Discrete Infomax Codes for Supervised Representation Learning.

For high-dimensional data such as images, learning an encoder that can output a compact yet informative representation is a key task on its own, in addition to facilitating subsequent processing of data. We present a model that produces discrete infomax codes (DIMCO); we train a probabilistic encoder that yields k-way d-dimensional codes associated with input data. Our model maximizes the mutual information between codes and ground-truth class labels, with a regularization which encourages entries of a codeword to be statistically independent. In this context, we show that the infomax principle also justifies existing loss functions, such as cross-entropy as its special cases. Our analysis also shows that using shorter codes reduces overfitting in the context of few-shot classification, and our various experiments show this implicit task-level regularization effect of DIMCO. Furthermore, we show that the codes learned by DIMCO are efficient in terms of both memory and retrieval time compared to prior methods.

Flexible Ternary Combination of Gellan Gum, Sodium Carboxymethyl Cellulose, and Silicon Dioxide Nanocomposites Fabricated by Quaternary Ammonium Silane: Rheological, Thermal, and Antimicrobial Properties.

Gellan gum-sodium carboxymethyl cellulose (GC)-based composite films with various concentrations of silicon dioxide (SiO2) nanoparticles and octadecyldimethyl-(3-triethoxy silylpropyl)ammonium chloride (ODDMAC) were successfully prepared by the traditional solution casting method to improve the antimicrobial and water repellent properties. Fourier transform infrared (FT-IR) spectra confirm the formation of hydrogen bonds between the GC and nano-SiO2. The microstructure and physicochemical properties were investigated by FT-IR, wide-angle X-ray diffraction, and scanning electron microscopy (SEM) analyses. The rheological properties of the GC-SiO2 hydrogel were also characterized. The results show that the inclusion of SiO2 nanoparticles significantly improved the viscosity and viscoelastic properties of the GC hydrogel. The GC-SiO2 hydrogel exhibited shear-thinning behavior and its viscosity decreased at high shear rates. The storage and loss moduli of the GC composites increased as the frequency and SiO2 concentration increased. The tensile strength and elongation at break of the GC composites increased by 75.9 and 62%, respectively, with the addition of SiO2 and ODDMAC. In addition, nano-SiO2 decreased the water vapor permeability and increased the hydrophobic properties of the GC-SiO2 composites. Thermogravimetric analysis showed that the T 5% loss was in the range of 99.4-128.6 °C and the char yield was in the range of 20.1-29.9%, which was significantly enhanced by the incorporation of SiO2 nanoparticles. The GC-SiO2 (ODDMAC) nanocomposites effectively shielded the UV light and exhibited high antimicrobial activity against six different pathogens. The simple and cost-effective GC-SiO2 (ODDMAC) nanocomposites gained importance in food packaging and biomedical applications.

Population-Based Epidemiology of Pediatric Patients with Treated Tic Disorders from Real-World Evidence in Korea.

Objectives: Although tic disorder (TD) is a common mental disorder in children and adolescents, epidemiological data based on real-world evidence (RWE) are insufficient. Using RWE, this study sought to examine the prevalence of treated TD, use of medical utilization, and use of prescription drugs among patients with TD with respect to TD type and comorbid psychiatric illness. Methods: We performed a retrospective cross-sectional study. Using the Korean Health Insurance Review and Assessment Service Pediatric Patient Sample data from 2009 to 2016, we analyzed 20,599 patients with TD (Korean Standard Classification of Diseases-6/7 code: F95.x) aged 2-19 years. Results: The annual average TD prevalence was 2.6/1000 population (95% confidence interval, 2.3-2.8/1000). Between 2009 and 2016, a slight increase in TD prevalence was observed from 1.9 to 2.9/1000 population. The TD prevalence rate in male patients was four times higher than that in female patients. Differences were observed in health care utilization and drug prescription types between patients with Tourette syndrome and chronic or transient TD. In addition, more than half of patients with TD had comorbid psychiatric disorders, and one-third of patients with TD had attention-deficit/hyperactivity disorder (ADHD). Patients with TD without comorbidities were frequently prescribed aripiprazole, while patients with TD and comorbid ADHD were frequently prescribed atomoxetine, methylphenidate, risperidone, and aripiprazole. Conclusion: This study described the epidemiological characteristics of TD based on recent RWE from Korea, and its findings can help establish future TD evidence-based clinical guidelines and related policies.

What is a Right Distal Fusion Level for Prevention of Sagittal Imbalance in Multilevel Posterior Cervical Spine Surgery: C7 or T1?

