Overcoming the clinical-MR imaging paradox of multiple sclerosis: MR imaging data assessed with a random forest approach

Meta-analysis of cellular toxicity for graphene via data-mining the literature and machine learning

Since graphene is currently incorporated into various consumer products and used in a variety of applications, determining the relationships between the physicochemical properties of graphene and its toxicity is critical for conducting environmental and health risk analyses. Data from the literature suggest that exposure to graphene may result in cytotoxicity. However, existing graphene toxicity data are complex and heterogeneous, making it difficult to conduct risk assessments. Here, we conducted a meta-analysis of published data on the cytotoxicity of graphene based on 792 publications, including 986 cell viability data points, 762 half maximal inhibitory concentration (IC₅₀) data points, and 100 lactate dehydrogenase (LDH) release data points. Models to predict graphene cytotoxicity were then developed based on cell viability, IC₅₀, and LDH release as toxicity endpoints using random forests learning algorithms. The most influential attributes influencing graphene cytotoxicity were revealed to be exposure dose and detection method for cell viability, diameter and surface modification for IC₅₀, and detection method and organ source for LDH release. The meta-analysis produced three sets of key attributes for the three abovementioned toxicity endpoints that can be used in future studies of graphene toxicity. The findings indicate that rigorous data mining protocols can be combined with suitable machine learning tools to develop models with good predictive power and accuracy. The results also provide guidance for the design of safe graphene materials.

Network Damage Predicts Clinical Worsening in Multiple Sclerosis: A 6.4-Year Study

OBJECTIVE

In multiple sclerosis (MS), clinical impairment is likely due to both structural damage and abnormal brain function. We assessed the added value of integrating structural and functional network MRI measures to predict 6.4-year MS clinical disability deterioration.

METHODS

Baseline 3D T1-weighted and resting-state functional MRI scans were obtained from 233 patients with MS and 77 healthy controls. Patients underwent a neurologic evaluation at baseline and at 6.4-year median follow-up (interquartile range = 5.06-7.51 years). At follow-up, patients were classified as clinically stable/worsened according to disability changes. In relapsing-remitting (RR) MS, secondary progressive (SP) MS conversion was evaluated. Global brain volumetry was obtained. Furthermore, independent component analysis identified the main functional connectivity (FC) and gray matter (GM) network patterns.

RESULTS

At follow-up, 105/233 (45%) patients were clinically worsened; 26/157 (16%) patients with RRMS evolved to SPMS. The treatment-adjusted random forest model identified normalized GM and brain volumes, decreased FC between default-mode networks, increased FC of the left precentral gyrus in the sensorimotor network (SMN), and GM atrophy in the fronto-parietal network (false discovery rate [FDR]-corrected p = range 0.01-0.09) as predictors of clinical worsening (out-of-bag [OOB] accuracy = 0.74). An expected contribution of baseline disability was also present (FDR-p = 0.01). Baseline disability, normalized GM volume, and GM atrophy in the SMN (FDR-p = range 0.01-0.09) were independently associated with SPMS conversion (OOB accuracy = 0.84). At receiver operating characteristic analysis, including network MRI variables improved disability worsening (p = 0.05) and SPMS conversion (p = 0.02) prediction.

CONCLUSIONS

Integration of MRI network measures helped determining the relative contributions of global/local GM damage and functional reorganization to clinical deterioration in MS.

Relationships between circulating branched chain amino acid concentrations and risk of adverse cardiovascular events in patients with STEMI treated with PCI

The incidence of in-hospital cardiovascular adverse events (AEs) in patients with ST-segment elevation myocardial infarction (STEMI) following primary percutaneous coronary intervention (PCI) is relatively high. Identification of metabolic markers could improve our understanding of the underlying pathological changes in these patients. We aimed to identify associations between concentrations of plasma metabolites on admission and development of in-hospital AEs in post-PCI patients with STEMI. We used targeted mass spectrometry to measure plasma concentrations of 26 amino acid metabolites on admission in 96 patients with STEMI who subsequently developed post-PCI AEs and in 96 age- and sex-matched patients without post-PCI cardiovascular AEs. Principal component analysis (PCA) revealed that PCA-derived factors, including branched chain amino acids (BCAAs), were associated with increased risks of all three pre-specified outcomes: cardiovascular mortality/acute heart failure (AHF), cardiovascular mortality, and AHF. Addition of BCAA to the Global Registry of Acute Coronary Events risk score increased the concordance C statistic from 0.702 to 0.814 (p < 0.001), and had a net reclassification index of 0.729 (95% confidence interval, 0.466–0.992, p < 0.001). These findings demonstrate that high circulating BCAA concentrations on admission are associated with subsequent in-hospital AEs after revascularization in patients with STEMI.

