Automatic Post-Stroke Severity Assessment Using Novel Unsupervised Consensus Learning for Wearable and Camera-Based Sensor Datasets

BackgroundCompensatory movements are commonly employed by stroke survivors during seated reaching and may have negative effects on their long-term recovery. Detecting compensation is useful for coaching the patient to reduce compensatory trunk movements and improving the motor function of the paretic arm. Sensor-based and camera-based systems have been developed to detect compensatory movements, but they still have some limitations, such as causing object obstructions, requiring complex setups and raising privacy concerns. To overcome these drawbacks, this paper proposes a compensatory movement detection system based on pressure distribution data and is unobtrusive, simple and practical. Machine learning algorithms were applied to classify compensatory movements automatically. Therefore, the purpose of this study was to develop and test a pressure distribution-based system for the automatic detection of compensation movements of stroke survivors using machine learning algorithms.MethodsEight stroke survivors performed three types of reaching tasks (back-and-forth, side-to-side, and up-and-down reaching tasks) with both the healthy side and the affected side. The pressure distribution data were recorded, and five features were extracted for classification. The k-nearest neighbor (k-NN) and support vector machine (SVM) algorithms were applied to detect and categorize the compensatory movements. The surface electromyography (sEMG) signals of nine trunk muscles were acquired to provide a detailed description and explanation of compensatory movements.ResultsCross-validation yielded high classification accuracies (F1-score>0.95) for both the k-NN and SVM classifiers in detecting compensation movements during all the reaching tasks. In detail, an excellent performance was achieved in discriminating between compensation and noncompensation (NC) movements, with an average F1-score of 0.993. For the multiclass classification of compensatory movement patterns, an average F1-score of 0.981 was achieved in recognizing the NC, trunk lean-forward (TLF), trunk rotation (TR) and shoulder elevation (SE) movements.ConclusionsGood classification performance in detecting and categorizing compensatory movements validated the feasibility of the proposed pressure distribution-based system. Reliable classification accuracy achieved by the machine learning algorithms indicated the potential to monitor compensation movements automatically by using the pressure distribution-based system when stroke survivors perform seated reaching tasks.

In the United States, the incidence rate of new or recurrent stroke is approximately 795 000 per year, and stroke prevalence for individuals over the age of 20 years is estimated at 6.5 million.1 Mortality rates in the first 30 days after stroke have decreased because of advances in emergency medicine and acute stroke care. In addition, there is strong evidence that organized postacute, inpatient stroke care delivered within the first 4 weeks by an interdisciplinary healthcare team results in an absolute reduction in the number of deaths.2,3 Despite these positive achievements, stroke continues to represent the leading cause of long-term disability in Americans: An estimated 50 million stroke survivors worldwide currently cope with significant physical, cognitive, and emotional deficits, and 25% to 74% of these survivors require some assistance or are fully dependent on caregivers for activities of daily living (ADLs).4,5 Notwithstanding the substantial progress in acute stroke care over the past 15 years, the focus of stroke medical advances and healthcare resources has been on acute and subacute recovery phases, which has resulted in substantial health disparities in later phases of stroke care. Additionally, healthcare providers (HCPs) are often unaware of not only patients’ potential for improvement during more chronic recovery phases but also common issues that stroke survivors and their caregivers experience. Furthermore, even with evidence that documents neuroplasticity potential regardless of age and time after stroke,6 the mean lifetime cost of ischemic stroke (which accounts for 87% of all strokes) in the United States is an estimated $140 000 (for inpatient, rehabilitation, and follow-up costs), with 70% of first-year stroke costs attributed to acute inpatient hospital care1; therefore, fewer financial resources appear to be dedicated to providing optimal care during the later phases of stroke recovery. Because there remains a …

Characterization of Compensatory Trunk Movements During Prosthetic Upper Limb Reaching Tasks

