Search

IntroductionClassical deep learning research requires lots of centralised data. However, data sets are often stored at different clinical centers, and sharing sensitive patient data such as brain images is difficult. In this manuscript, we investigated the feasibility of federated learning, sending models to the data instead of the other way round, for research on brain magnetic resonant images of people with multiple sclerosis (MS).MethodsUsing transfer learning on a previously published brain age model, we trained a model to decode performance on the symbol digit modalities test (SDMT) of patients with MS from structural T1 weighted MRI. Three international centers in Brussels, Greifswald and Prague participated in the project. In Brussels, one computer served as the server coordinating the FL project, while the other served as client for model training on local data (n=97). The other two clients were Greifswald (n=104) and Prague (n=100). Each FL round, the server sent a global model to the clients, where its fully connected layer was updated on the local data. After collecting the local models, the server applied a weighted average of two randomly picked clients, yielding a new global model.ResultsAfter 22 federated learning rounds, the average validation loss across clients reached a minimum. The model appeared to have learned to assign SDMT values close to the mean with a mean absolute error of 9.04, 10.59 and 10.71 points between true and predicted SDMT on the test data sets of Brussels, Greifswald and Prague respectively. The overall test MAE across all clients was 10.13 points.ConclusionFederated learning is feasible for machine learning research on brain MRI of persons with MS, setting the stage for larger transfer learning studies to investigate the utility of brain age latent representations in cognitive decoding tasks.

BACKGROUND AND PURPOSE: Data from neuro-imaging techniques allow us to estimate a brain's age. Brain age is easily interpretable as 'how old the brain looks' and could therefore be an attractive communication tool for brain health in clinical practice. This study aimed to investigate its clinical utility by investigating the relationship between brain age and cognitive performance in multiple sclerosis (MS). METHODS: A linear regression model was trained to predict age from brain magnetic resonance imaging volumetric features and sex in a healthy control dataset (HC_train, n = 1673). This model was used to predict brain age in two test sets: HC_test (n = 50) and MS_test (n = 201). Brain-predicted age difference (BPAD) was calculated as BPAD = brain age minus chronological age. Cognitive performance was assessed by the Symbol Digit Modalities Test (SDMT). RESULTS: Brain age was significantly related to SDMT scores in the MS_test dataset (r = -0.46, p

Graph-theoretical analysis is a novel tool to understand the organisation of the brain.We assessed whether altered graph theoretical parameters, as observed in multiple sclerosis (MS), reflect pathology-induced restructuring of the brain's functioning or result from a reduced signal quality in functional MRI (fMRI). In a cohort of 49 people with MS and a matched group of 25 healthy subjects (HS), we performed a cognitive evaluation and acquired fMRI. From the fMRI measurement, Pearson correlation-based networks were calculated and graph theoretical parameters reflecting global and local brain organisation were obtained. Additionally, we assessed metrics of scanning quality (signal to noise ratio (SNR)) and fMRI signal quality (temporal SNR and contrast to noise ratio (CNR)). In accordance with the literature, we found that the network parameters were altered in MS compared to HS. However, no significant link was found with cognition. Scanning quality (SNR) did not differ between both cohorts. In contrast, measures of fMRI signal quality were significantly different and explained the observed differences in GTA parameters. Our results suggest that differences in network parameters between MS and HS in fMRI do not reflect a functional reorganisation of the brain, but rather occur due to reduced fMRI signal quality.

BackgroundData from neuro-imaging techniques allow us to estimate a brain’s age. Brain age is easily interpretable as “how old the brain looks”, and could therefore be an attractive communication tool for brain health in clinical practice. This study aimed to investigate its clinical utility.ObjectivesTo investigate the relationship between brain age and information processing speed in MS.MethodsA ridge-regression model was trained to predict age from brain MRI volumetric features and sex in a healthy control dataset (HC_train, n=1690). This model was used to predict brain age in two test sets: HC_test (n=50) and MS_test (n=201). Brain-Predicted Age Difference (BPAD) was calculated as BPAD=brain age minus chronological age. Information processing speed was assessed with the Symbol Digit Modalities Test (SDMT).ResultsBrain age was significantly related to SDMT scores in the MS_test dataset (r=-0.44, pConclusionsBrain age is a candidate biomarker for information processing speed in MS and an easy to grasp metric for brain health.

