An interpretable machine learning model to explain the interplay between brain lesions and cortical atrophy in multiple sclerosis.

Multiple Sclerosis (MS) is the most common cause, (after trauma) of neurological disability in young adults in Western countries. While several Magnetic Resonance Imaging (MRI) studies have demonstrated a strong association between the presence of cortical grey matter atrophy and the progression of neurological impairment in MS patients, the neurobiological substrates of cortical atrophy in MS, and in particular its relationship with white matter (WM) and cortical lesions, remain unknown. The aim of this study was to investigate the interplay between cortical atrophy and different types of lesions at Ultra-High Field (UHF) 7 T MRI, including cortical lesions and lesions with a susceptibility rim (a feature which histopathological studies have associated with impaired remyelination and progressive tissue destruction). We combined lesion characterization with a recent machine learning (ML) framework which includes explainability, and we were able to predict cortical atrophy in MS from a handful of lesion-related features extracted from 7 T MR imaging. This highlights not only the importance of UHF MRI for accurately evaluating intracortical and rim lesion load, but also the differential contributions that these types of lesions may bring to determine disease evolution and severity. Also, we found that a small subset of features [WM lesion volume (not considering rim lesions), patient age and WM lesion count (not considering rim lesions), intracortical lesion volume] carried most of the prediction power. Interestingly, an almost opposite pattern emerged when contrasting cortical with WM lesion load: WM lesion load is most important when it is small, whereas cortical lesion load behaves in the opposite way.Clinical Relevance- Our results suggest that disconnection and axonal degeneration due to WM lesions and local cortical demyelination are the main factors determining cortical thinning. These findings further elucidate the complexity of MS pathology across the whole brain and the need for both statistical and mechanistic approaches to understanding the etiopathogenesis of lesions.

Whole-brain white matter network reorganization in HIV.

The human immunodeficiency virus (HIV) causes an infectious disease with a high viral tropism toward CD4 T-lymphocytes and macrophage. Since the advent of combined antiretroviral therapy (CART), the number of opportunistic infectious disease has diminished, turning HIV into a chronic condition. Nevertheless, HIV-infected patients suffer from several life-long symptoms, including the HIV-associated neurocognitive disorder (HAND), whose biological substrates remain unclear. HAND includes a range of cognitive impairments which have a huge impact on daily patient life. The aim of this study was to examine putative structural brain network changes in HIV-infected patient to test whether diffusion-imaging-related biomarkers could be used to discover and characterize subtle neurological alterations in HIV infection. To this end, we employed multi-shell, multi-tissue constrained spherical deconvolution in conjunction with probabilistic tractography and graph-theoretical analyses. We found several statistically significant effects in both local (right postcentral gyrus, right precuneus, right inferior parietal lobule, right transverse temporal gyrus, right inferior temporal gyrus, right putamen and right pallidum) and global graph-theoretical measures (global clustering coefficient, global efficiency and transitivity). Our study highlights a global and local reorganization of the structural connectome which support the possible application of graph theory to detect subtle alteration of brain regions in HIV patients.Clinical Relevance-Brain measures able to detect subtle alteration in HIV patients could also be used in e.g. evaluating therapeutic responses, hence empowering clinical trials.

Is thulium laser vapoenucleation of the prostate equally safe and effective in elderly patients? A propensity score matched analysis of early perioperative and functional outcomes. / Vapoenucleación prostática con láser Tulio. ¿Es igual de segura y efectiva en pacientes ancianos? Análisis por puntuación de propensión de los resultados funcionales y perioperatorios tempranos.