STUDY DESIGN: This was a retrospective cohort study. OBJECTIVE: To evaluate the sagittal alignment and T1 slope after multilevel posterior cervical fusion surgery depending on the distal fusion level; C7 or T1, and find out the appropriate distal fusion level. SUMMARY OF BACKGROUND DATA: The sagittal balance of the cervical spine is known to be affected by cervical lordosis and T1 slope. However, T1 slope is not a constant parameter that can be frequently changed after the surgery. Furthermore, useful studies to help guide surgeons in decision-making as to the most appropriate distal level of fusion for cervical sagittal balance are very limited. MATERIALS AND METHODS: From 2014 to 2015, 50 patients who underwent multilevel posterior cervical fusion surgery were evaluated and followed up for >2 years. Group 1 was composed of 29 patients whose distal fusion level was C7. Group 2 was composed of 21 patients whose distal fusion level was T1. C1-C2 lordosis, C2-C7 lordosis, C2-C7 sagittal vertical axis (SVA), and T1 slope were measured on preoperative and the last follow-up. RESULTS: In group 1, C2-C7 SVA (23.1→30.4 mm, P=0.043) was worsened, and T1 slope (22.3→32.9 degrees, P=0.001) was increased after the surgery. In group 2, no significant change occurred in C2-C7 SVA after the surgery (25.3 →23.6 mm, P=0.648). The last follow-up T1 slope was similar with preoperative T1 slope (22.7→21.8 degrees, P=0.04) in group 2. CONCLUSIONS: This study showed that sagittal alignment became worse after the multilevel posterior cervical surgery when distal fusion level was stopped at C7, which was associated with increase of T1 slope. However, when we extended the distal fusion level to T1, T1 slope was not changed after the surgery. Therefore, sagittal alignment was maintained after the surgery. On the basis of the results of this study, we recommend distal fusion extends to T1. LEVEL OF EVIDENCE: Level III.

Author Correction: Interrupting oral infection of Porphyromonas gingivalis with anti-FimA antibody attenuates bacterial dissemination to the arthritic joint and improves experimental arthritis.

After online publication of this article, the authors noticed an error in the Figure section. The correct statement of this article should have read as below.

Kruppel-Like Factor 4 Positively Regulates Autoimmune Arthritis in Mouse Models and Rheumatoid Arthritis in Patients via Modulating Cell Survival and Inflammation Factors of Fibroblast-Like Synoviocyte.

Rheumatoid arthritis (RA) is a chronic autoimmune disease that causes mild to severe joint inflammation. During RA pathogenesis, fibroblast-like synoviocytes (FLS) acquire a tumor-like phenotype and mediate cartilage destruction both directly and indirectly by producing proinflammatory cytokines and matrix metalloproteinases (MMPs). Kruppel-like factor (KLF) 4, a member of the KLF family, plays significant roles in cell survival, proliferation, and differentiation. A recent study reported increased expression of KLF4 in synovial tissue from RA patients. However, its precise role in RA in different models, including mouse autoimmune disease models, remains unclear. In this study, we examined the role of KLF4 during development of autoimmune arthritis in mouse models. To do this, we used KLF4 knockout mice rendered by ribonucleic acid (RNA)-guided endonuclease (RGEN) and performed collagen antibody-induced arthritis (CAIA). We found that deletion of KLF4 reduces inflammation induced by CAIA. In addition, we assessed collagen-induced arthritis (CIA) in control mice and KLF4-overexpressing mice generated by a minicircle vector treatment. Severity of CIA in mice overexpressing KLF4 was greater than that in mice injected with control vector. Finally, we verified the inflammatory roles of KLF4 in CIA by treating Kenpaullone which is used as KLF4 inhibitor. Next, we focused on human/mouse FLS to discover the cellular process involved in RA pathogenesis including proliferation, apoptosis, and inflammation including MMPs. In FLS, KLF4 upregulated expression of mRNA encoding proinflammatory cytokines interleukin (IL)-1ß and IL-6. KLF4 also regulated expression of matrix metallopeptidase 13 in the synovium. We found that blockade of KLF4 in FLS increased apoptosis and suppressed proliferation followed by downregulation of antiapoptotic factor BCL2. Our results indicate that KLF4 plays a crucial role in pathogenesis of inflammatory arthritis in vivo, by regulating apoptosis, MMP expression, and cytokine expression by FLS. Thus, KLF4 might be a novel transcription factor for generating RA by modulating cellular process of FLS.

Interrupting oral infection of Porphyromonas gingivalis with anti-FimA antibody attenuates bacterial dissemination to the arthritic joint and improves experimental arthritis.