This is a protocol for a Cochrane Review (intervention). The objectives are as follows: 1) To determine the effectiveness of virtual reality interventions compared with alternative or no intervention in people with Multiple Sclerosis (MS) on: gait and balance function, 2) To determine the effectiveness of virtual reality interventions compared with alternative or no intervention in people with MS on upper limb function; cognitive function; fatigue; global motor function; activity limitation; participation restriction and quality of life and adverse events.

This is a protocol for a Cochrane Review (intervention). The objectives are as follows: To determine the effectiveness of trunk training after stroke in comparison to no therapy or to control therapy, either dose‐matched or not. The review will assign the outcome measures to the levels defined in the ICF model and the review will evaluate the effect of different types of trunk training.

Background Falls are one of the most common complications after stroke, with a reported incidence ranging between 7% in the first week and 73% in the first year post stroke. This is an updated version of the original Cochrane Review published in 2013. Objectives To evaluate the effectiveness of interventions aimed at preventing falls in people after stroke. Our primary objective was to determine the effect of interventions on the rate of falls (number of falls per person-year) and the number of fallers. Our secondary objectives were to determine the effects of interventions aimed at preventing falls on 1) the number of fall-related fractures; 2) the number of fall-related hospital admissions; 3) near-fall events; 4) economic evaluation; 5) quality of life; and 6) adverse effects of the interventions. Search methods We searched the trials registers of the Cochrane Stroke Group (September 2018) and the Cochrane Bone, Joint and Muscle Trauma Group (October 2018); the Cochrane Central Register of Controlled Trials (CENTRAL; 2018, Issue 9) in the Cochrane Library; MEDLINE (1950 to September 2018); Embase (1980 to September 2018); CINAHL (1982 to September 2018); PsycINFO (1806 to August 2018); AMED (1985 to December 2017); and PEDro (September 2018). We also searched trials registers and checked reference lists. Selection criteria Randomised controlled trials of interventions where the primary or secondary aim was to prevent falls in people after stroke. Data collection and analysis Two review authors (SD and WS) independently selected studies for inclusion, assessed trial quality and risk of bias, and extracted data. We resolved disagreements through discussion, and contacted study authors for additional information where required. We used a rate ratio and 95% confidence interval (CI) to compare the rate of falls (e.g. falls per person-year) between intervention and control groups. For risk of falling we used a risk ratio and 95% CI based on the number of people falling (fallers) in each group. We pooled results where appropriate and applied GRADE to assess the quality of the evidence. Main results We included 14 studies (of which six have been published since the first version of this review in 2013), with a total of 1358 participants. We found studies that investigated exercises, predischarge home visits for hospitalised patients, the provision of single lens distance vision glasses instead of multifocal glasses, a servo-assistive rollator and non-invasive brain stimulation for preventing falls. Exercise compared to control for preventing falls in people after stroke The pooled result of eight studies showed that exercise may reduce the rate of falls but we are uncertain about this result (rate ratio 0.72, 95% CI 0.54 to 0.94, 765 participants, low-quality evidence). Sensitivity analysis for single exercise interventions, omitting studies using multiple/multifactorial interventions, also found that exercise may reduce the rate of falls (rate ratio 0.66, 95% CI 0.50 to 0.87, 626 participants). Sensitivity analysis for the effect in the chronic phase post stroke resulted in little or no difference in rate of falls (rate ratio 0.58, 95% CI 0.31 to 1.12, 205 participants). A sensitivity analysis including only studies with low risk of bias found little or no difference in rate of falls (rate ratio 0.88, 95% CI 0.65 to 1.20, 462 participants). Methodological limitations mean that we have very low confidence in the results of these sensitivity analyses. For the outcome of number of fallers, we are very uncertain of the effect of exercises compared to the control condition, based on the pooled result of 10 studies (risk ratio 1.03, 95% CI 0.90 to 1.19, 969 participants, very low quality evidence). The same sensitivity analyses as described above gives us very low certainty that there are little or no differences in number of fallers (single interventions: risk ratio 1.09, 95% CI 0.93 to 1.28, 796 participants; chronic phase post stroke: risk ratio 0.94, 95% CI 0.73 to 1.22, 375 participants; low risk of bias studies: risk ratio 0.96, 95% CI 0.77 to 1.21, 462 participants). Other interventions for preventing falls in people after stroke We are very uncertain whether interventions other than exercise reduce the rate of falls or number of fallers. We identified very low certainty evidence when investigating the effect of predischarge home visits (rate ratio 0.85, 95% CI 0.43 to 1.69; risk ratio 1.48, 95% CI 0.71 to 3.09; 85 participants), provision of single lens distance glasses to regular wearers of multifocal glasses (rate ratio 1.08, 95% CI 0.52 to 2.25; risk ratio 0.74, 95% CI 0.47 to 1.18; 46 participants) and a servo-assistive rollator (rate ratio 0.44, 95% CI 0.16 to 1.21; risk ratio 0.44, 95% CI 0.16 to 1.22; 42 participants). Finally, transcranial direct current stimulation (tDCS) was used in one study to examine the effect on falls post stroke. We have low certainty that active tDCS may reduce the number of fallers compared to sham tDCS (risk ratio 0.30, 95% CI 0.14 to 0.63; 60 participants). Authors' conclusions At present there exists very little evidence about interventions other than exercises to reduce falling post stroke. Low to very low quality evidence exists that this population benefits from exercises to prevent falls, but not to reduce number of fallers. Fall research does not in general or consistently follow methodological gold standards, especially with regard to fall definition and time post stroke. More well-reported, adequately-powered research should further establish the value of exercises in reducing falling, in particular per phase, post stroke. ispartof: Cochrane Database Of Systematic Reviews vol:10 issue:10 ispartof: location:England status: published

