Rehabilitation medicine is facing a new development phase thanks to a recent wave of rigorous clinical trials aimed at improving the scientific evidence of protocols. This phenomenon, combined with ...new trends in personalised medical therapies, is expected to change clinical practice dramatically. The emerging field of Rehabilomics is only possible if methodologies are based on biomedical data collection and analysis. In this framework, the objective of this work is to develop a systematic review of machine learning algorithms as solutions to predict motor functional recovery of post-stroke patients after treatment.
We conducted a comprehensive search of five electronic databases using the Patient, Intervention, Comparison and Outcome (PICO) format. We extracted health conditions, population characteristics, outcome assessed, the method for feature extraction and selection, the algorithm used, and the validation approach. The methodological quality of included studies was assessed using the prediction model risk of bias assessment tool (PROBAST). A qualitative description of the characteristics of the included studies as well as a narrative data synthesis was performed.
A total of 19 primary studies were included. The predictors most frequently used belonged to the areas of demographic characteristics and stroke assessment through clinical examination. Regarding the methods, linear and logistic regressions were the most frequently used and cross-validation was the preferred validation approach.
We identified several methodological limitations: small sample sizes, a limited number of external validation approaches, and high heterogeneity among input and output variables. Although these elements prevented a quantitative comparison across models, we defined the most frequently used models given a specific outcome, providing useful indications for the application of more complex machine learning algorithms in rehabilitation medicine.
Stroke related motor function deficits affect patients' likelihood of returning to professional activities, limit their participation in society and functionality in daily living. Hence, robot-aided ...gait rehabilitation needs to be fruitful and effective from a motor learning perspective. For this reason, optimal human-robot interaction strategies are necessary to foster neuroplastic shaping during therapy. Therefore, we performed a systematic search on the effects of different control algorithms on quantitative objective gait parameters of post-acute stroke patients.
We conducted a systematic search on four electronic databases using the Population Intervention Comparison and Outcome format. The heterogeneity of performance assessment, study designs and patients' numerosity prevented the possibility to conduct a rigorous meta-analysis, thus, the results were presented through narrative synthesis.
A total of 31 studies (out of 1036) met the inclusion criteria, without applying any temporal constraints. No controller preference with respect to gait parameters improvements was found. However, preferred solutions were encountered in the implementation of force control strategies mostly on rigid devices in therapeutic scenarios. Conversely, soft devices, which were all position-controlled, were found to be more commonly used in assistive scenarios. The effect of different controllers on gait could not be evaluated since conspicuous heterogeneity was found for both performance metrics and study designs.
Overall, due to the impossibility of performing a meta-analysis, this systematic review calls for an outcome standardisation in the evaluation of robot-aided gait rehabilitation. This could allow for the comparison of adaptive and human-dependent controllers with conventional ones, identifying the most suitable control strategies for specific pathologic gait patterns. This latter aspect could bolster individualized and personalized choices of control strategies during the therapeutic or assistive path.
Abstract
Background
Rehabilitation treatments and services are essential for the recovery of post-stroke patients’ functions; however, the increasing number of available therapies and the lack of ...consensus among outcome measures compromises the possibility to determine an appropriate level of evidence. Machine learning techniques for prognostic applications offer accurate and interpretable predictions, supporting the clinical decision for personalised treatment. The aim of this study is to develop and cross-validate predictive models for the functional prognosis of patients, highlighting the contributions of each predictor.
Methods
A dataset of 278 post-stroke patients was used for the prediction of the class transition, obtained from the modified Barthel Index. Four classification algorithms were cross-validated and compared. On the best performing model on the validation set, an analysis of predictors contribution was conducted.
Results
The Random Forest obtained the best overall results on the accuracy (76.2%), balanced accuracy (74.3%), sensitivity (0.80), and specificity (0.68). The combination of all the classification results on the test set, by weighted voting, reached 80.2% accuracy. The predictors analysis applied on the Support Vector Machine, showed that a good trunk control and communication level, and the absence of bedsores retain the major contribution in the prediction of a good functional outcome.
Conclusions
Despite a more comprehensive assessment of the patients is needed, this work paves the way for the implementation of solutions for clinical decision support in the rehabilitation of post-stroke patients. Indeed, offering good prognostic accuracies for class transition and patient-wise view of the predictors contributions, it might help in a personalised optimisation of the patients’ rehabilitation path.
COVID-19 cases are increasing around the globe with almost 5 million of deaths. We propose here a deep learning model capable of predicting the duration of the infection by means of information ...available at hospital admission. A total of 222 patients were enrolled in our observational study. Anagraphical and anamnestic data, COVID-19 signs and symptoms, COVID-19 therapy, hematochemical test results, and prior therapies administered to patients are used as predictors. A set of 55 features, all of which can be taken in the first hours of the patient’s hospitalization, was considered. Different solutions were compared achieving the best performance with a sequential convolutional neural network-based model merged in an ensemble with two different meta-learners linked in cascade. We obtained a median absolute error of 2.7 days (IQR = 3.0) in predicting the duration of the infection; the error was equally distributed in the infection duration range. This tool could preemptively give an outlook of the COVID-19 patients’ expected path and the associated hospitalization effort. The proposed solution could be viable in tackling the huge burden and the logistics complexity of hospitals or rehabilitation centers during the pandemic waves.