PURPOSE: To evaluate if thulium laser vapoenucleation of the prostate (ThuVEP) is equally safe and effective in a selected cohort of elderly patients when compared to "younger" patients. MATERIALS AND METHODS: We performed a retrospective analysis of consecutive patients who underwent ThuVEP between September 2018 and February 2020. After application of the inclusion/exclusion criteria, patients were stratified according to the 75 years-old cut-off point suggested by the WHO. Group A included patients < 75 years-old; Group B included patients > 75 years-old. Preoperative assessment included urological consultation, prostate specific antigen (PSA), International Prostate Symptom Score (IPSS) and quality of life index, transrectal ultrasound to estimate prostate volume (PVol), and uroflowmetry to assess preoperative Qmax, Qave and post-void residual volume (PVR). Perioperative and postoperative data were analyzed during 3-month follow-up. RESULTS: After propensity-score analysis, 51 versus 51 patients were 1:1 matched according to PVol, PSA, Qmax, IPSS and QoL. Patients were comparable at baseline excluding age (65 (IQR 59-70) versus 79 (IQR 77-82) years, Group A versus B, respectively, p-value < 0.001). No differences were found in terms of hemoglobin drop, complications rate, catheterization time and length of hospital stay. Group A (younger) patients had more significant improvement in 30-days absolute Qmax, Qave and ΔQmax. At 90-days follow-up, the differences between the groups disappeared. Within the 90-days follow-up, no significant differences were found in the readmission rate, with no need of reinterventions. CONCLUSIONS: In our hands, even in elderly patients affected by BPH, ThuVEP appears to be a safe and effective treatment option.

Disruption of structural brain networks in Primary Open Angle Glaucoma.

Primary open angle glaucoma (POAG) is one of the most common causes of permanent blindness in the world. Recent studies have originated the hypothesis that POAG could be considered as a central nervous system pathology which results in secondary visual involvement. The aim of this study is to assess possible structural whole brain connectivity alterations in POAG by combining multi-shell diffusion weighted imaging, multi-shell multi-tissue probabilistic tractography, graph theoretical measures and a newly designed disruption index, which evaluates the global reorganization of brain networks in group-wise comparisons. We found global differences in structural connectivity between Glaucoma patients and controls, as well as in local graph theoretical measures. These changes extended well beyond the primary visual pathway. Furthermore, group-wise and subject-wise disruption indices were found to be statistically different between glaucoma patients and controls, with a positive slope. Overall, our results support the hypothesis of a whole-brain structural reorganization in glaucoma which is specific to structural connectivity, possibly placing this disease within the recently defined groups of brain disconnection syndrome.

Time-resolved connectome of the five-factor model of personality.

The human brain is characterized by highly dynamic patterns of functional connectivity. However, it is unknown whether this time-variant 'connectome' is related to the individual differences in the behavioural and cognitive traits described in the five-factor model of personality. To answer this question, inter-network time-variant connectivity was computed in n = 818 healthy people via a dynamical conditional correlation model. Next, network dynamicity was quantified throughout an ad-hoc measure (T-index) and the generalizability of the multi-variate associations between personality traits and network dynamicity was assessed using a train/test split approach. Conscientiousness, reflecting enhanced cognitive and emotional control, was the sole trait linked to stationary connectivity across several circuits such as the default mode and prefronto-parietal network. The stationarity in the 'communication' across large-scale networks offers a mechanistic description of the capacity of conscientious people to 'protect' non-immediate goals against interference over-time. This study informs future research aiming at developing more realistic models of the brain dynamics mediating personality differences.

The central autonomic network at rest: Uncovering functional MRI correlates of time-varying autonomic outflow.