Multiple sclerosis (MS) manifests heterogeneously among persons suffering from it, making its disease course highly challenging to predict. At present, prognosis mostly relies on biomarkers that are unable to predict disease course on an individual level. Machine learning is a promising technique, both in terms of its ability to combine multimodal data and through the capability of making personalized predictions. However, most investigations on machine learning for prognosis in MS were geared towards predicting physical deterioration, while cognitive deterioration, although prevalent and burdensome, remained largely overlooked. This review aims to boost the field of machine learning for cognitive prognosis in MS by means of an introduction to machine learning and its pitfalls, an overview of important elements for study design, and an overview of the current literature on cognitive prognosis in MS using machine learning. Furthermore, the review discusses new trends in the field of machine learning that might be adopted for future studies in the field. ispartof: JOURNAL OF PERSONALIZED MEDICINE vol:11 issue:12 ispartof: location:Switzerland status: published

Background: Cognitive dysfunction is a frequent manifestation of multiple sclerosis (MS) but its effect on locomotor rehabilitation is unknown. Objective: To study the impact of cognitive impairment on locomotor rehabilitation outcome in people with MS. Methods: We performed a retrospective analysis involving ambulatory patients with MS who were admitted for intensive, inpatient, multidisciplinary rehabilitation at the National Multiple Sclerosis Center of Melsbroek between the years 2012 and 2017. The Brief Repeatable Battery of Neuropsychological Tests (BRB-N) was used to determine the cognitive status of subjects as either impaired (COG–) or preserved (COG+). Locomotor outcome was compared between groups with the difference in 6-minute walk test (6MWT) measured at admission and discharge (Δ6MWT). In addition, individual test scores of the BRB-N for attention (Paced Auditory Serial Addition Test 2” and 3”), visuospatial learning/memory (7/24 Spatial Recall Test), verbal learning/memory (Selective Reminding Test) and verbal fluency (Controlled Oral Word Association Test) were correlated to the Δ6MWT. Results: A total of 318 complete and unique records were identified. Both groups showed a significant within-group Δ6MWT during hospitalization (COG+: 47.51 m; COG–: 40.97 m; P Conclusion: Cognitive impairment based on BRB-N results appears not to impede locomotor rehabilitation in ambulatory patients with MS. Attentional deficits are correlated to the extent of locomotor rehabilitation, suggesting the presence of a subtle effect of cognition.