The use of machine learning and deep learning techniques to assess proprioceptive impairments of the upper limb after stroke

Artificial Intelligence in Medical Education and Mentoring in Rehabilitation Medicine

ABSTRACT

Artificial intelligence emerges as a transformative force, offering novel solutions to enhance medical education and mentorship in the specialty of physical medicine and rehabilitation. Artificial intelligence is a transformative technology that is being adopted in nearly every industry. In medicine, the use of artificial intelligence in medical education is growing. Artificial intelligence may also assist with some of the challenges of mentorship, including the limited availability of experienced mentors, and the logistical difficulties of time and geography are some constraints of traditional mentorship. In this commentary, we discuss various models of artificial intelligence in medical education and mentoring, including expert systems, conversational agents, and hybrid models. These models enable tailored guidance, broaden outreach within the physical medicine and rehabilitation community, and support continuous learning and development. Balancing artificial intelligence's technical advantages with the essential human elements while addressing ethical considerations, artificial intelligence integration into medical education and mentorship presents a paradigm shift toward a more accessible, responsive, and enriched experience in rehabilitation medicine.

Towards Transforming Neurorehabilitation: The Impact of Artificial Intelligence on Diagnosis and Treatment of Neurological Disorders

Background and Objectives: Neurological disorders like stroke, spinal cord injury (SCI), and Parkinson's disease (PD) significantly affect global health, requiring accurate diagnosis and long-term neurorehabilitation. Artificial intelligence (AI), such as machine learning (ML), may enhance early diagnosis, personalize treatment, and optimize rehabilitation through predictive analytics, robotic systems, and brain-computer interfaces, improving outcomes for patients. This systematic review examines how AI and ML systems influence diagnosis and treatment in neurorehabilitation among neurological disorders. Materials and Methods: Studies were identified from an online search of PubMed, Web of Science, and Scopus databases with a search time range from 2014 to 2024. This review has been registered on Open OSF (n) EH9PT. Results: Recent advancements in AI and ML are revolutionizing motor rehabilitation and diagnosis for conditions like stroke, SCI, and PD, offering new opportunities for personalized care and improved outcomes. These technologies enhance clinical assessments, therapy personalization, and remote monitoring, providing more precise interventions and better long-term management. Conclusions: AI is revolutionizing neurorehabilitation, offering personalized, data-driven treatments that enhance recovery in neurological disorders. Future efforts should focus on large-scale validation, ethical considerations, and expanding access to advanced, home-based care.

Bibliometric and visualized analysis of the application of artificial intelligence in stroke

Background

Stroke stands as a prominent cause of mortality and disability worldwide, posing a major public health concern. Recent years have witnessed rapid advancements in artificial intelligence (AI). Studies have explored the utilization of AI in imaging analysis, assistive rehabilitation, treatment, clinical decision-making, and outcome and risk prediction concerning stroke. However, there is still a lack of systematic bibliometric analysis to discern the current research status, hotspots, and possible future development trends of AI applications in stroke.

Methods

The publications on the application of AI in stroke were retrieved from the Web of Science Core Collection, spanning 2004-2024. Only articles or reviews published in English were included in this study. Subsequently, a manual screening process was employed to eliminate literature not pertinent to the topic. Visualization diagrams for comprehensive and in-depth analysis of the included literature were generated using CiteSpace, VOSviewer, and Charticulator.

Results

This bibliometric analysis included a total of 2,447 papers, and the annual publication volume shows a notable upward trajectory. The most prolific authors, countries, and institutions are Dukelow, Sean P., China, and the University of Calgary, respectively, making significant contributions to the advancement of this field. Notably, stable collaborative networks among authors and institutions have formed. Through clustering and citation burst analysis of keywords and references, the current research hotspots have been identified, including machine learning, deep learning, and AI applications in stroke rehabilitation and imaging for early diagnosis. Moreover, emerging research trends focus on machine learning as well as stroke outcomes and risk prediction.

Conclusion

This study provides a comprehensive and in-depth analysis of the literature regarding AI in stroke, facilitating a rapid comprehension of the development status, cooperative networks, and research priorities within the field. Furthermore, our analysis may provide a certain reference and guidance for future research endeavors.

The independence of impairments in proprioception and visuomotor adaptation after stroke

BACKGROUND

Proprioceptive impairments are common after stroke and are associated with worse motor recovery and poor rehabilitation outcomes. Motor learning may also be an important factor in motor recovery, and some evidence in healthy adults suggests that reduced proprioceptive function is associated with reductions in motor learning. It is unclear how impairments in proprioception and motor learning relate after stroke. Here we used robotics and a traditional clinical assessment to examine the link between impairments in proprioception after stroke and a type of motor learning known as visuomotor adaptation.

METHODS

We recruited participants with first-time unilateral stroke and controls matched for overall age and sex. Proprioceptive impairments in the more affected arm were assessed using robotic arm position- (APM) and movement-matching (AMM) tasks. We also assessed proprioceptive impairments using a clinical scale (Thumb Localization Test; TLT). Visuomotor adaptation was assessed using a task that systematically rotated hand cursor feedback during reaching movements (VMR). We quantified how much participants adapted to the disturbance and how many trials they took to adapt to the same levels as controls. Spearman's rho was used to examine the relationship between proprioception, assessed using robotics and the TLT, and visuomotor adaptation. Data from healthy adults were used to identify participants with stroke who were impaired in proprioception and visuomotor adaptation. The independence of impairments in proprioception and adaptation were examined using Fisher's exact tests.