Peripheral measures of autonomic nervous system (ANS) activity at rest have been extensively employed as putative biomarkers of autonomic cardiac control. However, a comprehensive characterization of the brain-based central autonomic network (CAN) sustaining cardiovascular oscillations at rest is missing, limiting the interpretability of these ANS measures as biomarkers of cardiac control. We evaluated combined cardiac and fMRI data from 34 healthy subjects from the Human Connectome Project to detect brain areas functionally linked to cardiovagal modulation at rest. Specifically, we combined voxel-wise fMRI analysis with instantaneous heartbeat and spectral estimates obtained from inhomogeneous linear point-process models. We found exclusively negative associations between cardiac parasympathetic activity at rest and a widespread network including bilateral anterior insulae, right dorsal middle and left posterior insula, right parietal operculum, bilateral medial dorsal and ventrolateral posterior thalamic nuclei, anterior and posterior mid-cingulate cortex, medial frontal gyrus/pre-supplementary motor area. Conversely, we found only positive associations between instantaneous heart rate and brain activity in areas including frontopolar cortex, dorsomedial prefrontal cortex, anterior, middle and posterior cingulate cortices, superior frontal gyrus, and precuneus. Taken together, our data suggests a much wider involvement of diverse brain areas in the CAN at rest than previously thought, which could reflect a differential (both spatially and directionally) CAN activation according to the underlying task. Our insight into CAN activity at rest also allows the investigation of its impairment in clinical populations in which task-based fMRI is difficult to obtain (e.g., comatose patients or infants).

Resting State Neural Correlates of Cardiac Sympathetic Dynamics in Healthy Subjects.

Recent advances in functional Magnetic Resonance Imaging (fMRI) research have uncovered the existence of the central autonomic network (CAN), which comprises brain regions whose activity correlates with autonomic nervous system dynamics. By exploiting the spectral paradigm of heartbeat dynamics, cortical and sub-cortical areas functionally linked to vagal activity have been identified. However, due to methodological limitations, functional neural correlates of cardiac sympathetic dynamics remain uncharacterized. To this extent, we exploit the high spatiotemporal resolution of fMRI data from the Human Connectome Project to study the CAN activity by correlating a recently proposed instantaneous characterization of sympathetic activity (the sympathetic activity index - SAI) from heartbeat dynamics. SAI estimates are embedded into the probabilistic modeling of inhomogeneous point-processes, and are derived from a combination of disentangling coefficients linked to the orthonormal Laguerre functions. By analyzing resting state recordings from 34 young healthy people, we obtain positive correlations between instantaneous SAI estimates and a number of brain regions including frontal pole, insular cortex, frontal and temporal gyri, lateral occipital cortex, paracingulate and cingulate gyri, precuneus and temporal fusiform cortices, as well as thalamus, caudate nucleus, putamen, brain-stem, hippocampus, amygdala, and nucleus accumbens. Our findings significantly extend current knowledge on the CAN, opening new avenues in the characterization of healthy and pathological states in humans.

Reduced betweenness centrality of a sensory-motor vestibular network in subclinical agoraphobia.

Agoraphobic patients feel dizzy in crowded open spaces and respond to this symptom with excessive fear and avoidance. These clinical features show great similitude with the newly defined syndrome of persistent postural perceptual dizziness (PPPD). Patients with PPPD show decreased activity and connectivity in regions of the vestibular cortex. Due to the great overlap between these two conditions, we hypothesized that individuals with sub-clinical agoraphobia would show reduction in the connectivity features of these regions. We selected a group of healthy individuals from the Human Connectome Project that self-reported agoraphobia episodes, and compared it with a control group. We accurately matched the two groups for psychological measures and personality traits in order to study the neural correlates of vestibular symptoms independently of possible psychiatric vulnerabilities. We found that the agoraphobia group showed reduced betweenness centrality of a network encompassing key regions of the vestibular cortex. Dysfunctions of the vestibular cortex may explain the dizziness symptom for a disorder previously labelled as psychogenic.

Structural connectome of the human vestibular, pre-motor, and navigation network.

The aim of this study is to characterize modules and hubs within the multimodal vestibular system and, particularly, to test the centrality of posterior peri-sylvian regions. Structural connectivity matrices from 50 unrelated healthy right-handed subjects from the Human Connectome Project (HCP) database were analyzed using multishell diffusion-weighted data, probabilistic tractography (constrained spherical-deconvolution informed filtering of tractograms) in combination with subject-specific grey matter parcellations. Network nodes included parcellated regions within the vestibular, pre-motor and navigation system. Module calculation produced two and three modules in the right and left hemisphere, respectively. On the right, regions were grouped into a vestibular and pre-motor module, and into a visual-navigation module. On the left this last module was split into an inferior and superior component. In the thalamus, a region comprising the mediodorsal and anterior complex, and lateral and inferior pulvinar, was included in the ipsilateral navigation module, while the remaining thalamus was clustered with the ipsilateral vestibular pre-motor module. Hubs were located bilaterally in regions encompassing the inferior parietal cortex and the precuneus. This analysis revealed a dorso-lateral path within the multi-modal vestibular system related to vestibular / motor control, and a ventro-medial path related to spatial orientation / navigation. Posterior peri-sylvian regions may represent the main hubs of the whole modular network.

