A decision support system based on recurrent neural networks to predict medication dosage for patients with Parkinson's disease

Using deep learning has demonstrated significant potential in making informed decisions based on clinical evidence. In this study, we deal with optimizing medication and quantitatively present the role of deep learning in predicting the medication dosage for patients with Parkinson's disease (PD). The proposed method is based on recurrent neural networks (RNNs) and tries to predict the dosage of five critical medication types for PD, including levodopa, dopamine agonists, monoamine oxidase-B inhibitors, catechol-O-methyltransferase inhibitors, and amantadine. Recurrent neural networks have memory blocks that retain crucial information from previous patient visits. This feature is helpful for patients with PD, as the neurologist can refer to the patient's previous state and the prescribed medication to make informed decisions. We employed data from the Parkinson's Progression Markers Initiative. The dataset included information on the Unified Parkinson's Disease Rating Scale, Activities of Daily Living, Hoehn and Yahr scale, demographic details, and medication use logs for each patient. We evaluated several models, such as multi-layer perceptron (MLP), Simple-RNN, long short-term memory (LSTM), and gated recurrent units (GRU). Our analysis found that recurrent neural networks (LSTM and GRU) performed the best. More specifically, when using LSTM, we were able to predict levodopa and dopamine agonist dosage with a mean squared error of 0.009 and 0.003, mean absolute error of 0.062 and 0.030, root mean square error of 0.099 and 0.053, and R-squared of 0.514 and 0.711, respectively.

A Decision-Aware Ambient Assisted Living System with IoT Embedded Device for In-Home Monitoring of Older Adults

Older adults' independent life is compromised due to various problems, such as memory impairments and decision-making difficulties. This work initially proposes an integrated conceptual model for assisted living systems capable of providing helping means for older adults with mild memory impairments and their caregivers. The proposed model has four main components: (1) an indoor location and heading measurement unit in the local fog layer, (2) an augmented reality (AR) application to make interactions with the user, (3) an IoT-based fuzzy decision-making system to handle the direct and environmental interactions with the user, and (4) a user interface for caregivers to monitor the situation in real time and send reminders once required. Then, a preliminary proof-of-concept implementation is performed to evaluate the suggested mode's feasibility. Functional experiments are carried out based on various factual scenarios, which validate the effectiveness of the proposed approach. The accuracy and response time of the proposed proof-of-concept system are further examined. The results suggest that implementing such a system is feasible and has the potential to promote assisted living. The suggested system has the potential to promote scalable and customizable assisted living systems to reduce the challenges of independent living for older adults.

Evaluating the Effect of Increasing Working Memory Load on EEG-Based Functional Brain Networks

Purpose: Working Memory (WM) plays a crucial role in many cognitive functions of the human brain. Examining how the inter-regional connectivity and characteristics of functional brain networks modulate with increasing WM load could lead to a more in-depth understanding of the WM system. Materials and Methods: To investigate the effect of WM load alterations on the inter-regional synchronization and functional network characteristics, we used Electroencephalogram (EEG) data recorded from 21 healthy participants during an n-back task with three load levels (0-back, 2-back, and 3-back). The networks were constructed based on the weighted Phase Lag Index (wPLI) in the theta, alpha, beta, low-gamma, and high-gamma frequency bands. After constructing the fully connected, weighted, and undirected networks, the node-to-node connections, graph-theory metrics consisting of mean Clustering coefficient (C), characteristic path Length (L), and node strength were analyzed by statistical tests. Results: It was revealed that in the presence of WM load (2- and 3-back tasks) compared with the WM-free condition (0-back task) within the alpha range, the Inter-Regional Functional Connectivity (IRFC), functional integration, functional segregation, and node strength in channels located at the frontal, parietal and occipital regions were significantly reduced. In the high-gamma band, IRFC was significantly higher in the difficult task (3-back) compared to the easy and moderate tasks (0- and 2-back). Besides, locally clustered connections were significantly increased in 3-back relative to the 2-back task. Conclusion: Inter-regional alpha synchronization and alpha-band network metrics can distinguish between the WM and WM-free tasks. In contrast, phase synchronization of high-gamma oscillations can differentiate between the levels of WM load, which demonstrates the potential of the phase-based functional connectivity and brain network metrics for predicting the WM load level.

