Machine learning allows expert level classification of intraoperative motor evoked potentials during neurosurgical procedures.

OBJECTIVE: To develop and evaluate machine learning (ML) approaches for muscle identification using intraoperative motor evoked potentials (MEPs), and to compare their performance to human experts. BACKGROUND: There is an unseized opportunity to apply ML analytic techniques to the world of intraoperative neuromonitoring (IOM). MEPs are the ideal candidates given the importance of their correct interpretation during a surgical operation to the brain or the spine. In this work, we develop and test a set of different ML models for muscle identification using intraoperative MEPs and compare their performance to human experts. In addition, we provide a review of the available literature on current ML applications to IOM data in neurosurgery. METHODS: We trained and tested five different ML classifiers on a MEP database developed from six different muscles in patients who underwent brain or spinal cord surgery. MEPs were obtained by both transcranial (TES) and direct cortical stimulation (DCS) protocols. The models were evaluated within a single patient and on previously unseen patients, considering signals from TES and DCS both independently and mixed. Ten expert neurophysiologists classified a set of 50 randomly selected MEPs, and their performance was compared to the best performing model. RESULTS: A total of 25.423 MEPs were included in the study. Random Forest proved to be the best performing model with 99 % accuracy in the single patient dataset task and a 78 %-94 % accuracy range on previously unseen patients. The model performance was maximized by representing MEPs as a set of features typically employed in signal processing compared to traditional neurophysiological parameters. The classification ability of the Random Forest model between six different muscles and across different MEP acquisition modalities (79 %) significantly exceeded that of human experts (mean 48 %). CONCLUSIONS: Carefully selected ML models proved to have reliable capacity of extracting meaningful information to classify intraoperative MEPs using a limited number of features, proving robustness across patients and signal acquisition modalities, outperforming human experts, and with the potential to act as decision support systems to the IOM team. Such encouraging results lay the path to further explore the underlying nature of clinically important signals, with the aim to continue to produce useful applications to make surgeries safer and more efficient.

Metabolic-associated fatty liver disease and liver fibrosis scores as COVID-19 outcome predictors: a machine-learning application.

Patients with COVID-19 and metabolic-dysfunction associated fatty liver disease (MAFLD) appear to be at higher risk for severe manifestations, especially in the youngest decades. Our aim was to examine whether patients with MAFLD and/or with increased liver fibrosis scores (FIB-4) are at risk for severe COVID-19 illness, using a machine learning (ML) model. Six hundred and seventy two patients were enrolled for SARS-CoV-2 pneumonia between February 2020 and May 2021. Steatosis was detected by ultrasound or computed tomography (CT). ML model valuated the risks of both in-hospital death and prolonged hospitalizations (> 28 days), considering MAFLD, blood hepatic profile (HP), and FIB-4 score. 49.6% had MAFLD. The accuracy in predicting in-hospital death was 0.709 for the HP alone and 0.721 for HP + FIB-4; in the 55-75 age subgroup, 0.842/0.855; in the MAFLD subgroup, 0.739/ 0.772; in the MAFLD 55-75 years, 0.825/0.833. Similar results were obtained when considering the accuracy in predicting prolonged hospitalization. In our cohort of COVID-19 patients, the presence of a worse HP and a higher FIB-4 correlated with a higher risk of death and prolonged hospitalization, regardless of the presence of MAFLD. These findings could improve the clinical risk stratification of patients diagnosed with SARS-CoV-2 pneumonia.

Classification of multiple sclerosis patients based on structural disconnection: A robust feature selection approach.

BACKGROUND AND PURPOSE: Although structural disconnection represents the hallmark of multiple sclerosis (MS) pathophysiology, classification attempts based on structural connectivity have achieved low accuracy levels. Here, we set out to fill this gap, exploring the performance of supervised classifiers on features derived from microstructure informed tractography and selected applying a novel robust approach. METHODS: Using microstructure informed tractography with diffusion MRI data, we created quantitative connectomes of 55 MS patients and 24 healthy controls. We then used a robust approach-based on two classical methods of feature selection- to select relevant features from three network representations (whole connectivity matrices, node strength, and local efficiency). Classification accuracy of the selected features was tested with five different classifiers, while their meaningfulness was tested via correlation with clinical scales. As a comparison, the same classifiers were run on features selected with the standard procedure in network analysis (thresholding). RESULTS: Our procedure identified 11 features for the whole net, five for local efficiency, and seven for node strength. For all classifiers, the accuracy was in the range 64.5%-91.1%, with features extracted from the whole net reaching the maximum, and overcoming results obtained with the standard procedure in all cases. Correlations with clinical scales were identified across functional domains, from motor and cognitive abilities to fatigue and depression. CONCLUSION: Applying a robust feature selection procedure to quantitative structural connectomes, we were able to classify MS patients with excellent accuracy, while providing information on the white matter connections and gray matter regions more affected by MS pathology.