Functional connectome of the five-factor model of personality.

A key objective of the emerging field of personality neuroscience is to link the great variety of the enduring dispositions of human behaviour with reliable markers of brain function. This can be achieved by analyzing large sets of data with methods that model whole-brain connectivity patterns. To meet these expectations, we exploited a large repository of personality and neuroimaging measures made publicly available via the Human Connectome Project. Using connectomic analyses based on graph theory, we computed global and local indices of functional connectivity (e.g., nodal strength, efficiency, clustering, betweenness centrality) and related these metrics to the five-factor-model (FFM) personality traits (i.e., neuroticism, extraversion, openness, agreeableness, and conscientiousness). The maximal information coefficient was used to assess for linear and non-linear statistical dependencies across the graph 'nodes', which were defined as distinct brain circuits identified via independent component analysis. Multi-variate regression models and 'train/test' machine-learning approaches were also used to examine the associations between FFM traits and connectomic indices as well as to test for the generalizability of the main findings, whilst accounting for age and sex differences. Conscientiousness was the sole FFM trait linked to measures of higher functional connectivity in the fronto-parietal and default mode networks. This might provide a mechanistic explanation of the behavioural observation that conscientious people are reliable and efficient in goal-setting or planning. Our study provides new inputs to understanding the neurological basis of personality and contributes to the development of more realistic models of the brain dynamics that mediate personality differences.

Dynamical brain connectivity estimation using GARCH models: An application to personality neuroscience.

It has recently become evident that the functional connectome of the human brain is a dynamical entity whose time evolution carries important information underpinning physiological brain function as well as its disease-related aberrations. While simple sliding window approaches have had some success in estimating dynamical brain connectivity in a functional MRI (fMRI) context, these methods suffer from limitations related to the arbitrary choice of window length and limited time resolution. Recently, Generalized autoregressive conditional heteroscedastic (GARCH) models have been employed to generate dynamical covariance models which can be applied to fMRI. Here, we employ a GARCH-based method (dynamic conditional correlation - DCC) to estimate dynamical brain connectivity in the Human Connectome Project (HCP) dataset and study how the dynamic functional connectivity behaviors related to personality as described by the five-factor model. Openness, a trait related to curiosity and creativity, is the only trait associated with significant differences in the amount of time-variability (but not in absolute median connectivity) of several inter-network functional connections in the human brain. The DCC method offers a novel window to extract dynamical information which can aid in elucidating the neurophysiological underpinning of phenomena to which conventional static brain connectivity estimates are insensitive.

Dynamic inter-network connectivity in the human brain.

Recently, the field of functional brain connectivity has shifted its attention on studying how functional connectivity (FC) between remote regions changes over time. It is becoming increasingly evident that the human "connectome" is a dynamical entity whose variations are effected over very short timescales and reflect crucial mechanisms which underline the physiological functioning of the brain. In this study, we employ ad-hoc statistical and surrogate data generation methods to quantify whether and which brain networks displayed dynamic behaviors in a very large sample of healthy subjects provided by the Human Connectome Project (HCP). Our findings provided evidences that there are specific pairs of networks and specific networks within the healthy brain that are more likely to display dynamic behaviors. This new set of findings supports the notion that studying the time-variant connectivity in the brain could reveal useful and important properties about brain functioning in health and disease.

Resting-state brain correlates of cardiovascular complexity.