Background Many stroke survivors do not regain full function of their upper limb, which can have a devastating effect for participation in daily activities. Objectives This research aimed to systematically review stroke survivors’ lived experiences of upper limb recovery. Our objectives explored stroke survivors’ barriers and facilitators to recovery and the clinical implications for health professionals. Methods We searched the databases of MEDLINE, EMCARE, CINAHL, Scopus, and Web of Science using keywords and synonyms of ‘stroke survivors’, ‘upper limb’, ‘recovery’, and ‘perspectives’. Studies were appraised using the Critical Appraisal Skills Programme (CASP) Qualitative checklist. Data were analyzed descriptively and analytically using a thematic synthesis approach. We graded the emerging descriptive themes using the Confidence in the Evidence from Reviews of Qualitative Research (GRADE-CERQual). A protocol was registered apriori in PROSPERO (ID:181379). Results Eleven studies were included for review. Nine descriptive themes and three analytical themes emerged from the data that influenced upper limb recovery: 1) psychological wellbeing, 2) rehabilitation, and 3) relationships. Conclusions Barriers to upper limb recovery included health professionals’ attitudes particularly towards a lack of inclusivity in goal setting. In addition, stroke survivors perceived therapists focused on upper limb compensation rather than functional recovery, which was disempowering. Facilitators that enhanced recovery included the stroke survivors’ personal attitudes (psychological well-being) and external supports from family, friends, health professionals, and stroke survivor peers. Such supports enhanced the stroke survivors’ journey towards acceptance and adapting to their ‘new normal’ thus building confidence and self-efficacy to re-establish identity.

Purpose: Practitioners can use stroke survivors’ lived experiences associated with everyday activities to understand and assess recovery from stroke, to plan interventions and to facilitate goal setting. This study explored the experiences associated with daily activities among community dwelling survivors of stroke to identify factors that influenced these experiences. Method: In this mixed method study, 23 stroke survivors completed the Daily Experiences of Pleasure, Productivity and Restoration Profile, a time use survey, on three separate, self-chosen days within 1 week. Participants subsequently completed individual, semi-structured interviews. Results: Descriptive statistics and constant comparative analysis revealed that high levels of pleasure were associated with low-demand leisure activities, high levels of productivity were associated with home management activities and engaging in hobbies provided the highest combination of pleasure, productivity and restoration. The study also revealed that lived experiences of pleasure, productivity and restoration were influenced by the convergence of internal and external factors. Conclusions: Stroke survivors’ lived experiences of pleasure, productivity and restoration in the context of everyday activities yields insight into the survivors’ recovery and adjustment processes and into their quality of life. Practitioners should therefore incorporate stroke survivors’ experiences as they facilitate goal setting and plan interventions.Implications for RehabilitationUnderstanding stroke survivors’ lived experiences provides practitioners valuable information to assist with goal setting and intervention planning.A sense of accomplishment may be gained outside of work in home management and high demand leisure activities for survivors with low self-efficacy.Practitioners should consider facilitating stroke survivors’ participation in their hobbies as these activities were associated with high level of pleasure, productivity and restoration.Attending to stroke survivors’ self-perceptions and self-expectations is critical as these internal factors influence their lived experience.