The Topography of Visually Guided Grasping in the Premotor Cortex: A Dense-Transcranial Magnetic Stimulation (TMS) Mapping Study.

Visuomotor transformations at the cortical level occur along a network where posterior parietal regions are connected to homologous premotor regions. Grasping-related activity is represented in a diffuse, ventral and dorsal system in the posterior parietal regions, but no systematic causal description of a premotor counterpart of a similar diffuse grasping representation is available. To fill this gap, we measured the kinematics of right finger movements in 17 male and female human participants during grasping of three objects of different sizes. Single-pulse transcranial magnetic stimulation was applied 100 ms after visual presentation of the object over a regular grid of 8 spots covering the left premotor cortex (PMC) and 2 Sham stimulations. Maximum finger aperture during reach was used as the feature to classify object size in different types of classifiers. Classification accuracy was taken as a measure of the efficiency of visuomotor transformations for grasping. Results showed that transcranial magnetic stimulation reduced classification accuracy compared with Sham stimulation when it was applied to 2 spots in the ventral PMC and 1 spot in the medial PMC, corresponding approximately to the ventral PMC and the dorsal portion of the supplementary motor area. Our results indicate a multifocal representation of object geometry for grasping in the PMC that matches the known multifocal parietal maps of grasping representations. Additionally, we confirm that, by applying a uniform spatial sampling procedure, transcranial magnetic stimulation can produce cortical functional maps independent of a priori spatial assumptions.SIGNIFICANCE STATEMENT Visually guided actions activate a large frontoparietal network. Here, we used a dense grid of transcranial magnetic stimulation spots covering the whole premotor cortex (PMC), to identify with accurate spatial mapping the functional specialization of the human PMC during grasping movement. Results corroborate previous findings about the role of the ventral PMC in preshaping the fingers according to the size of the target. Crucially, we found that the medial part of PMC, putatively covering the supplementary motor area, plays a direct role in object grasping. In concert with findings in nonhuman primates, these results indicate a multifocal representation of object geometry for grasping in the PMC and expand our understanding of how our brain integrates visual and motor information to perform visually guided actions.

Soft Ngram Representation and Modeling for Protein Remote Homology Detection.

Remote homology detection represents a central problem in bioinformatics, where the challenge is to detect functionally related proteins when their sequence similarity is low. Recent solutions employ representations derived from the sequence profile, obtained by replacing each amino acid of the sequence by the corresponding most probable amino acid in the profile. However, the information contained in the profile could be exploited more deeply, provided that there is a representation able to capture and properly model such crucial evolutionary information. In this paper, we propose a novel profile-based representation for sequences, called soft Ngram. This representation, which extends the traditional Ngram scheme (obtained by grouping N consecutive amino acids), permits considering all of the evolutionary information in the profile: this is achieved by extracting Ngrams from the whole profile, equipping them with a weight directly computed from the corresponding evolutionary frequencies. We illustrate two different approaches to model the proposed representation and to derive a feature vector, which can be effectively used for classification using a support vector machine (SVM). A thorough evaluation on three benchmarks demonstrates that the new approach outperforms other Ngram-based methods, and shows very promising results also in comparison with a broader spectrum of techniques.

Traveling on discrete embeddings of gene expression.