Detecting ADHD Based on Brain Functional Connectivity Using Resting-State MEG Signals

Purpose: Attention Deficit Hyperactivity Disorder (ADHD) is now recognized as the most common childhood behavioral disorder. This disorder causes school problems and social incompatibility. Thus an accurate diagnosis can help diminish such problems. In this paper, we propose a brain connectomics approach based on eyes-open resting state Magnetoencephalography (rs-MEG) to diagnose subjects with ADHD from Healthy Controls (HC). Materials and Methods: We used the eyes-open rs-MEG signals recorded from 25 subjects with ADHD and 25 HC. We calculated Coherence (COH) between the MEG sensors in the conventional frequency bands (i.e., delta, theta, alpha, beta, and gamma), selected the most discriminative COH measures by the Neighborhood Component Analysis (NCA), and fed them to three classifiers, including Support Vector Machine (SVM) with Radial Basis Function (RBF) kernel, K-Nearest Neighbors (KNN), and Decision Tree to classify ADHD and HC. Results: We achieved the best average accuracy of 91.1% for a single-band classifier based on the COH in the delta-band as an input feature of the SVM. However, when we integrated the COH values of all frequency bands as input features, the average accuracy was slightly improved to 92.7% using the SVM classifier. Conclusion: Our results demonstrate the capability of a functional connectomics approach based on rs-MEG for the diagnosis of ADHD. It is noteworthy that, to the best of our knowledge, COH has not yet been used to diagnose ADHD using rs-MEG data. Furthermore, there is no study on diagnosing ADHD using eyes-open rs-MEG. Thus, a novelty of our proposed method is to use COH and eyes-open rs-MEG data to diagnose ADHD. Moreover, our proposed method showed promising results compared with previous rs-MEG studies for the diagnosis of ADHD.

Quantification of whole-body bradykinesia in Parkinson's disease participants using multiple inertial sensors

Bradykinesia (slowness of movement) is a common motor symptom of Parkinson's disease (PD) that can severely affect quality of life for those living with the disease. Assessment and treatment of PD motor symptoms largely depends on clinical scales such as the Unified Parkinson's Disease Rating Scale (UPDRS). However, such clinical scales rely on the visual assessment by a human observer, naturally resulting in inter-rater variability. Although previous studies have developed objective means for measuring bradykinesia in PD patients, their evaluation was restricted by the type of movement and number of joints assessed. These studies failed to provide a more comprehensive, whole-body evaluation capable of measuring multiple joints simultaneously. This study utilizes wearable inertial measurement units (IMUs) to quantify whole-body movements, providing novel bradykinesia indices for walking (WBI) and standing up from a chair (sit-to-stand; SBI). The proposed bradykinesia indices include the joint angles at both upper and lower limbs and trunk motion to compute a complete, objective score for whole body bradykinesia. Thirty PD and 11 age-matched healthy control participants were recruited for the study. The participants performed two standard walking tasks that involved multiple body joints in the upper and lower limbs. The WBI and SBI successfully identified differences between control and PD participants. The indices also effectively identified differences within the PD population, distinguishing participants assessed with (ON) and without (OFF) levodopa; the gold-standard of treatment for PD. The goal of this study is to provide health professionals with an objective score for whole body bradykinesia by simultaneously measuring the upper and lower extremities along with truncal movement. This method demonstrates potential to be used in conjunction with current clinical standards for motor symptom assessment, and may also be promising for the remote assessment of PD patients and in cases where experienced clinicians may not be available. In conclusion, the intelligent use of this technology for the measurement of bradykinesia (among other symptoms) has vast implications for optimizing treatment in Parkinson's disease, ultimately leading to an improvement in quality of life.

Towards remote monitoring of Parkinson's disease tremor using wearable motion capture systems

The management of movement disorders is shifting from a centralized-clinical assessment towards remote monitoring and individualized therapy. While a variety of treatment options are available, ranging from pharmaceutical drugs to invasive neuromodulation, the clinical effects are inconsistent and often poorly measured. For instance, the lack of remote monitoring has been a major limitation to optimize therapeutic interventions for patients with Parkinson's Disease (PD). In this work, we focus on the assessment of full-body tremor as the most recognized PD symptom. Forty PD and twenty two healthy participants were recruited. The main assessment tool was an inertial measurement unit (IMU)-based motion capture system to quantify full-body tremor and to separate tremor-dominant from non-tremor-dominant PD patients as well as from healthy controls. We developed a new measure and evaluated its clinical utility by correlating the results with the Unified Parkinson's Disease Rating Scale (UPDRS) scores as the gold standard. Significant correlation was observed between the UPDRS and the tremor severity scores for the selected tasks. The results suggest that it is feasible and clinically meaningful to utilize the suggested objective tremor score for the assessment of PD patients. Furthermore, this portable assessment tool could potentially be used in the home environment to monitor PD tremor and facilitate optimizing therapeutic interventions.

