evaluation were highest for ease and simplicity of use (86%) and quality of experience (85%), followed by comparison to in-person evaluation (58%). Eighty-nine percent (89%) of respondents expressed satisfaction with the overall quality of experience. Results suggest that the system was effective at conveying audio-visual and touch data in real-time across 20.3 miles, and warrants further development.

Motion Capture With Ellipsoidal Skeleton Using Multiple Depth Cameras.

This paper introduces a novel motion capturing framework which works by minimizing the fitting error between an ellipsoid based skeleton and the input point cloud data captured by multiple depth cameras. The novelty of this method comes from that it uses the ellipsoids equipped with the spherical harmonics encoded displacement and normal functions to capture the geometry details of the tracked object. This method is also integrated with a mechanism to avoid collisions of bones during the motion capturing process. The method is implemented parallelly with CUDA on GPU and has a fast running speed without dedicated code optimization. The errors of the proposed method on the data from Berkeley Multimodal Human Action Database (MHAD) are within a reasonable range compared with the ground truth results. Our experiment shows that this method succeeds on many challenging motions which are failed to be reported by Microsoft Kinect SDK and not tested by existing works. In the comparison with the state-of-art marker-less depth camera based motion tracking work our method shows advantages in both robustness and input data modality.

Discovering Multidimensional Motifs in Physiological Signals for Personalized Healthcare.

Personalized diagnosis and therapy requires monitoring patient activity using various body sensors. Sensor data generated during personalized exercises or tasks may be too specific or inadequate to be evaluated using supervised methods such as classification. We propose multidimensional motif (MDM) discovery as a means for patient activity monitoring, since such motifs can capture repeating patterns across multiple dimensions of the data, and can serve as conformance indicators. Previous studies pertaining to mining MDMs have proposed approaches that lack the capability of concurrently processing multiple dimensions, thus limiting their utility in online scenarios. In this paper, we propose an efficient real-time approach to MDM discovery in body sensor generated time series data for monitoring performance of patients during therapy. We present two alternative models for MDMs based on motif co-occurrences and temporal ordering among motifs across multiple dimensions, with detailed formulation of the concepts proposed. The proposed method uses an efficient hashing based record to enable speedy update and retrieval of motif sets, and identification of MDMs. Performance evaluation using synthetic and real body sensor data in unsupervised motif discovery tasks shows that the approach is effective for (a) concurrent processing of multidimensional time series information suitable for real-time applications, (b) finding unknown naturally occurring patterns with minimal delay, and

Distributed haptic interactions with physically based 3D deformable models over lossy networks.

Researchers have faced great challenges when simulating complicated 3D volumetric deformable models in haptics-enabled collaborative/cooperative virtual environments (HCVEs) due to the expensive simulation cost, heavy communication load, and unstable network conditions. When general network services are applied to HCVEs, network problems such as packet loss, delay, and jitter can cause severe visual distortion, haptic instability, and system inconsistency. In this paper, we propose a novel approach to support haptic interactions with physically based 3D deformable models in a distributed virtual environment. Our objective is to achieve real-time sharing of deformable and force simulations over general networks. Combining linear modal analysis and corotational methods, we can effectively simulate physical behaviors of 3D objects, even for large rotational deformations. We analyze different factors that influence HCVEs' performance and focus on exploring solutions for streaming over lossy networks. In our system, 3D deformation can be described by a fairly small amount of data (several KB) using accelerations in the spectral domain, so that we can achieve low communication load and effective streaming. We develop a loss compensation and prediction algorithm to correct the errors/distortions caused by network problem, and a force prediction method to simulate force at users' side to ensure the haptic stability, and the visual and haptic consistency. Our system works well under both the client-server and the peer-to-peer distribution structures, and can be easily extended to other topologies. In addition to theoretical analysis, we have tested the proposed system and algorithms under various network conditions. The experimental results are remarkably good, confirming the effectiveness, robustness, and validity of our approach.

Point-based manifold harmonics.

This paper proposes an algorithm to build a set of orthogonal Point-Based Manifold Harmonic Bases (PB-MHB) for spectral analysis over point-sampled manifold surfaces. To ensure that PB-MHB are orthogonal to each other, it is necessary to have symmetrizable discrete Laplace-Beltrami Operator (LBO) over the surfaces. Existing converging discrete LBO for point clouds, as proposed by Belkin et al., is not guaranteed to be symmetrizable. We build a new point-wisely discrete LBO over the point-sampled surface that is guaranteed to be symmetrizable, and prove its convergence. By solving the eigen problem related to the new operator, we define a set of orthogonal bases over the point cloud. Experiments show that the new operator is converging better than other symmetrizable discrete Laplacian operators (such as graph Laplacian) defined on point-sampled surfaces, and can provide orthogonal bases for further spectral geometric analysis and processing tasks.

Mining pattern sequences in respiratory tumor motion data.

Management of respiration induced tumor motion during radiation therapy is crucial to effective treatment. Pattern sequences in the tumor motion signals can be valuable features in the analysis and prediction of irregular tumor motion. In this study, we put forward an approach towards mining pattern sequences in respiratory tumor motion data. We discuss the use of pattern sequence distributions as effective representations of motion characteristics, and find similarities between individual tumor motion instances. We also explore grouping of patients based on similarities in pattern sequence distributions exhibited by their respiratory motion traces.

A body sensor network with electromyogram and inertial sensors: multimodal interpretation of muscular activities.

The evaluation of the postural control system (PCS) has applications in rehabilitation, sports medicine, gait analysis, fall detection, and diagnosis of many diseases associated with a reduction in balance ability. Standing involves significant muscle use to maintain balance, making standing balance a good indicator of the health of the PCS. Inertial sensor systems have been used to quantify standing balance by assessing displacement of the center of mass, resulting in several standardized measures. Electromyogram (EMG) sensors directly measure the muscle control signals. Despite strong evidence of the potential of muscle activity for balance evaluation, less study has been done on extracting unique features from EMG data that express balance abnormalities. In this paper, we present machine learning and statistical techniques to extract parameters from EMG sensors placed on the tibialis anterior and gastrocnemius muscles, which show a strong correlation to the standard parameters extracted from accelerometer data. This novel interpretation of the neuromuscular system provides a unique method of assessing human balance based on EMG signals. In order to verify the effectiveness of the introduced features in measuring postural sway, we conduct several classification tests that operate on the EMG features and predict significance of different balance measures.

Indexing 3-D human motion repositories for content-based retrieval.

Content-based retrieval of the similar motions for the human joints has significant impact in the fields of physical medicine, biomedicine, rehabilitation, and motion therapy. In this paper, we propose an efficient indexing approach for 3-D human motion capture data, supporting queries involving both subbody motions as well as whole-body motions.