OBJECTIVE: High-throughput technologies have generated an unprecedented amount of high-dimensional gene expression data. Algorithmic approaches could be extremely useful to distill information and derive compact interpretable representations of the statistical patterns present in the data. This paper proposes a mining approach to extract an informative representation of gene expression profiles based on a generative model called the Counting Grid (CG). METHOD: Using the CG model, gene expression values are arranged on a discrete grid, learned in a way that "similar" co-expression patterns are arranged in close proximity, thus resulting in an intuitive visualization of the dataset. More than this, the model permits to identify the genes that distinguish between classes (e.g. different types of cancer). Finally, each sample can be characterized with a discriminative signature - extracted from the model - that can be effectively employed for classification. RESULTS: A thorough evaluation on several gene expression datasets demonstrate the suitability of the proposed approach from a twofold perspective: numerically, we reached state-of-the-art classification accuracies on 5 datasets out of 7, and similar results when the approach is tested in a gene selection setting (with a stability always above 0.87); clinically, by confirming that many of the genes highlighted by the model as significant play also a key role for cancer biology. CONCLUSION: The proposed framework can be successfully exploited to meaningfully visualize the samples; detect medically relevant genes; properly classify samples.

A Multimodal Approach for Protein Remote Homology Detection.

Protein remote homology detection represents a crucial and challenging task in bioinformatics: even if effective methods appeared in recent years, in several cases a proper characterization of remote evolutionary correlation can not be derived. In such situations, it may be possible that information derived from other sources helps, provided that it is possible to properly integrate such (even partial) information into existing models. In this paper, we provide some evidence that this route is feasible: inspired by the multimodal retrieval literature, we show how it is possible to exploit a simple multimodal approach to improve a model learned from a set of sequences, by using knowledge derived from a partial set of corresponding 3D structures. We investigate (with the SCOP 1.53 benchmark) the suitability of the proposed multimodal scheme, showing that a beneficial effect can be obtained even when a very reduced amount of structures are available. A further detailed analysis on a member of the GPCR superfamily confirms that this multimodal approach can extract information that cannot be obtained from sequence-based techniques.

In silico deconstruction of ATP-competitive inhibitors of glycogen synthase kinase-3ß.

Fragment-based methods have emerged in the last two decades as alternatives to traditional high throughput screenings for the identification of chemical starting points in drug discovery. One arguable yet popular assumption about fragment-based design is that the fragment binding mode remains conserved upon chemical expansion. For instance, the question of the binding conservation upon fragmentation of a molecule is still unclear. A number of papers have challenged this hypothesis by means of experimental techniques, with controversial results, "underlining" the idea that a simple generalization, maybe, is not possible. From a computational standpoint, the issue has been rarely addressed and mostly to test novel protocols on limited data sets. To fill this gap, we here report on a computational retrospective study concerned with the in silico deconstruction of leadlike compounds, active on the pharmaceutically relevant enzyme glycogen synthase kinase-3ß.

Investigating topic models' capabilities in expression microarray data classification.

In recent years a particular class of probabilistic graphical models-called topic models-has proven to represent an useful and interpretable tool for understanding and mining microarray data. In this context, such models have been almost only applied in the clustering scenario, whereas the classification task has been disregarded by researchers. In this paper, we thoroughly investigate the use of topic models for classification of microarray data, starting from ideas proposed in other fields (e.g., computer vision). A classification scheme is proposed, based on highly interpretable features extracted from topic models, resulting in a hybrid generative-discriminative approach; an extensive experimental evaluation, involving 10 different literature benchmarks, confirms the suitability of the topic models for classifying expression microarray data.

The grapevine expression atlas reveals a deep transcriptome shift driving the entire plant into a maturation program.

We developed a genome-wide transcriptomic atlas of grapevine (Vitis vinifera) based on 54 samples representing green and woody tissues and organs at different developmental stages as well as specialized tissues such as pollen and senescent leaves. Together, these samples expressed ∼91% of the predicted grapevine genes. Pollen and senescent leaves had unique transcriptomes reflecting their specialized functions and physiological status. However, microarray and RNA-seq analysis grouped all the other samples into two major classes based on maturity rather than organ identity, namely, the vegetative/green and mature/woody categories. This division represents a fundamental transcriptomic reprogramming during the maturation process and was highlighted by three statistical approaches identifying the transcriptional relationships among samples (correlation analysis), putative biomarkers (O2PLS-DA approach), and sets of strongly and consistently expressed genes that define groups (topics) of similar samples (biclustering analysis). Gene coexpression analysis indicated that the mature/woody developmental program results from the reiterative coactivation of pathways that are largely inactive in vegetative/green tissues, often involving the coregulation of clusters of neighboring genes and global regulation based on codon preference. This global transcriptomic reprogramming during maturation has not been observed in herbaceous annual species and may be a defining characteristic of perennial woody plants.