Graphical abstract
With data taken ad admission, entering a PCA-based feature selection, a
k
-fold cross-validated CNN-based model was implemented. After external texting, a median absolute error of 2.7 days IQR = 3 days
Patients with Disorder of Consciousness (DoC) entering Intensive Rehabilitation Units after a severe Acquired Brain Injury have a highly variable evolution of the state of consciousness which is a ...complex aspect to predict. Besides clinical factors, electroencephalography has clearly shown its potential into the identification of prognostic biomarkers of consciousness recovery. In this retrospective study, with a dataset of 271 patients with DoC, we proposed three different Elastic-Net regressors trained on different datasets to predict the Coma Recovery Scale-Revised value at discharge based on data collected at admission. One dataset was completely EEG-based, one solely clinical data-based and the last was composed by the union of the two. Each model was optimized, validated and tested with a robust nested cross-validation pipeline. The best models resulted in a median absolute test error of 4.54 IQR = 4.56, 3.39 IQR = 4.36, 3.16 IQR = 4.13 for respectively the EEG, clinical and hybrid model. Furthermore, the hybrid model for what concerns overcoming an unresponsive wakefulness state and exiting a DoC results in an AUC of 0.91 and 0.88 respectively. Small but useful improvements are added by the EEG dataset to the clinical model for what concerns overcoming an unresponsive wakefulness state. Data-driven techniques and namely, machine learning models are hereby shown to be capable of supporting the complex decision-making process the practitioners must face.
The use of stereophotogrammetry systems is challenging when targeting children's gait analysis due to the time required and the need to keep physical markers in place. For this reason, marker-less ...photoelectric systems appear to be a solution for accurate and fast gait analysis in youth. The aim of this study is to validate a photoelectric system and its configurations (LED filter setting) on healthy children, comparing the kinematic gait parameters with those obtained from a three-dimensional stereophotogrammetry system. Twenty-seven healthy children were enrolled. Three LED filter settings for the OptoGait were compared to the BTS P6000. The analysis included the non-parametric 80% limits of agreement and the intraclass correlation coefficient (ICC). Additionally, normalised limits of agreement and bias (NLoAs and Nbias) were compared to the clinical experience of physical therapists (i.e., assuming an error lower than 5% is acceptable). ICCs showed excellent consistency for most of the parameters and filter settings; NLoAs varied between 1.39% and 12.62%. An inverse association between the number of LEDs for filter setting and the bias values was also observed. Observations confirm the validity of the OptoGait system for the evaluation of spatiotemporal gait parameters in children.
Intensive treadmill training (TT) has been documented to improve gait parameters and functional independence in Parkinson's Disease (PD), but the optimal intervention protocol and the criteria for ...tailoring the intervention to patients' performances are lacking. TT may be integrated with augmented virtual reality (AVR), however, evidence of the effectiveness of this combined treatment is still limited. Moreover, prognostic biomarkers of rehabilitation, potentially useful to customize the treatment, are currently missing. The primary aim of this study is to compare the effects on gait performances of TT + AVR versus TT alone in II-III stage PD patients with gait disturbance. Secondary aims are to assess the effects on balance, gait parameters and other motor and non-motor symptoms, and patient's satisfaction and adherence to the treatment. As an exploratory aim, the study attempts to identify biomarkers of neuroplasticity detecting changes in Neurofilament Light Chain concentration T0-T1 and to identify prognostic biomarkers associated to blood-derived Extracellular Vesicles.
Single-center, randomized controlled single-blind trial comparing TT + AVR vs. TT in II-III stage PD patients with gait disturbances. Assessment will be performed at baseline (T0), end of training (T1), 3 (T2) and 6 months (T3, phone interview) from T1. The primary outcome is difference in gait performance assessed with the Tinetti Performance-Oriented Mobility Assessment gait scale at T1. Secondary outcomes are differences in gait performance at T2, in balance and spatial-temporal gait parameters at T1 and T2, patients' satisfaction and adherence. Changes in falls, functional mobility, functional autonomy, cognition, mood, and quality of life will be also assessed at different timepoints. The G*Power software was used to estimate a sample size of 20 subjects per group (power 0.95,
< 0.05), raised to 24 per group to compensate for potential drop-outs. Both interventions will be customized and progressive, based on the participant's performance, according to a predefined protocol.
This study will provide data on the possible superiority of AVR-associated TT over conventional TT in improving gait and other motor and non-motor symptoms in persons with PD and gait disturbances. Results of the exploratory analysis could add information in the field of biomarker research in PD rehabilitation.
•Three microstate maps (A, B, and C) sufficiently represent sABI EEG recordings.•Occurrence of frontally-oriented microstates were indicators of consciousness level and presence.•Transitioning from ...left-right to antero-posterior dynamics correlates with consciousness.•Inexpensive, non-invasive markers of spatio-temporal brain activity may complement diagnosis.