While estimates of complex heartbeat dynamics have provided effective prognostic and diagnostic markers for a wide range of pathologies, brain correlates of complex cardiac measures in general and of complex sympatho-vagal dynamics in particular are still unknown. In this study we combine resting state functional Magnetic Resonance Imaging (fMRI) and physiological signal acquisition from 34 healthy subjects selected from the Human Connectome Project (HCP) repository with inhomogeneous point-process approximate and sample heartbeat entropy measures (ipApEn and ipSampEn) to investigate brain areas involved in complex cardiovascular control. Our results show that activity in the Temporal Gyrus, Frontal Orbital Cortex, Temporal Fusiform and Opercular cortices, Planum Temporale, and Paracingulate cortex are negatively correlated with ipApEn dynamics. Activity in the same cortical areas as well as in the Temporal Fusiform cortex are negatively correlated with ipSampEn dynamics. No significant positive correlations were found. These pioneering results suggest that cardiovascular complexity at rest is linked to a few specific cortical brain structures, including crucial areas connected with parasympathetic outflow. This corroborates the hypothesis of a multidimensional central network which controls nonlinear cardiac dynamics under a predominantly vagally-driven tone.

Resting-state brain correlates of instantaneous autonomic outflow.

A prominent pathway of brain-heart interaction is represented by autonomic nervous system (ANS) heartbeat modulation. While within-brain resting state networks have been the object of intense functional Magnetic Resonance Imaging (fMRI) research, technological and methodological limitations have hampered research on the central correlates of cardiovascular control dynamics. Here we combine the high temporal and spatial resolution as well as data volume afforded by the Human Connectome Project with a probabilistic model of heartbeat dynamics to characterize central correlates of sympathetic and parasympathetic ANS activity at rest. We demonstrate an involvement of a number of brain regions such as the Insular cortex, Frontal Gyrus, Lateral Occipital Cortex, Paracingulate and Cingulate Gyrus and Precuneous Cortex, as well as subcortical structures (Thalamus, Putamen, Pallidum, Brain-Stem, Hippocampus, Amygdala, and Right Caudate) in the modulation of ANS-mediated cardiovascular control, possibly indicating a broader definition of the central autonomic network (CAN). Our findings provide a basis for an informed neurobiological interpretation of the numerous studies which employ HRV-related measures as standalone biomarkers in health and disease.

Estimating directed brain-brain and brain-heart connectivity through globally conditioned Granger causality approaches.

While a large body of research has focused on the study of within-brain physiological networks (i.e. brain connectivity) as well as their disease-related aberration, few investigators have focused on estimating the directionality of these brain-brain interaction which, given the complexity of brain networks, should be properly conditioned in order to avoid the high number of false positives commonly encountered when using bivariate approaches to brain connectivity estimation. Additionally, the constituents of a number of brain subnetworks, and in particular of the central autonomic network (CAN), are still not completely determined. In this study we present and validate a global conditioning approach to reconstructing directed networks using complex synthetic networks of nonlinear oscillators. We then employ our framework, along with a probabilistic model for heartbeat generation, to characterize the directed functional connectome of the human brain and to establish which parts of this connectome effect the directed central modulation of peripheral autonomic cardiovascular control. We demonstrate the effectiveness of our conditioning approach and unveil a top-down directed influence of the default mode network on the salience network, which in turn is seen to be the strongest modulator of directed autonomic cardiovascular control.

Simultaneous estimation of the in-mean and in-variance causal connectomes of the human brain.

