Krisztian Buza

Resting State fMRI Functional Connectivity Analysis Using Dynamic Time Warping

Traditional resting-state network concept is based on calculating linear dependence of spontaneous low frequency fluctuations of the BOLD signals of different brain areas, which assumes temporally stable zero-lag synchrony across regions. However, growing amount of experimental findings suggest that functional connectivity exhibits dynamic changes and a complex time-lag structure, which cannot be captured by the static zero-lag correlation analysis. Here we propose a new approach applying Dynamic Time Warping (DTW) distance to evaluate functional connectivity strength that accounts for non-stationarity and phase-lags between the observed signals. Using simulated fMRI data we found that DTW captures dynamic interactions and it is less sensitive to linearly combined global noise in the data as compared to traditional correlation analysis. We tested our method using resting-state fMRI data from repeated measurements of an individual subject and showed that DTW analysis results in more stable connectivity patterns by reducing the within-subject variability and increasing robustness for preprocessing strategies. Classification results on a public dataset revealed a superior sensitivity of the DTW analysis to group differences by showing that DTW based classifiers outperform the zero-lag correlation and maximal lag cross-correlation based classifiers significantly. Our findings suggest that analysing resting-state functional connectivity using DTW provides an efficient new way for characterizing functional networks.

Resting state fMRI functional connectivity-based classification using a convolutional neural network architecture

Machine learning techniques have become increasingly popular in the field of resting state fMRI (functional magnetic resonance imaging) network based classification. However, the application of convolutional networks has been proposed only very recently and has remained largely unexplored. In this paper we describe a convolutional neural network architecture for functional connectome classification called connectome-convolutional neural network (CCNN). Our results on simulated datasets and a publicly available dataset for amnestic mild cognitive impairment classification demonstrate that our CCNN model can efficiently distinguish between subject groups. We also show that the connectome-convolutional network is capable to combine information from diverse functional connectivity metrics and that models using a combination of different connectivity descriptors are able to outperform classifiers using only one metric. From this flexibility follows that our proposed CCNN model can be easily adapted to a wide range of connectome based classification or regression tasks, by varying which connectivity descriptor combinations are used to train the network.

Classification of fMRI data using dynamic time warping based functional connectivity analysis

The synchronized spontaneous low frequency fluctuations of the BOLD signal, as captured by functional MRI measurements, is known to represent the functional connections of different brain areas. The aforementioned MRI measurements result in high-dimensional time series, the dimensions of which correspond to the activity of different brain regions. Recently we have shown that Dynamic Time Warping (DTW) distance can be used as a similarity measure between BOLD signals of brain regions as an alternative of the traditionally used correlation coefficient. We have characterized the new metrics stability in multiple measurements, and between subjects in homogenous groups. In this paper we investigated the DTW metrics sensitivity and demonstrated that DTW-based models outperform correlation-based models in resting-state fMRI data classification tasks. Additionally, we show that functional connectivity networks resulting from DTW-based models as compared to the correlation-based models are more stable and sensitive to differences between healthy subjects and patient groups.

Drug-target interaction prediction: A Bayesian ranking approach

BACKGROUND AND OBJECTIVE In silico prediction of drug-target interactions (DTI) could provide valuable information and speed-up the process of drug repositioning - finding novel usage for existing drugs. In our work, we focus on machine learning algorithms supporting drug-centric repositioning approach, which aims to find novel usage for existing or abandoned drugs. We aim at proposing a per-drug ranking-based method, which reflects the needs of drug-centric repositioning research better than conventional drug-target prediction approaches. METHODS We propose Bayesian Ranking Prediction of Drug-Target Interactions (BRDTI). The method is based on Bayesian Personalized Ranking matrix factorization (BPR) which has been shown to be an excellent approach for various preference learning tasks, however, it has not been used for DTI prediction previously. In order to successfully deal with DTI challenges, we extended BPR by proposing: (i) the incorporation of target bias, (ii) a technique to handle new drugs and (iii) content alignment to take structural similarities of drugs and targets into account. RESULTS Evaluation on five benchmark datasets shows that BRDTI outperforms several state-of-the-art approaches in terms of per-drug nDCG and AUC. BRDTI results w.r.t. nDCG are 0.929, 0.953, 0.948, 0.897 and 0.690 for G-Protein Coupled Receptors (GPCR), Ion Channels (IC), Nuclear Receptors (NR), Enzymes (E) and Kinase (K) datasets respectively. Additionally, BRDTI significantly outperformed other methods (BLM-NII, WNN-GIP, NetLapRLS and CMF) w.r.t. nDCG in 17 out of 20 cases. Furthermore, BRDTI was also shown to be able to predict novel drug-target interactions not contained in the original datasets. The average recall at top-10 predicted targets for each drug was 0.762, 0.560, 1.000 and 0.404 for GPCR, IC, NR, and E datasets respectively. CONCLUSIONS Based on the evaluation, we can conclude that BRDTI is an appropriate choice for researchers looking for an in silico DTI prediction technique to be used in drug-centric repositioning scenarios. BRDTI Software and supplementary materials are available online at www.ksi.mff.cuni.cz/∼peska/BRDTI.