RESULTS

Impairments in proprioception (58.3%) and adaptation (52.1%) were common in participants with stroke (n = 48; 2.10% acute, 70.8% subacute, 27.1% chronic stroke). Performance on the APM task, AMM task, and TLT scores correlated weakly with measures of visuomotor adaptation. Fisher's exact tests demonstrated that impairments in proprioception, assessed using robotics and the TLT, were independent from impairments in visuomotor adaptation in our sample.

CONCLUSION

Our results suggest impairments in proprioception may be independent from impairments in visuomotor adaptation after stroke. Further studies are needed to understand factors that influence the relationship between motor learning, proprioception and other rehabilitation outcomes throughout stroke recovery.

Sensory deficits of the paretic and non-paretic upper limbs relate with the motor recovery of the poststroke subjects

BACKGROUND

Post stroke, motor paresis has usually been considered to be a crucial factor responsible for the disability; other impairments such as somatosensory deficits may also play a role.

OBJECTIVE

To determine the relation between the sensory deficits (paretic and non-paretic upper limbs) and the motor recovery of the paretic upper limb and to predict the potential of motor recovery based on the sensory deficits among stroke subjects.

METHODS

The study was a cross-sectional study conducted in a rehabilitation institute. Ninety-five poststroke hemiparetic subjects having sensory impairment in any of the modalities were considered for this study. Sensory deficits were assessed on both the upper limbs (paretic and non-paretic) primarily using Erasmus MC modification of the revised version of Nottingham Sensory Assessment (Em-NSA) and Nottingham Sensory Assessment (Stereognosis) (NSA-S). The motor recovery was assessed using the Fugl-Meyer assessment (FMA).

RESULTS

The measures of sensory deficits exhibited weak but significant correlation [the paretic (Em-NSA and NSA; r = .38 to .58; p < .001) and the non-paretic (Em-NSA and NSA; r = .24 to .38; p = .03 to .001)] with the motor recovery of the paretic upper limb as measured by FMA. The potential of favorable recovery of the paretic upper limb may be predicted using the cutoff scores of Em-NSA (30, 21, and 24) and NSA-S (5, 8, and 5) of the paretic side.

CONCLUSION

In stroke, sensory deficits relate weakly with the recovery of the paretic upper limb and can predict recovery potential of the paretic upper limb.

Robotic assessment of sensorimotor and cognitive deficits in patients with temporal lobe epilepsy

OBJECTIVE

Individuals with temporal lobe epilepsy (TLE) frequently demonstrate impairments in executive function, working memory, and/or declarative memory. It is recommended that screening for cognitive impairment is undertaken in all people newly diagnosed with epilepsy. However, standard neuropsychological assessments are a limited resource and thus not available to all. Our study investigated the use of robotic technology (the Kinarm robot) for cognitive screening.

METHODS

27 participants with TLE (17 left) underwent both a brief neuropsychological screening and a robotic (Kinarm) assessment. The degree of impairments and correlations between standardized scores from both approaches to assessments were analysed across different neurocognitive domains. Performance was compared between people with left and right TLE to look for laterality effects. Finally, the association between the duration of epilepsy and performance was assessed.

RESULTS

Across the 6 neurocognitive domains (attention, executive function, language, memory, motor and visuospatial) assessed by our neuropsychological screening, all showed scores that significantly correlated with Kinarm tasks assessing the same cognitive domains except language and memory that were not adequately assessed with Kinarm. Participants with right TLE performed worse on most tasks than those with left TLE, including both visuospatial (typically considered right hemisphere), and verbal memory and language tasks (typically considered left hemisphere). No correlations were found between the duration of epilepsy and either the neuropsychological screening or Kinarm assessment.

SIGNIFICANCE

Our findings suggest that Kinarm may be a useful tool in screening for neurocognitive impairment in people with TLE. Further development may facilitate an easier and more rapid screening of cognition in people with epilepsy and distinguishing patterns of cognitive impairment.

The Use of Kinematic Features in Evaluating Upper Limb Motor Function Learning Progress Based on Machine Learning

Evaluating progress throughout a patient's rehabilitation process helps choose effective treatment and formulate personalised and evidence-based rehabilitation interventions. The evaluation process is difficult due to the limitations of current clinical assessments. They lack the ability to reflect sensitive changes continuously throughout the rehabilitation process. Kinematic features have been extracted from individual's movement to address this problem due to their sensitivity and continuity. However, choosing appropriate kinematic features for rehabilitation evaluation has always been challenging. This paper exploits the application of kinematic features to classify movement patterns and movement qualities. 12 kinematic features were firstly extracted from a 7-segment triangle pattern of motion to monitor the learning progress with more numbers of drawing attempts. A statistical analysis was then conducted to compare the selected kinematic features with the clinically validated normalised jerk. Two supervised machine learning models were finally developed to classify movement patterns and movement qualities based on the selected kinematic features. The study was based on data recorded from 14 participants using a single position sensor. 6 kinematic features were able to reflect sensitive changes during the experiment and all kinematic features contributed to the classification tasks. Consistent with the literature, the results indicated that features based on movement velocity were the most beneficial in the classification tasks.