In recent years, the study of the human connectome (i.e. of statistical relationships between non spatially contiguous neurophysiological events in the human brain) has been enormously fuelled by technological advances in high-field functional magnetic resonance imaging (fMRI) as well as by coordinated world wide data-collection efforts like the Human Connectome Project (HCP). In this context, Granger Causality (GC) approaches have recently been employed to incorporate information about the directionality of the influence exerted by a brain region on another. However, while fluctuations in the Blood Oxygenation Level Dependent (BOLD) signal at rest also contain important information about the physiological processes that underlie neurovascular coupling and associations between disjoint brain regions, so far all connectivity estimation frameworks have focused on central tendencies, hence completely disregarding so-called in-variance causality (i.e. the directed influence of the volatility of one signal on the volatility of another). In this paper, we develop a framework for simultaneous estimation of both in-mean and in-variance causality in complex networks. We validate our approach using synthetic data from complex ensembles of coupled nonlinear oscillators, and successively employ HCP data to provide the very first estimate of the in-variance connectome of the human brain.

Differences in Gaussian diffusion tensor imaging and non-Gaussian diffusion kurtosis imaging model-based estimates of diffusion tensor invariants in the human brain.

PURPOSE: An increasing number of studies have aimed to compare diffusion tensor imaging (DTI)-related parameters [e.g., mean diffusivity (MD), fractional anisotropy (FA), radial diffusivity (RD), and axial diffusivity (AD)] to complementary new indexes [e.g., mean kurtosis (MK)/radial kurtosis (RK)/axial kurtosis (AK)] derived through diffusion kurtosis imaging (DKI) in terms of their discriminative potential about tissue disease-related microstructural alterations. Given that the DTI and DKI models provide conceptually and quantitatively different estimates of the diffusion tensor, which can also depend on fitting routine, the aim of this study was to investigate model- and algorithm-dependent differences in MD/FA/RD/AD and anisotropy mode (MO) estimates in diffusion-weighted imaging of human brain white matter. METHODS: The authors employed (a) data collected from 33 healthy subjects (20-59 yr, F: 15, M: 18) within the Human Connectome Project (HCP) on a customized 3 T scanner, and (b) data from 34 healthy subjects (26-61 yr, F: 5, M: 29) acquired on a clinical 3 T scanner. The DTI model was fitted to b-value =0 and b-value =1000 s/mm(2) data while the DKI model was fitted to data comprising b-value =0, 1000 and 3000/2500 s/mm(2) [for dataset (a)/(b), respectively] through nonlinear and weighted linear least squares algorithms. In addition to MK/RK/AK maps, MD/FA/MO/RD/AD maps were estimated from both models and both algorithms. Using tract-based spatial statistics, the authors tested the null hypothesis of zero difference between the two MD/FA/MO/RD/AD estimates in brain white matter for both datasets and both algorithms. RESULTS: DKI-derived MD/FA/RD/AD and MO estimates were significantly higher and lower, respectively, than corresponding DTI-derived estimates. All voxelwise differences extended over most of the white matter skeleton. Fractional differences between the two estimates [(DKI - DTI)/DTI] of most invariants were seen to vary with the invariant value itself as well as with MK/RK/AK values, indicating substantial anatomical variability of these discrepancies. In the HCP dataset, the median voxelwise percentage differences across the whole white matter skeleton were (nonlinear least squares algorithm) 14.5% (8.2%-23.1%) for MD, 4.3% (1.4%-17.3%) for FA, -5.2% (-48.7% to -0.8%) for MO, 12.5% (6.4%-21.2%) for RD, and 16.1% (9.9%-25.6%) for AD (all ranges computed as 0.01 and 0.99 quantiles). All differences/trends were consistent between the discovery (HCP) and replication (local) datasets and between estimation algorithms. However, the relationships between such trends, estimated diffusion tensor invariants, and kurtosis estimates were impacted by the choice of fitting routine. CONCLUSIONS: Model-dependent differences in the estimation of conventional indexes of MD/FA/MO/RD/AD can be well beyond commonly seen disease-related alterations. While estimating diffusion tensor-derived indexes using the DKI model may be advantageous in terms of mitigating b-value dependence of diffusivity estimates, such estimates should not be referred to as conventional DTI-derived indexes in order to avoid confusion in interpretation as well as multicenter comparisons. In order to assess the potential and advantages of DKI with respect to DTI as well as to standardize diffusion-weighted imaging methods between centers, both conventional DTI-derived indexes and diffusion tensor invariants derived by fitting the non-Gaussian DKI model should be separately estimated and analyzed using the same combination of fitting routines.