Defining an Analytic Framework to Evaluate Quantitative MRI Markers of Traumatic Axonal Injury: Preliminary Results in a Mouse Closed Head Injury Model

Diffuse axonal injury (DAI) is a hallmark of traumatic brain injury (TBI) pathology. Recently, the Closed Head Injury Model of Engineered Rotational Acceleration (CHIMERA) was developed to generate an experimental model of DAI in a mouse. The characterization of DAI using diffusion tensor magnetic resonance imaging (MRI; diffusion tensor imaging, DTI) may provide a useful set of outcome measures for preclinical and clinical studies. The objective of this study was to identify the complex neurobiological underpinnings of DTI features following DAI using a comprehensive and quantitative evaluation of DTI and histopathology in the CHIMERA mouse model. A consistent neuroanatomical pattern of pathology in specific white matter tracts was identified across ex vivo DTI maps and photomicrographs of histology. These observations were confirmed by voxelwise and regional analysis of DTI maps, demonstrating reduced fractional anisotropy (FA) in distinct regions such as the optic tract. Similar regions were identified by quantitative histology and exhibited axonal damage as well as robust gliosis. Additional analysis using a machine-learning algorithm was performed to identify regions and metrics important for injury classification in a manner free from potential user bias. This analysis found that diffusion metrics were able to identify injured brains almost with the same degree of accuracy as the histology metrics. Good agreement between regions detected as abnormal by histology and MRI was also found. The findings of this work elucidate the complexity of cellular changes that give rise to imaging abnormalities and provide a comprehensive and quantitative evaluation of the relative importance of DTI and histological measures to detect brain injury.

Relevance of asymptomatic spinal MRI lesions in patients with multiple sclerosis

BACKGROUND

The impact of new asymptomatic spinal cord lesions (a-SL) in multiple sclerosis (MS) course is poorly characterized.

OBJECTIVE

The objective of this research paper is to assess the prognostic value of a-SL in predicting MS course.

METHODS

Relapsing-remitting MS patients who received serial MRI (brain and spinal) at baseline (t1) and within 12 to 36 months (t2) during clinical stability, and had a follow-up (t2-t3) ⩾24 months were included. Relapses and disability progression were evaluated between t2 and t3.

RESULTS

Of 413 consecutive screened MS patients, 103 patients (65 females, median age 43 years) were included. After a median t1-t2 interval of 17 (IQR 13-26) months, 25.2% and 43.7% patients had ⩾1 new a-SL (a-SL+) and asymptomatic brain lesions (a-BL+), respectively. Relapse risk between t2 and t3 (median interval: 42 (IQR 32-57.5) months) was significantly increased in a-SL+ and/or a-BL+ vs a-BL- and a-SL- (HR = 2.31, 95% CI = 1.13-4.72, p = 0.02). No differences in the risk of disability progression were found in a-SL+ and/or a-BL+ vs a-SL- and a-BL-.

CONCLUSION

a-SL occur in one-quarter of clinically stable RRMS, and combined with a-BL contribute significantly in predicting future disease course.

Identifying Chinese Microblog Users With High Suicide Probability Using Internet-Based Profile and Linguistic Features: Classification Model

BACKGROUND

Traditional offline assessment of suicide probability is time consuming and difficult in convincing at-risk individuals to participate. Identifying individuals with high suicide probability through online social media has an advantage in its efficiency and potential to reach out to hidden individuals, yet little research has been focused on this specific field.

OBJECTIVE

The objective of this study was to apply two classification models, Simple Logistic Regression (SLR) and Random Forest (RF), to examine the feasibility and effectiveness of identifying high suicide possibility microblog users in China through profile and linguistic features extracted from Internet-based data.

METHODS

There were nine hundred and nine Chinese microblog users that completed an Internet survey, and those scoring one SD above the mean of the total Suicide Probability Scale (SPS) score, as well as one SD above the mean in each of the four subscale scores in the participant sample were labeled as high-risk individuals, respectively. Profile and linguistic features were fed into two machine learning algorithms (SLR and RF) to train the model that aims to identify high-risk individuals in general suicide probability and in its four dimensions. Models were trained and then tested by 5-fold cross validation; in which both training set and test set were generated under the stratified random sampling rule from the whole sample. There were three classic performance metrics (Precision, Recall, F1 measure) and a specifically defined metric "Screening Efficiency" that were adopted to evaluate model effectiveness.

