Search

BackgroundDisability progression is a key milestone in the disease evolution of people with multiple sclerosis (PwMS). Prediction models of disability progression have not yet reached the level of trust needed to be adopted in the clinic. A common benchmark to assess model development in multiple sclerosis is also currently lacking.MethodsData of adult PwMS with a follow-up of at least three years from 146 MS centers, spread over 40 countries and collected by the MSBase consortium was used. With basic inclusion criteria for quality requirements, it represents a total of 15, 240 PwMS. External validation was performed and repeated five times to assess the significance of the results. TRIPOD guidelines were followed.Confirmed disability progression after two years was predicted, with a confirmation window of six months. Only routinely collected variables were used such as the expended disability status scale, treatment, relapse information, and MS course.To learn the probability of disability progression, state-of-the-art machine learning models were investigated. The discrimination performance of the models is evaluated on their area under the receiver operator curve (ROC-AUC) and under the precision recall curve (AUC-PR), and their calibration via the Brier score and the expected calibration error.FindingsA temporal attention model was the best model. It achieved a ROC-AUC of 0·71 ± 0·01, an AUC-PR of 0·26 ± 0·02, a Brier score of 0·1 ± 0·01 and an expected calibration error of 0·07 ± 0·04. The history of disability progression is more predictive for future disability progression than the treatment or relapses.InterpretationGood discrimination and calibration performance on an external validation set is achieved, using only routinely collected variables. This makes these models ready for a clinical impact study. All our preprocessing and model code is available at https://gitlab.com/edebrouwer/ms_benchmark, making this task an ideal benchmark for predicting disability progression in MS.

Multiple sclerosis (MS) is a chronic disease affecting millions of people worldwide. Through the demyelinating and axonal pathology of MS, the signal conduction in the central nervous system is affected. Evoked potential measurements allow clinicians to monitor this process and can be used for decision support. We share a dataset that contains motor evoked potential (MEP) measurements, in which the brain is stimulated and the resulting signal is measured in the hands and feet. This results in time series of 100 milliseconds long. Typically, both hands and feet are measured in one hospital visit. The dataset contains 5586 visits of 963 patients, performed in day-to-day clinical care over a period of 6 years. The dataset consists of approximately 100,000 MEP. Clinical metadata such as the expanded disability status scale, sex, and age is also available. This dataset can be used to explore the role of evoked potentials in MS research and patient care. It may also be used as a benchmark for time series analysis and predictive modelling. Te authors would like to thank the personnel working at the MS clinic for their help while collecting the data. In particular, we would like to thank Henny Strackx for his help with extracting the data from the local servers, Daisy Mobers for her input on how the measurements were performed, and Filip Van Elsen for his input on the technical aspects of the measurement devices used. Furthermore, we thank Patricia Tielens, Dieter Flament, An Voets, Jorina Nickmans and Benny Daems for their legal counsel and Wim Lamotte for answering our questions about the security measures that had to be taken for data transfer. Finally, we thank P95 for their work in assessing the re-identifcation risk of the dataset. Te computational resources and services used in this work were provided by the VSC (Flemish Supercomputer Center), funded by the Research Foundation - Flanders (FWO) and the Flemish Government - department EWI. Tis research received funding from the Flemish Government under the “Onderzoeksprogramma Artifciële Intelligentie (AI) Vlaanderen” programme. This work was supported by Research Foundation—Flanders (FWO) for ELIXIR Belgium (I002819N), and Hermesfonds for ELIXIR Belgium (AH.2017.051, IO 17001306).

Background Evoked potentials (EPs) are a measure of the conductivity of the central nervous system. They are used to monitor disease progression of multiple sclerosis patients. Previous studies only extracted a few variables from the EPs, which are often further condensed into a single variable: the EP score. We perform a machine learning analysis of motor EP that uses the whole time series, instead of a few variables, to predict disability progression after two years. Obtaining realistic performance estimates of this task has been difficult because of small data set sizes. We recently extracted a dataset of EPs from the Rehabiliation & MS Center in Overpelt, Belgium. Our data set is large enough to obtain, for the first time, a performance estimate on an independent test set containing different patients. Methods We extracted a large number of time series features from the motor EPs with the highly comparative time series analysis software package. Mutual information with the target and the Boruta method are used to find features which contain information not included in the features studied in the literature. We use random forests (RF) and logistic regression (LR) classifiers to predict disability progression after two years. Statistical significance of the performance increase when adding extra features is checked. Results Including extra time series features in motor EPs leads to a statistically significant improvement compared to using only the known features, although the effect is limited in magnitude (ΔAUC = 0.02 for RF and ΔAUC = 0.05 for LR). RF with extra time series features obtains the best performance (AUC = 0.75±0.07 (mean and standard deviation)), which is good considering the limited number of biomarkers in the model. RF (a nonlinear classifier) outperforms LR (a linear classifier). Conclusions Using machine learning methods on EPs shows promising predictive performance. Using additional EP time series features beyond those already in use leads to a modest increase in performance. Larger datasets, preferably multi-center, are needed for further research. Given a large enough dataset, these models may be used to support clinicians in their decision making process regarding future treatment.