<p dir="ltr">According to the World Bank report, 15% of the world population suffers from a disability such as a stroke, myopathy, neuropathy, spinal cord injuries, bone atrophy, etc. Stroke annually occurs for 15 million people worldwide, and 50,000 are in Canada. For stroke survivors, assessment rehabilitation is often required to assist the affected body parts in regaining control over their mobility. With a substantial portion of the global population experiencing disabilities, including stroke survivors, remote patient monitoring and assessment using sensor technology and IoT devices has become feasible. Identifying and assessing the impaired body part is essential to providing proper treatment and rehabilitation plans. Therefore, monitoring Activities of Daily Living (ADLs) and complex body movements of post-stroke survivors could deliver a wealth of clinically applicable information. Specifically, evaluating stroke ADLs in a clinical setting is constrained to information provided by a healthcare professional and subjective rating scales based on their judgment. Thus, the assessment score and tests for stroke patients are done manually with a physiotherapist's opinion, which is subjective. </p><p dir="ltr">To address the above-mentioned challenges for stroke survivors and utilize artificial intelligence (AI) in the rehabilitation assessment area, this thesis investigated the problem of iii automatically identifying affected body parts and the severity level of the affected hand. Furthermore, incorporating diverse sensor technologies, encompassing open-source wearable sensor-based (Xsens) and camera-based (Vicon) datasets, enriches the scope and depth of this thesis. Additionally, addressing the scalability of the developed AI model and maintaining the privacy of the patients' dataset is investigated. A novel Multi-Level Meta Learner (MLML) diagnosis model was developed using ensemble learning to accurately classify stroke patients' affected hands from non-affected hands virtually. Moreover, provides an intelligent post stroke severity assessment using a consensus clustering algorithm called Post-Stroke Assessment-Midified Nonnegative Matrix Factorization (PSA-MNMF) inspired by the advances in consensus learning that combine various clustering methods into one united clustering to produce more stable and robust results than individual clustering. The method is the first to investigate the severity levels using unsupervised learning and trunk displacement features in the frequency domain for post-stroke smart assessment. To ensure scalability and patient privacy, a federated learning approach called the Post-Stroke Federated Learning Consensus-Driven Model (PSA-FL-CDM) was proposed to harness the vast datasets, enhance the model's performance, and reduce computational time compared to the centralized model. </p><p dir="ltr">From a clinical standpoint, there is an increasing recognition of the shift from traditional clinical assessments to a more technologically advanced approach. This shift involves utilizing AI-powered sensors to comprehensively assess neurological deficits, functional capabilities, and activities of daily living and even potentially estimate the quality of life. The primary objective of this thesis was to satisfy the needs of clinically assessing post-stroke, assess and compare the functional capabilities of the affected hand in post-stroke patients and a control group.</p>

ObjectivesThe five-repetition sit-to-stand (5R-STS) test was designed to capture objective functional impairment (OFI), and thus provides an adjunctive dimension in patient assessment. It is conceivable that there are different subsets of patients with OFI and degenerative lumbar disease. We aim to identify clusters of objectively functionally impaired individuals based on 5R-STS and unsupervised machine learning (ML).MethodsData from two prospective cohort studies on patients with surgery for degenerative lumbar disease and 5R-STS times of ≥ 10.5 s—indicating presence of OFI. K-means clustering—an unsupervised ML algorithm—was applied to identify clusters of OFI. Cluster hallmarks were then identified using descriptive and inferential statistical analyses.ResultsWe included 173 patients (mean age [standard deviation]: 46.7 [12.7] years, 45% male) and identified three types of OFI. OFI Type 1 (57 pts., 32.9%), Type 2 (81 pts., 46.8%), and Type 3 (35 pts., 20.2%) exhibited mean 5R-STS test times of 14.0 (3.2), 14.5 (3.3), and 27.1 (4.4) seconds, respectively. The grades of OFI according to the validated baseline severity stratification of the 5R-STS increased significantly with each OFI type, as did extreme anxiety and depression symptoms, issues with mobility and daily activities. Types 1 and 2 are characterized by mild to moderate OFI—with female gender, lower body mass index, and less smokers as Type I hallmarks.ConclusionsUnsupervised learning techniques identified three distinct clusters of patients with OFI that may represent a more holistic clinical classification of patients with OFI than test-time stratifications alone, by accounting for individual patient characteristics.

