Evaluation of machine learning algorithms and structural features for optimal MRI-based diagnostic prediction in psychosis.

A relatively large number of studies have investigated the power of structural magnetic resonance imaging (sMRI) data to discriminate patients with schizophrenia from healthy controls. However, very few of them have also included patients with bipolar disorder, allowing the clinically relevant discrimination between both psychotic diagnostics. To assess the efficacy of sMRI data for diagnostic prediction in psychosis we objectively evaluated the discriminative power of a wide range of commonly used machine learning algorithms (ridge, lasso, elastic net and L0 norm regularized logistic regressions, a support vector classifier, regularized discriminant analysis, random forests and a Gaussian process classifier) on main sMRI features including grey and white matter voxel-based morphometry (VBM), vertex-based cortical thickness and volume, region of interest volumetric measures and wavelet-based morphometry (WBM) maps. All possible combinations of algorithms and data features were considered in pairwise classifications of matched samples of healthy controls (N = 127), patients with schizophrenia (N = 128) and patients with bipolar disorder (N = 128). Results show that the selection of feature type is important, with grey matter VBM (without data reduction) delivering the best diagnostic prediction rates (averaging over classifiers: schizophrenia vs. healthy 75%, bipolar disorder vs. healthy 63% and schizophrenia vs. bipolar disorder 62%) whereas algorithms usually yielded very similar results. Indeed, those grey matter VBM accuracy rates were not even improved by combining all feature types in a single prediction model. Further multi-class classifications considering the three groups simultaneously made evident a lack of predictive power for the bipolar group, probably due to its intermediate anatomical features, located between those observed in healthy controls and those found in patients with schizophrenia. Finally, we provide MRIPredict (https://www.nitrc.org/projects/mripredict/), a free tool for SPM, FSL and R, to easily carry out voxelwise predictions based on VBM images.

[Systematic review of the Genomewide Association Studies (GWAS) in schizophrenia]. / Revisión sistemática de los Genomewide Association Studies (GWAS) en esquizofrenia.

INTRODUCTION: Heritability in schizophrenia can reach up to 80% and the risk in families is 5-10 times higher than in the general population. The large contribution of genetics in this disorder has led to a growing interest in its study. OBJECTIVES: To review the findings of genetic studies known as Genomewide Association Studies (GWAS) on schizophrenia. METHOD: Systematic search using Pubmed with the key words GWAS and (psychosis) or (schizophrenia). The following web pages have been reviewed: http://www.szgene.org/largescale.asp and www.genome.gov/gwastudies/. RESULTS: The GWAS have focused on causal biological aspects, such as the histocompatibility complex, glutamate metabolism, apoptosis and inflammatory processes, and the immune system (TNF-ß, TNFR1). Also focused in the search were the genes that modulate the appearance of secondary metabolic and cardiac effects and secondary effects in subjects with schizophrenia and on anti-psychotic treatment. In neurorecognition, over-expression of the MET proto-oncogene (MET) has been associated with a low susceptibility for schizophrenia and a better cognitive performance, as well as a lower susceptibility for the incidence of cancer. Mention is also made of the different genes that mediate in cognitive functioning depending on the anti-psychotic treatment received. CONCLUSIONS: The main interests of the GWAS during the last few years have been the neurobiological pathways involved in schizophrenia. The discoveries arising from these studies have been limited. This has led to an innovative approach on the aetiological study of the disorder by studying gene-environment interactions.

Effects of vascular flow and PEEP in a multiple hit model of lung injury in isolated perfused rabbit lungs.

BACKGROUND: High vascular flow aggravates lung damage in animal models of ventilator-induced lung injury. Positive end-expiratory pressure (PEEP) can attenuate ventilator-induced lung injury, but its continued effectiveness in the setting of antecedent lung injury is unclear. The objective of the present study was to evaluate whether the application of PEEP diminishes lung injury induced by concurrent high vascular flow and high alveolar pressures in normal lungs and in a preinjury lung model. METHODS: Two series of experiments were performed. Fifteen sets of isolated rabbit lungs were randomized into three groups (n = 5): low vascular flow/low PEEP; high vascular flow/low PEEP, and high vascular flow/high PEEP. Subsequently, the same protocol was applied in an additional 15 sets of isolated rabbit lungs in which oleic acid was added to the vascular perfusate to produce mild to moderate lung injury. All lungs were ventilated with peak airway pressure of 30 cm H2O for 30 minutes. Outcome measures included frequency of gross structural failure, pulmonary hemorrhage, edema formation, changes in static compliance, pulmonary vascular resistance, and pulmonary ultrafiltration coefficient. RESULTS: In the context of high vascular flow, application of a moderate level of PEEP reduced pulmonary rupture, edema formation, and lung hemorrhage. The protective effects of PEEP were not observed in lungs concurrently injured with oleic acid. CONCLUSIONS: Under these experimental conditions, PEEP attenuates lung injury in the setting of high vascular flow. The protective effect of PEEP is lost in a two-hit model of lung injury.

Contributions of vascular flow and pulmonary capillary pressure to ventilator-induced lung injury.

OBJECTIVE: To evaluate the influence of vascular flow on ventilator-induced lung injury independent of vascular pressures. DESIGN: Laboratory study. SETTING: Hospital laboratory. SUBJECTS: Thirty-two New Zealand White rabbits. INTERVENTIONS: Thirty-two isolated perfused rabbit lungs were allocated into four groups: low flow/low pulmonary capillary pressure; high flow/high pulmonary capillary pressure; low flow/high pulmonary capillary pressure, and high flow/low pulmonary capillary pressure. All lungs were ventilated with peak airway pressure 30 cm H2O and positive end-expiratory pressure 5 cm H2O for 30 mins. MEASUREMENTS AND MAIN RESULTS: Outcome measures included frequency of gross structural failure (pulmonary rupture), pulmonary hemorrhage, edema formation, changes in lung compliance, pulmonary vascular resistance, and pulmonary ultrafiltration coefficient. Lungs exposed to high pulmonary vascular flow ruptured more frequently, displayed more hemorrhage, developed more edema, suffered larger decreases in compliance, and had larger increases in vascular resistance than lungs exposed to low vascular flows (p < .05 for each pairwise comparison between groups). CONCLUSIONS: These findings suggest that high pulmonary vascular flows might exacerbate ventilator-induced lung injury independent of their effects on pulmonary vascular pressures.