Neural networks for automated image segmentation are typically trained to achieve maximum accuracy, while less attention has been given to the calibration of their confidence scores. However, well-calibrated confidence scores provide valuable information towards the user. We investigate several post hoc calibration methods that are straightforward to implement, some of which are novel. They are compared to Monte Carlo (MC) dropout. They are applied to neural networks trained with cross-entropy (CE) and soft Dice (SD) losses on BraTS 2018 and ISLES 2018. Surprisingly, models trained on SD loss are not necessarily less calibrated than those trained on CE loss. In all cases, at least one post hoc method improves the calibration. There is limited consistency across the results, so we can't conclude on one method being superior. In all cases, post hoc calibration is competitive with MC dropout. Although average calibration improves compared to the base model, subject-level variance of the calibration remains similar., Comment: 6 pages, 3 figures, submitted to ISBI'21

BACKGROUND AND OBJECTIVES: Research in Multiple Sclerosis (MS) has recently focused on extracting knowledge from real-world clinical data sources. This type of data is more abundant than data produced during clinical trials and potentially more informative about real-world clinical practice. However, this comes at the cost of less curated and controlled data sets. In this work we aim to predict disability progression by optimally extracting information from longitudinal patient data in the real-world setting, with a special focus on the sporadic sampling problem. METHODS: We use machine learning methods suited for patient trajectories modeling, such as recurrent neural networks and tensor factorization. A subset of 6682 patients from the MSBase registry is used. RESULTS: We can predict disability progression of patients in a two-year horizon with an ROC-AUC of 0.85, which represents a 32% decrease in the ranking pair error (1-AUC) compared to reference methods using static clinical features. CONCLUSIONS: Compared to the models available in the literature, this work uses the most complete patient history for MS disease progression prediction and represents a step forward towards AI-assisted precision medicine in MS. ispartof: COMPUTER METHODS AND PROGRAMS IN BIOMEDICINE vol:208 ispartof: location:Ireland status: published

In a recent paper Arita et al. [Phys. Rev. E 90, 052108 (2014)] consider the transport properties of a class of generalized exclusion processes. Analytical expressions for the transport-diffusion coefficient are derived by ignoring correlations. It is claimed that these expressions become exact in the hydrodynamic limit. In this Comment, we point out that (i) the influence of correlations upon the diffusion does not vanish in the hydrodynamic limit, and (ii) the expressions for the self- and transport diffusion derived by Arita et al. are special cases of results derived in [Phys. Rev. Lett. 111, 110601 (2013)]., (citation added, published version)

Detailed fluctuation theorems are statements about the probability distribution for the stochastic entropy production along a trajectory. It involves the consideration of a suitably transformed dynamics, such as the time reversed, the adjoint, or a combination of these. We identify specific, typically unique, initial conditions, called echo states, for which the final probability distribution of the transformed dynamics reproduces the initial distribution. In this case the detailed fluctuation theorems relate the stochastic entropy production of the direct process to that of the transformed one. We illustrate our results by an explicit analytical calculation and numerical simulations for a modulated two-state quantum dot., Comment: 8 pages, 6 figures, published version

We study the adsorption and desorption kinetics of interacting particles moving on a one-dimensional lattice. Confinement is introduced by limiting the number of particles on a lattice site. Adsorption and desorption are found to proceed at different rates, and are strongly influenced by the concentration-dependent transport diffusion. Analytical solutions for the transport and self-diffusion are given for systems of length 1 and 2 and for a zero-range process. In the last situation the self- and transport diffusion can be calculated analytically for any length., Published in EPJ ST volume "Brownian Motion in Confined Geometries"

We expand on a recent study of a lattice model of interacting particles [Phys. Rev. Lett. 111, 110601 (2013)]. The adsorption isotherm and equilibrium fluctuations in particle number are discussed as a function of the interaction. Their behavior is similar to that of interacting particles in porous materials. Different expressions for the particle jump rates are derived from transition state theory. Which expression should be used depends on the strength of the inter-particle interactions. Analytical expressions for the self- and transport diffusion are derived when correlations, caused by memory effects in the environment, are neglected. The diffusive behavior is studied numerically with kinetic Monte Carlo (kMC) simulations, which reproduces the diffusion including correlations. The effect of correlations is studied by comparing the analytical expressions with the kMC simulations. It is found that the Maxwell-Stefan diffusion can exceed the self-diffusion. To our knowledge, this is the first time this is observed. The diffusive behavior in one-dimensional and higher dimensional systems is qualitatively the same, with the effect of correlations decreasing for increasing dimension. The length dependence of both the self- and transport diffusion is studied for one-dimensional systems. For long lengths the self-diffusion shows a one over length dependence. Finally, we discuss when agreement with experiments and simulations can be expected. The assumption that particles in different cavities do not interact is expected to hold quantitatively at low and medium particle concentrations, if the particles are not strongly interacting., Comment: (18 pages, 16 figures, published version)

We evaluate the self-diffusion and transport diffusion of interacting particles in a discrete geometry consisting of a linear chain of cavities, with interactions within a cavity described by a free-energy function. Exact analytical expressions are obtained in the absence of correlations, showing that the self-diffusion can exceed the transport diffusion if the free-energy function is concave. The effect of correlations is elucidated by comparison with numerical results. Quantitative agreement is obtained with recent experimental data for diffusion in a nanoporous zeolitic imidazolate framework material, ZIF-8., Comment: 5 pages main text (3 figures); 9 pages supplemental material (2 figures). (minor changes, published version)

We use kinetic Monte Carlo simulations to investigate current fluctuations in boundary driven generalized exclusion processes, in different dimensions. Simulation results are in full agreement with predictions based on the additivity principle and the macroscopic fluctuation theory. The current statistics are independent of the shape of the contacts with the reservoirs, provided they are macroscopic in size. In general, the current distribution depends on the spatial dimension. For the special cases of the symmetric simple exclusion process and the zero-range process, the current statistics are the same for all spatial dimensions., 21 pages, 11 figures, published version