The advent of gravitational wave astronomy (GW) has revolutionized the observation of cataclysmic cosmic events, such as black hole mergers and neutron star collisions. The Laser Interferometer Gravitational-Wave Observatory (LIGO) has been at the forefront of these discoveries. However, the immense volume and complexity of gravitational wave data present significant challenges for traditional analysis methods. This paper investigates the growing synergy between artificial intelligence (AI) and GW science, emphasizing how AI enhances signal detection, noise reduction, and data interpretation. It begins with an overview of GW fundamentals and the role of machine learning in increasing detector sensitivity. Notable GW events observed by LIGO are discussed alongside persistent analytical challenges such as data quality, generalization, and computational constraints. A comprehensive performance review of AI techniques, including supervised learning, unsupervised learning, deep learning, and reinforcement learning is presented based on data spanning 2021 to 2024. Evaluation metrics include accuracy, precision, true positive rate (TPR), false positive rate (FPR), and computational efficiency. Findings indicate that deep learning and supervised learning outperform other approaches, particularly in enhancing TPR and minimizing FPR. While unsupervised and reinforcement learning models offer less precision, they demonstrate high efficiency and potential for real-time applications. The study also explores AI’s integration into next-generation detectors and waveform reconstruction techniques. Overall, the integration of AI into GW research significantly improves the reliability and speed of event detection, unlocking new possibilities for exploring the dynamic universe. This paper provides a comprehensive outlook on the transformative role of AI in shaping the future of GW astronomy.

As global CO2 emissions continue to rise, identifying meaningful patterns across countries has become increasingly vital for shaping effective climate policies. However, many existing approaches rely on uniform benchmarks that overlook national emission heterogeneity. To address this gap, this study applies two unsupervised machine learning techniques K-Means and Gaussian Mixture Model (GMM) to cluster countries based on CO2 emissions from both energy and industrial sectors. The dataset consists of six key indicators, including total emissions, growth rate, and sectoral shares. After handling missing values and applying Min-Max normalization, Principal Component Analysis (PCA) was used to reduce dimensionality and aid visualization. The core objective is to compare the effectiveness of K-Means and GMM in identifying emission-based country groupings. K-Means produced three distinct clusters with strong separation, including a unique cluster dominated solely by China due to its exceptional emission profile. GMM, by contrast, generated more flexible probabilistic clusters, better capturing overlapping patterns and internal variabilities among countries. Evaluation metrics showed that K-Means outperformed GMM in silhouette score and inertia, indicating clearer boundaries, while GMM was more adept at modeling complex, non-spherical distributions. These findings reveal the trade-offs between clarity and adaptability in clustering approaches. The study demonstrates how unsupervised learning can offer actionable insights for emission-based segmentation, enabling more nuanced and differentiated mitigation strategies. By highlighting algorithm-specific strengths, this research contributes to the advancement of machine learning applications in climate informatics and supports the development of targeted international environmental responses.

Stroke survivors are one of the largest consumer groups of rehabilitation services. Despite improvement in daily activities while in inpatient rehabilitation, many have difficulty performing daily activities at home after discharge. The difference in performance between a standard clinical context and at home is poorly understood. To better understand differences in activity performance during transition from inpatient rehabilitation facility (IRF) to home, we examined daily activity performance scores from 2 different environments (IRF and home) at the same time point (discharge). This was a cross-sectional analysis using baseline data from a randomized controlled trial. Participants were stroke survivors aged ≥50 who planned to discharge home from the IRF. The Functional Independence Measure and Section GG codes (both converted to International Classification of Functioning, Disability, and Health scores) were conducted per protocol first at home and then in the IRF at discharge (≤3 days apart, order not randomized). Among 57 participants, activity scores at home were significantly worse than scores at IRF discharge. Over 40% of participants had discharge scores indicating no-to-mild impairment for shower/tub transfer, walking, and going up/down stairs, while home visit scores indicated moderate-to-complete impairment for those activities. The greatest differences in scores were for shower/tub transfer (median difference 1.5, 95% CI 1.00-2.00) and going up/down stairs (median difference 1.50, 95% CI 1.00-2.00). The environment plays an important role in stroke survivors' functioning at home. Future studies should further examine how the environment impacts activity performance upon returning home following stroke.