Haptic Feedback Manipulation During Botulinum Toxin Injection Therapy for Focal Hand Dystonia Patients: A Possible New Assistive Strategy

Abnormality of sensorimotor integration in the basal ganglia and cortex has been reported in the literature for patients with task-specific focal hand dystonia (FHD). In this study, we investigate the effect of manipulation of kinesthetic input in people living with writer's cramp disorder (a major form of FHD). For this purpose, severity of dystonia is studied for 11 participants while the symptoms of seven participants have been tracked during five sessions of assessment and Botulinum toxin injection (BoNT-A) therapy (one of the current suggested therapies for dystonia). BoNT-A therapy is delivered in the first and the third session. The goal is to analyze the effect of haptic manipulation as a potential assistive technique during BoNT-A therapy. The trial includes writing, hovering, and spiral/sinusoidal drawing subtasks. In each session, the subtasks are repeated twice when (a) a participant uses a normal pen, and (b) when the participant uses a robotics-assisted system (supporting the pen) which provides a compliant virtual writing surface and manipulates the kinesthetic sensory input. The results show (p-value using one-sample t-tests) that reducing the writing surface rigidity significantly decreases the severity of dystonia and results in better control of grip pressure (an indicator of dystonic cramping). It is also shown that (p-value based on paired-samples t-test) using the proposed haptic manipulation strategy, it is possible to augment the effectiveness of BoNT-A therapy. The outcome of this study is then used in the design of an actuated pen as a writing-assistance tool that can provide compliant haptic interaction during writing for FHD patients.

Predicting Improvement in Writer's Cramp Symptoms following Botulinum Neurotoxin Injection Therapy

INTRODUCTION

Writer's cramp is a specific focal hand dystonia causing abnormal posturing and tremor in the upper limb. The most popular medical intervention, botulinum neurotoxin type A (BoNT-A) therapy, is variably effective for 50-70% of patients. BoNT-A non-responders undergo ineffective treatment and may experience significant side effects. Various assessments have been used to determine response prediction to BoNT-A, but not in the same population of patients.

METHODS

A comprehensive assessment was employed to measure various symptom aspects. Clinical scales, full upper-limb kinematic measures, self-report, and task performance measures were assessed for nine writer's cramp patients at baseline. Patients received two BoNT-A injections then were classified as responders or non-responders based on a quantified self-report measure. Baseline scores were compared between groups, across all measures, to determine which scores predicted a positive BoNT-A response.

RESULTS

Five of nine patients were responders. No kinematic measures were predictably different between groups. Analyses revealed three features that predicted a favorable response and separated the two groups: higher than average cramp severity and cramp frequency, and below average cramp latency.

DISCUSSION

Non-kinematic measures appear to be superior in making such predictions. Specifically, measures of cramp severity, frequency, and latency during performance of a specific set of writing and drawing tasks were predictive factors. Since kinematic was not used to determine the injection pattern and the injections were visually guided, it may still be possible to use individual patient kinematics for better outcomes.

Kinematic and kinetic assessment of upper limb movements in patients with writer's cramp

Background

The assessment and treatment of writer’s cramp is complicated due to the variations in the forces and angles of involved joints. Additionally, in some cases compensatory movements for cramp relief further complicates assessment. Currently these variables are subjectively measured with clinical scales and visual assessments. This subjectivity makes it difficult to successfully administer interventions such as Botulinum toxin injection or orthotics resulting in poor efficacy and significant side effects.

Method

A multi-sensor system was used to record finger and wrist forces along with deviation angles at the wrist, elbow and shoulder while 9 patients with writer's cramp performed a series of standardized tasks on surfaces inclined at different angles. Clinical, kinetic, and kinematic information regarding cramping was collected.

Results

First, four tasks appeared to best predict cramp occurrence. Second, unique biomechanical profiles emerged for patients regarding force, angles and cramp severity. Third, cluster analyses using these features showed a clear separation of patients into two severity classes. Finally, a relationship between severity and kinetic-kinematic information suggested that primary cramping versus compensatory movements could be potentially inferred.

Conclusions

The results demonstrate that using a set of standardized tasks and objective measures, individual profiles for arm movements and applied forces associated with writer’s cramp can be generated. The clinician can then accurately target the biomechanics specifically, whether it is with injection or other rehabilitative measures, fulfilling an important unmet need in the treatment of writer’s cramp.

Electronic supplementary material

The online version of this article (doi:10.1186/s12984-016-0122-0) contains supplementary material, which is available to authorized users.