RESULTS

Classification performance was generally matched between SLR and RF. Given the best performance of the classification models, we were able to retrieve over 70% of the labeled high-risk individuals in overall suicide probability as well as in the four dimensions. Screening Efficiency of most models varied from 1/4 to 1/2. Precision of the models was generally below 30%.

CONCLUSIONS

Individuals in China with high suicide probability are recognizable by profile and text-based information from microblogs. Although there is still much space to improve the performance of classification models in the future, this study may shed light on preliminary screening of risky individuals via machine learning algorithms, which can work side-by-side with expert scrutiny to increase efficiency in large-scale-surveillance of suicide probability from online social media.

ANALYSIS OF MULTIPLE SCLEROSIS LESIONS VIA SPATIALLY VARYING COEFFICIENTS

Magnetic resonance imaging (MRI) plays a vital role in the scientific investigation and clinical management of multiple sclerosis. Analyses of binary multiple sclerosis lesion maps are typically "mass univariate" and conducted with standard linear models that are ill suited to the binary nature of the data and ignore the spatial dependence between nearby voxels (volume elements). Smoothing the lesion maps does not entirely eliminate the non-Gaussian nature of the data and requires an arbitrary choice of the smoothing parameter. Here we present a Bayesian spatial model to accurately model binary lesion maps and to determine if there is spatial dependence between lesion location and subject specific covariates such as MS subtype, age, gender, disease duration and disease severity measures. We apply our model to binary lesion maps derived from T2-weighted MRI images from 250 multiple sclerosis patients classified into five clinical subtypes, and demonstrate unique modeling and predictive capabilities over existing methods.

Metabolomics signature improves the prediction of cardiovascular events in elderly subjects

AIMS

Age is one of the most important determinants of cardiovascular health, therefore the management of cardiovascular diseases (CVD) in elderly people entails great challenge. A possible explanation of vascular senescence process is the mitochondrial damage and dysfunction. We hypothesized that metabolomic profiling would identify biomarkers predicting major cardiovascular events (MACEs) in elderly people, improving the clinical standard cardiovascular risk factors.

METHODS AND RESULTS

Targeted-mass-spectrometry-based profiling of 49 metabolites was performed in a group of very old participants (n = 67, mean age = 85 ± 3 years) with a high rate of previous CVD (68%). Principal Component Analysis, Random Survival Forest analysis and Cox proportional hazards regression modeling were used to evaluate the relation between the metabolite factors and recurring MACEs. We tested discrimination ability and reclassification of clinical and metabolomic models. At follow-up (median = 3.5 years), 17 MACEs occurred (5 cardiovascular deaths, 1 nonfatal myocardial infarction, 7 nonfatal strokes and 4 peripheral artery surgeries) (incidence = 7.3% person-years). Metabolite factor 1, composed by medium- and long-chain acylcarnitines, and factor 7 (alanine) were independently associated with MACEs, after adjustment for clinical CV covariates [HR = 1.77 (95%CI = 1.11-2.81, p = 0.016) and HR = 2.18 (95%CI = 1.17-4.07, p = 0.014), respectively]. However, only factor 1 significantly increases the prediction accuracy of the Framingham Recurring-Coronary-Heart-Disease-Score, with a significant improvement in discrimination (integrated discrimination improvement = 7%, p = 0.01) and correctly reclassifying 41% of events and 37% of non-events resulting in a cNRI = 0.79 (p = 0.005).

CONCLUSIONS

Aging mitochondrial dysfunction evaluated by metabolomic profiling is associated with MACEs, independently of standard predictors.

Cognitive impairment in multiple sclerosis: a forgotten disability remembered

Physicians first noted the presence of cognitive impairment in patients with multiple sclerosis (MS) more than 160 years ago, yet it took clinicians until 2001 to codify a standard test to measure cognitive function. We now know that cognitive impairment occurs in up to 65 percent of people with MS and usually lessens their ability to remember previously learned information. So far, trials of drugs formulated to treat cognitive impairment have failed, but the authors remain optimistic that new approaches to diagnosis and drug development could lead to effective therapies in the future.