Due to the increasing number of risk factors, energy sector has been experiencing interruptions (attack) in the normal operation both externally and internally. Different methods are used for the identification and evaluation of vulnerabilities due to these interruption in the complex and critical infrastructures like the smart grid. Based on the objective of the attack, the performance and effectiveness of the learning-based approaches may vary when compared with other approaches to identify critical components of the smart grid. In this work, we adopted two target selection strategies (one is an unsupervised learning algorithm and the other is a load ranking based approach) for attack and measured system performances based on two evaluation metrics. We conducted the experiments on four different standard power system test cases and compared the performances of the aforementioned target selection strategies by two evaluation metrics. We used K-means clustering as the unsupervised learning method for the target selection of contingencies. To evaluate the system damage, we used generation loss and number of transmission line outages. For different attack orders, with two different attack objectives (evaluation metrics), experiments were conducted on W&W 6 bus system, IEEE 7 bus system, IEEE 8 bus system, and IEEE 300 bus system. We showed that, a clustering based attack performs better when the system is relatively large (highly dense in terms of connection to other buses) and the objective is to achieve a high number of transmission line outages. On the other hand, load ranking based attack outperforms clustering based attack when the attack objective is to achieve higher generation loss, regardless of the size of the system.

Background. A significant factor in impaired movement caused by stroke is the inability to activate muscles independently. Although the pathophysiology behind this abnormal coactivation is not clear, reducing the coactivation could improve overall arm function. A myoelectric computer interface (MCI), which maps electromyographic signals to cursor movement, could be used as a treatment to help retrain muscle activation patterns. Objective. To investigate the use of MCI training to reduce abnormal muscle coactivation in chronic stroke survivors. Methods. A total of 5 healthy participants and 5 stroke survivors with hemiparesis participated in multiple sessions of MCI training. The level of arm impairment in stroke survivors was assessed using the upper-extremity portion of the Fugl-Meyer Motor Assessment (FMA-UE). Participants performed isometric activations of up to 5 muscles. Activation of each muscle was mapped to different directions of cursor movement. The MCI specifically targeted 1 pair of muscles in each participant for reduction of coactivation. Results. Both healthy participants and stroke survivors learned to reduce abnormal coactivation of the targeted muscles with MCI training. Out of 5 stroke survivors, 3 exhibited objective reduction in arm impairment as well (improvement in FMA-UE of 3 points in each of these patients). Conclusions. These results suggest that the MCI was an effective tool in directly retraining muscle activation patterns following stroke.

Stroke survivors often exhibit low motivation for rehabilitation, hindering their ability to effectively complete rehabilitation training task effectively and participate in daily activities actively. Reward strategies have been identified as an effective method for boosting rehabilitation motivation, but their long-term efficacy remains uncertain. Transcranial direct current stimulation (tDCS) has been recognized as a technique that facilitates plastic changes and functional reorganization of cortical areas. Particularly, tDCS can improve the functional connectivity between brain regions associated with goal-directed behavior when applied to the left dorsolateral prefrontal cortex (dlPFC). Combing reward strategies with tDCS (RStDCS) has been shown to motivate healthy individuals to exert more effort in task performance. However, research exploring the combined and sustained effects of these strategies on rehabilitation motivation in stroke survivors is lacking. Eighty-seven stroke survivors with low motivation and upper extremity dysfunction will be randomized to receive either conventional treatment, RS treatment, or RStDCS treatment. The RStDCS group will receive reward strategies combined with anodal tDCS stimulation of the left dlPFC. The RS group will receive reward strategies combined with sham stimulation. The conventional group will receive conventional treatment combined with sham stimulation. tDCS stimulation is performed over 3 weeks of hospitalization, 20 min/time, five times a week. Reward strategies refers to personalized active exercise programs for patients during hospitalization and at home. Patients can voluntarily choose tasks for active exercise and self-report to the therapist so as to punch a card for points and exchange gifts. The conventional group will receive home rehabilitation instructions prior to discharge. Rehabilitation motivation, measured using RMS. RMS, FMA, FIM, and ICF activity and social engagement scale will be compared at baseline, 3 weeks, 6 weeks, and 3 months post-enrollment to evaluate patients' multifaceted health condition based on the ICF framework. This study integrates knowledge from social cognitive science, economic behavioral science, and other relevant fields. We utilize straightforward and feasible reward strategies, combined with neuromodulation technology, to jointly improve patients' rehabilitation motivation. Behavioral observations and various assessment tools will be used to monitor patients' rehabilitation motivation and multifaceted health condition according to the ICF framework. The aim is to provide a preliminary exploration path for professionals to develop comprehensive strategies for improving patient rehabilitation motivation and facilitating a complete "hospital-home-society" rehabilitation process. https://www.chictr.org.cn/showproj.aspx?proj=182589, ChiCTR2300069068.