Classification of gene expression data

BACKGROUND AND OBJECTIVE Classification of gene expression data is the common denominator of various biomedical recognition tasks. However, obtaining class labels for large training samples may be difficult or even impossible in many cases. Therefore, semi-supervised classification techniques are required as semi-supervised classifiers take advantage of unlabeled data. METHODS Gene expression data is high-dimensional which gives rise to the phenomena known under the umbrella of the curse of dimensionality, one of its recently explored aspects being the presence of hubs or hubness for short. Therefore, hubness-aware classifiers have been developed recently, such as Naive Hubness-Bayesian k-Nearest Neighbor (NHBNN). In this paper, we propose a semi-supervised extension of NHBNN which follows the self-training schema. As one of the core components of self-training is the certainty score, we propose a new hubness-aware certainty score. RESULTS We performed experiments on publicly available gene expression data. These experiments show that the proposed classifier outperforms its competitors. We investigated the impact of each of the components (classification algorithm, semi-supervised technique, hubness-aware certainty score) separately and showed that each of these components are relevant to the performance of the proposed approach. CONCLUSIONS Our results imply that our approach may increase classification accuracy and reduce computational costs (i.e., runtime). Based on the promising results presented in the paper, we envision that hubness-aware techniques will be used in various other biomedical machine learning tasks. In order to accelerate this process, we made an implementation of hubness-aware machine learning techniques publicly available in the PyHubs software package (http://www.biointelligence.hu/pyhubs) implemented in Python, one of the most popular programming languages of data science.

Drug–target interaction prediction with Bipartite Local Models and hubness-aware regression

Abstract Computational prediction of drug–target interactions is an essential task with various applications in the pharmaceutical industry, such as adverse effect prediction or drug repositioning. Recently, expert systems based on machine learning have been applied to drug–target interaction prediction. Although hubness-aware machine learning techniques are among the most promising approaches, their potential to enhance drug–target interaction prediction methods has not been exploited yet. In this paper, we extend the Bipartite Local Model (BLM), one of the most prominent interaction prediction methods. In particular, we use BLM with a hubness-aware regression technique, ECkNN. We represent drugs and targets in the similarity space with rich set of features (i.e., chemical, genomic and interaction features), and build a projection-based ensemble of BLMs. In order to assist reproducibility of our work as well as comparison to published results, we perform experiments on widely used publicly available drug–target interaction datasets. The results show that our approach outperforms state-of-the-art drug–target prediction techniques. Additionally, we demonstrate the feasibility of predictions from the point of view of applications.

ParkinsoNET: Estimation of UPDRS Score Using Hubness-Aware Feedforward Neural Networks

ABSTRACT Parkinson’s disease is a worldwide, frequent, neurodegenerative disorder with increasing incidence. Speech disturbance appears during the progression of the disease. The Unified Parkinson’s Disease Rating Scale (UPDRS) is a gold-standard tool for diagnosis and follow-up of the disease. We aim at estimating the UPDRS score based on biomedical voice recordings. In this article, we study the hubness phenomenon in context of the UPDRS score estimation and propose hubness-aware error correction for feedforward neural networks to increase the accuracy of estimation. We perform experiments on publicly available datasets derived from real-voice data and show that the proposed technique systematically increases the accuracy of various feedforward neural networks.

A Model for Classification Based on the Functional Connectivity Pattern Dynamics of the Brain

Synchronized spontaneous low frequency fluctuations of the so called BOLD signal, as measured by functional Magnetic Resonance Imaging (fMRI), are known to represent the functional connections of different brain areas. Dynamic Time Warping (DTW) distance can be used as a similarity measure between BOLD signals of brain regions as an alternative of the traditionally used correlation coefficient and the usage of the DTW algorithm has further advantages: beside the DTW distance, the algorithm generates the warping path, i.e. the time-delay function between the compared two time-series. In this paper, we propose to use the relative length of the warping path as classification feature and demonstrate that the warping path itself carries important information when classifying patients according to cannabis addiction. We discuss biomedical relevance of our findings as well.

Feature Selection with a Genetic Algorithm for Classification of Brain Imaging Data

Recent advances in brain imaging technology, coupled with large-scale brain research projects, such as the BRAIN initiative in the U.S. and the European Human Brain Project, allow us to capture brain activity in unprecedented details. In principle, the observed data is expected to substantially shape our knowledge about brain activity, which includes the development of new biomarkers of brain disorders. However, due to the high dimensionality, the analysis of the data is challenging, and selection of relevant features is one of the most important analytic tasks. In many cases, due to the complexity of search space, evolutionary algorithms are appropriate to solve the aforementioned task. In this chapter, we consider the feature selection task from the point of view of classification tasks related to functional magnetic resonance imaging (fMRI) data. Furthermore, we present an empirical comparison of conventional LASSO-based feature selection and a novel feature selection approach designed for fMRI data based on a simple genetic algorithm.

How you type is who you are

The increasing interest in person identification based on typing patterns may be attributed to several factors. First, cheap and widely applicable person identification is essential due to wide-spread usage of internet based services, such as online courses or internet banking. Furthermore, introduction of new approaches is necessary because of the continuous development of attack techniques against existing identification methods. The dynamics of typing is characteristic to particular users, while a user is hardly able to mimic the typing dynamics of other users. According to recent observations, person identification based on machine learning using data about the dynamics of typing works surprisingly well. Hubness-aware regression techniques have been introduced recently, however they have not been applied to person identification previously. In this paper, we propose to use ECkNN, a hubness-aware regression technique together with dynamic time warping for person identification. We collected time-series data describing the dynamics of typing and used it to evaluate our approach. As baseline we used state-of-the-art time-series classifiers. Experimental results show that the proposed technique outperforms the baselines. In order to assist reproducibility of our work, we publish the data we collected.