Combination of atomic force microscopy and principal component analysis as a general method for direct recognition of functional and structural domains in nanonocomposite materials.

In this work, we report a simple method to direct identify nanometer sized textures in composite materials by means of AFM spectroscopy, aiming at recognizing structured region to be further investigated. It consists in acquiring a set of dynamic data organized in spectroscopy maps and subsequently extracting most valuable information by means of the principal component analysis (PCA) method. This algorithm projects the information of D spectroscopy curves, each containing P data, acquired at each point of an LxC grid into a subset of LxC maps without any assumption on the sample structure, filtering out redundancies and noise. As a consequence, a huge amount of 3D data is condensed into few 2D maps, easy to be examined. Results of this algorithm allow to find and locate regions of interest within the map, allowing a further reduction of data series to be extensively analyzed or modeled. In this work, we explain the main features of the method and show its application on a nanocomposite sample. Microsc. Res. Tech. 73:973-981, 2010. © 2010 Wiley-Liss, Inc.

Bag of Peaks: interpretation of NMR spectrometry.

MOTIVATION: The analysis of high-resolution proton nuclear magnetic resonance (NMR) spectrometry can assist human experts to implicate metabolites expressed by diseased biofluids. Here, we explore an intermediate representation, between spectral trace and classifier, able to furnish a communicative interface between expert and machine. This representation permits equivalent, or better, classification accuracies than either principal component analysis (PCA) or multi-dimensional scaling (MDS). In the training phase, the peaks in each trace are detected and clustered in order to compile a common dictionary, which could be visualized and adjusted by an expert. The dictionary is used to characterize each trace with a fixed-length feature vector, termed Bag of Peaks, ready to be classified with classical supervised methods. RESULTS: Our small-scale study, concerning Type I diabetes in Sardinian children, provides a preliminary indication of the effectiveness of the Bag of Peaks approach over standard PCA and MDS. Consistently, higher classification accuracies are obtained once a sufficient number of peaks (>10) are included in the dictionary. A large-scale simulation of noisy spectra further confirms this advantage. Finally, suggestions for metabolite-peak loci that may be implicated in the disease are obtained by applying standard feature selection techniques.

Multimodal phylogeny for taxonomy: integrating information from nucleotide and amino acid sequences.

The crucial role played by the analysis of microbial diversity in biotechnology-based innovations has increased the interest in the microbial taxonomy research area. Phylogenetic sequence analyses have contributed significantly to the advances in this field, also in the view of the large amount of sequence data collected in recent years. Phylogenetic analyses could be realized on the basis of protein-encoding nucleotide sequences or encoded amino acid molecules: these two mechanisms present different peculiarities, still starting from two alternative representations of the same information. This complementarity could be exploited to achieve a multimodal phylogenetic scheme that is able to integrate gene and protein information in order to realize a single final tree. This aspect has been poorly addressed in the literature. In this paper, we propose to integrate the two phylogenetic analyses using basic schemes derived from the multimodality fusion theory (or multiclassifier systems theory), a well-founded and rigorous branch for which its powerfulness has already been demonstrated in other pattern recognition contexts. The proposed approach could be applied to distance matrix-based phylogenetic techniques (like neighbor joining), resulting in a smart and fast method. The proposed methodology has been tested in a real case involving sequences of some species of lactic acid bacteria. With this dataset, both nucleotide sequence- and amino acid sequence-based phylogenetic analyses present some drawbacks, which are overcome with the multimodal analysis.

Investigating Hidden Markov Models' capabilities in 2D shape classification.

In this paper, Hidden Markov Models (HMMs) are investigated for the purpose of classifying planar shapes represented by their curvature coefficients. In the training phase, special attention is devoted to the initialization and model selection issues, which make the learning phase particularly effective. The results of tests on different data sets show that the proposed system is able to accurately classify objects that were translated, rotated, occluded, or deformed by shearing, also in the presence of noise.