Consciousness can be defined as a phenomenological experience continuously evolving. Current research showed how conscious mental activity can be subdivided into a series of atomic brain states converging to a discrete spatiotemporal pattern of global neuronal firing. Using the high temporal resolution of EEG recordings in patients with a severe Acquired Brain Injury (sABI) admitted to an Intensive Rehabilitation Unit (IRU), we detected a novel endotype of consciousness from the spatiotemporal brain dynamics identified via microstate analysis. Also, we investigated whether microstate features were associated with common neurophysiological alterations. Finally, the prognostic information comprised in such descriptors was analysed in a sub-cohort of patients with prolonged Disorder of Consciousness (pDoC). Occurrence of frontally-oriented microstates (C microstate), likelihood of maintaining such brain state or transitioning to the C topography and complexity were found to be indicators of consciousness presence and levels. Features of left–right asymmetric microstates and transitions toward them were found to be negatively correlated with antero-posterior brain reorganization and EEG symmetry. Substantial differences in microstates’ sequence complexity and presence of C topography were found between groups of patients with alpha dominant background, cortical reactivity and antero-posterior gradient. Also, transitioning from left-right to antero-posterior microstates was found to be an independent predictor of consciousness recovery, stronger than consciousness levels at IRU’s admission. In conclusions, global brain dynamics measured with scale-free estimators can be considered an indicator of consciousness presence and a candidate marker of short-term recovery in patients with a pDoC.
Due to continuous advances in intensive care technology and neurosurgical procedures, the number of survivors from severe acquired brain injuries (sABIs) has increased considerably, raising several ...delicate ethical issues. The heterogeneity and complex nature of the neurological damage of sABIs make the detection of predictive factors of a better outcome very challenging. Identifying the profile of those patients with better prospects of recovery will facilitate clinical and family choices and allow to personalize rehabilitation. This paper describes a multicenter prospective study protocol, to investigate outcomes and baseline predictors or biomarkers of functional recovery, on a large Italian cohort of sABI survivors undergoing postacute rehabilitation.
All patients with a diagnosis of sABI admitted to four intensive rehabilitation units (IRUs) within 4 months from the acute event, aged above 18, and providing informed consent, will be enrolled. No additional exclusion criteria will be considered. Measures will be taken at admission (T0), at three (T1) and 6 months (T2) from T0, and follow-up at 12 and 24 months from onset, including clinical and functional data, neurophysiological results, and analysis of neurogenetic biomarkers.
Advanced machine learning algorithms will be cross validated to achieve data-driven prediction models. To assess the clinical applicability of the solutions obtained, the prediction of recovery milestones will be compared to the evaluation of a multiprofessional, interdisciplinary rehabilitation team, performed within 2 weeks from admission.
Identifying the profiles of patients with a favorable prognosis would allow customization of rehabilitation strategies, to provide accurate information to the caregivers and, possibly, to optimize rehabilitation outcomes.
The application and validation of machine learning algorithms on a comprehensive pool of clinical, genetic, and neurophysiological data can pave the way toward the implementation of tools in support of the clinical prognosis for the rehabilitation pathways of patients after sABI.
The complex nature of stroke sequelae, the heterogeneity in rehabilitation pathways, and the lack of validated prediction models of rehabilitation outcomes challenge stroke rehabilitation quality ...assessment and clinical research. An integrated care pathway (ICP), defining a reproducible rehabilitation assessment and process, may provide a structured frame within investigated outcomes and individual predictors of response to treatment, including neurophysiological and neurogenetic biomarkers. Predictors may differ for different interventions, suggesting clues to personalize and optimize rehabilitation. To date, a large representative Italian cohort study focusing on individual variability of response to an evidence-based ICP is lacking, and predictors of individual response to rehabilitation are largely unexplored. This paper describes a multicenter study protocol to prospectively investigate outcomes and predictors of response to an evidence-based ICP in a large Italian cohort of stroke survivors undergoing post-acute inpatient rehabilitation.
All patients with diagnosis of ischemic or hemorrhagic stroke confirmed both by clinical and brain imaging evaluation, admitted to four intensive rehabilitation units (adopting the same stroke rehabilitation ICP) within 30 days from the acute event, aged 18+, and providing informed consent will be enrolled (expected sample: 270 patients). Measures will be taken at admission (T0), at discharge (T1), and at follow-up 6 months after a stroke (T2), including clinical data, nutritional, functional, neurological, and neuropsychological measures, electroencephalography and motor evoked potentials, and analysis of neurogenetic biomarkers.
In addition to classical multivariate logistic regression analysis, advanced machine learning algorithms will be cross-validated to achieve data-driven prognosis prediction models.
By identifying data-driven prognosis prediction models in stroke rehabilitation, this study might contribute to the development of patient-oriented therapy and to optimize rehabilitation outcomes.
ClinicalTrials.gov, NCT03968627. https://www.clinicaltrials.gov/ct2/show/NCT03968627?term=Cecchi&cond=Stroke&draw=2&rank=2.