BackgroundPracticing activities improves recovery after stroke, but many people in hospital do little activity. Feedback on activity using an accelerometer is a potential method to increase activity in hospital inpatients. This study’s goal is to investigate the effect of feedback, enabled by a Smart watch, on daily physical activity levels during inpatient stroke rehabilitation and the short-term effects on simple functional activities, primarily mobility.Methods/designA randomized controlled trial will be undertaken within the stroke rehabilitation wards of the Second Affiliated hospital of Anhui University of Traditional Chinese Medicine, Hefei, China. The study participants will be stroke survivors who meet inclusion criteria for the study, primarily: able to participate, no more than 4 months after stroke and walking independently before stroke. Participants will all receive standard local rehabilitation and will be randomly assigned either to receive regular feedback about activity levels, relative to a daily goal tailored by the smart watch over five time periods throughout a working day, or to no feedback, but still wearing the Smart watch. The intervention will last up to 3 weeks, ending sooner if discharged. The data to be collected in all participants include measures of daily activity (Smart watch measure); mobility (Rivermead Mobility Index and 10-metre walking time); independence in personal care (Barthel Activities of Daily Living (ADL) Index); overall activities (the World Health Organization (WHO) Disability Assessment Scale, 12-item version); and quality of life (the Euro-Qol 5L5D). Data will be collected by assessors blinded to allocation of the intervention at baseline, 3 weeks or at discharge (whichever is the sooner); and a reduced data set will be collected at 12 weeks by telephone interview. The primary outcome will be change in daily accelerometer activity scores. Secondary outcomes are compliance and adherence to wearing the watch, and changes in mobility, independence in personal care activities, and health-related quality of life.DiscussionThis project is being implemented in a large city hospital with limited resources and limited research experience. There has been a pilot feasibility study using the Smart watch, which highlighted some areas needing change and these are incorporated in this protocol.Trial registrationClinicalTrials.gov, NCT02587585. Registered on 30 September 2015. Chinese Clinical Trial Registry, ChiCTR-IOR-15007179. Registered on 8 August 2015.

Background: Informal caregivers have played a significant role since stroke survivors were discharged from the hospital. Caregivers carry out stroke survivors' daily activities, which can be burdensome. Caregiver coping strategies to reduce this burden have not been widely studied, especially in Indonesia.Purpose: This study was aimed at exploring the act of devotion as a coping strategy to reduce the burden of caring for stroke survivors.Method: This was an exploratory qualitative study of 30 informal caregivers of stroke survivors who live in a community in Surakarta, selected using purposeful sampling. In-depth interviews to explore the act of devoting as a coping strategy and observations of caregivers were conducted from June 2021 to December 2022.Finding: The data analysis revealed two primary themes: “family ties” and “physical.” The family ties theme emerged with five subthemes: pride, sincerity, resigned, destiny, and responsibility. The physical theme consisted of two subthemes: ability and spirit.Conclusion: Devotion as a coping strategy for stroke informal caregivers in the community plays a crucial role in maintaining the continuity of care. It involves pride, sincerity, resigned, destiny, responsibility, ability, and spirit. Further research on the results of these findings is needed on how caregivers maintain their quality of life as they begin to become caregivers.