Physics, Techniques and Review of Neuroradiological Applications of Diffusion Kurtosis Imaging (DKI).

In recent years many papers about diagnostic applications of diffusion tensor imaging (DTI) have been published. This is because DTI allows to evaluate in vivo and in a non-invasive way the process of diffusion of water molecules in biological tissues. However, the simplified description of the diffusion process assumed in DTI does not permit to completely map the complex underlying cellular components and structures, which hinder and restrict the diffusion of water molecules. These limitations can be partially overcome by means of diffusion kurtosis imaging (DKI). The aim of this paper is the description of the theory of DKI, a new topic of growing interest in radiology. DKI is a higher order diffusion model that is a straightforward extension of the DTI model. Here, we analyze the physics underlying this method, we report our MRI acquisition protocol with the preprocessing pipeline used and the DKI parametric maps obtained on a 1.5 T scanner, and we review the most relevant clinical applications of this technique in various neurological diseases.

Forecasting nanoparticle toxicity using nonlinear predictive regressor learning systems.

Nanoparticle (NP) toxicity is determined by a vast number of topological, sterical, physico-chemical as well as biological properties, rendering a priori evaluation of the effect of NP on biological tissue as arduous as it is necessary and urgent. We aimed at mining the HORIZON 2020 MODENA COST NP cytotoxicity database through nonlinear predictive regressor learning systems in order to assess the power of available NP descriptors and assay characteristics in predicting NP toxicity. Specifically, we assessed the results of cytotoxicity assays performed on 57 NP and trained two different nonlinear regressors (Support Vector Regressors [SVR] with polynomical kernels and Radial Basis Function [RBF] regressors) within a nested-cross validation scheme for parameter optimization to predict toxicity as quantified by EC25, EC50 and slope while using the regressional ReliefF algorithm (RReliefF) for feature selection. Available NP attributes were material, coating, cell type, dispersion protocol, shape, 1st and 2nd dimension, aspect ratio, surface area, zeta potential and size in situ. In most regressor learning systems, after feature selection with the RReliefF algorithm, the correlation between real and estimated toxicity endpoint values increased monotonically with the number of included features, reaching values above 0.90. The best performance was obtained with RBF regressors, and the most informative features in predicting toxicity endpoints were related to nanoparticle structure. These trends did not change significantly between toxicity endpoints. In conclusion, EC25, EC50 and slope can be predicted with high correlation using purely data-driven, machine learning methods in Adenosine triphosphate (ATP)-based NP cytotoxicity assays.

Predicting seizures in untreated temporal lobe epilepsy using point-process nonlinear models of heartbeat dynamics.

Symptoms of temporal lobe epilepsy (TLE) are frequently associated with autonomic dysregulation, whose underlying biological processes are thought to strongly contribute to sudden unexpected death in epilepsy (SUDEP). While abnormal cardiovascular patterns commonly occur during ictal events, putative patterns of autonomic cardiac effects during pre-ictal (PRE) periods (i.e. periods preceding seizures) are still unknown. In this study, we investigated TLE-related heart rate variability (HRV) through instantaneous, nonlinear estimates of cardiovascular oscillations during inter-ictal (INT) and PRE periods. ECG recordings from 12 patients with TLE were processed to extract standard HRV indices, as well as indices of instantaneous HRV complexity (dominant Lyapunov exponent and entropy) and higher-order statistics (bispectra) obtained through definition of inhomogeneous point-process nonlinear models, employing Volterra-Laguerre expansions of linear, quadratic, and cubic kernels. Experimental results demonstrate that the best INT vs. PRE classification performance (balanced accuracy: 73.91%) was achieved only when retaining the time-varying, nonlinear, and non-stationary structure of heartbeat dynamical features. The proposed approach opens novel important avenues in predicting ictal events using information gathered from cardiovascular signals exclusively.