Rheumatoid arthritis (RA) is a chronic autoimmune disease that typically results in strong inflammation and bone destruction in the joints. It is generally known that the pathogenesis of RA is linked to cardiovascular and periodontal diseases. Though rheumatoid arthritis and periodontitis share many pathologic features such as a perpetual inflammation and bone destruction, the precise mechanism underlying a link between these two diseases has not been fully elucidated. Collagen-induced arthritis (CIA) mice were orally infected with Porphyromonas gingivalis (Pg) or Pg preincubated with an anti-FimA antibody (FimA Ab) specific for fimbriae that are flexible appendages on the cell surface. Pg-infected CIA mice showed oral microbiota disruption and increased alveolar bone loss and had synovitis and joint bone destruction. However, preincubation with FimA Ab led to a significant reduction in the severity of both oral disease and arthritis. Moreover, FimA Ab attenuated bacterial attachment and aggregation on human gingival and rheumatoid arthritis synovial fibroblasts. In addition, we discovered bacteria may utilize dendritic cells, macrophages and neutrophils to migrate into the joints of CIA mice. These results suggest that disrupting Pg fimbriae function by FimA Ab ameliorates RA.

Publisher Correction: Manipulation of Rat Movement via Nigrostriatal Stimulation Controlled by Human Visually Evoked Potentials.

A correction to this article has been published and is linked from the HTML version of this paper. The error has been fixed in the paper.

Manipulation of Rat Movement via Nigrostriatal Stimulation Controlled by Human Visually Evoked Potentials.

Here, we report that the development of a brain-to-brain interface (BBI) system that enables a human user to manipulate rat movement without any previous training. In our model, the remotely-guided rats (known as ratbots) successfully navigated a T-maze via contralateral turning behaviour induced by electrical stimulation of the nigrostriatal (NS) pathway by a brain- computer interface (BCI) based on the human controller's steady-state visually evoked potentials (SSVEPs). The system allowed human participants to manipulate rat movement with an average success rate of 82.2% and at an average rat speed of approximately 1.9 m/min. The ratbots had no directional preference, showing average success rates of 81.1% and 83.3% for the left- and right-turning task, respectively. This is the first study to demonstrate the use of NS stimulation for developing a highly stable ratbot that does not require previous training, and is the first instance of a training-free BBI for rat navigation. The results of this study will facilitate the development of borderless communication between human and untrained animals, which could not only improve the understanding of animals in humans, but also allow untrained animals to more effectively provide humans with information obtained with their superior perception.

The SWI/SNF chromatin remodeling complex regulates germinal center formation by repressing Blimp-1 expression.

Germinal center (GC) reaction is crucial in adaptive immune responses. The formation of GC is coordinated by the expression of specific genes including Blimp-1 and Bcl-6. Although gene expression is critically influenced by the status of chromatin structure, little is known about the role of chromatin remodeling factors for regulation of GC formation. Here, we show that the SWI/SNF chromatin remodeling complex is required for GC reactions. Mice lacking Srg3/mBaf155, a core component of the SWI/SNF complex, showed impaired differentiation of GC B and follicular helper T cells in response to T cell-dependent antigen challenge. The SWI/SNF complex regulates chromatin structure at the Blimp-1 locus and represses its expression by interacting cooperatively with Bcl-6 and corepressors. The defect in GC reactions in mice lacking Srg3 was due to the derepression of Blimp-1 as supported by genetic studies with Blimp-1-ablated mice. Hence, our study identifies the SWI/SNF complex as a key mediator in GC reactions by modulating Bcl-6-dependent Blimp-1 repression.

A hybrid NIRS-EEG system for self-paced brain computer interface with online motor imagery.

BACKGROUND: For a self-paced motor imagery based brain-computer interface (BCI), the system should be able to recognize the occurrence of a motor imagery, as well as the type of the motor imagery. However, because of the difficulty of detecting the occurrence of a motor imagery, general motor imagery based BCI studies have been focusing on the cued motor imagery paradigm. NEW METHOD: In this paper, we present a novel hybrid BCI system that uses near infrared spectroscopy (NIRS) and electroencephalography (EEG) systems together to achieve online self-paced motor imagery based BCI. We designed a unique sensor frame that records NIRS and EEG simultaneously for the realization of our system. Based on this hybrid system, we proposed a novel analysis method that detects the occurrence of a motor imagery with the NIRS system, and classifies its type with the EEG system. RESULTS: An online experiment demonstrated that our hybrid system had a true positive rate of about 88%, a false positive rate of 7% with an average response time of 10.36 s. COMPARISON WITH EXISTING METHOD(S): As far as we know, there is no report that explored hemodynamic brain switch for self-paced motor imagery based BCI with hybrid EEG and NIRS system. CONCLUSIONS: From our experimental results, our hybrid system showed enough reliability for using in a practical self-paced motor imagery based BCI.

TEMPI: probabilistic modeling time-evolving differential PPI networks with multiPle information.

MOTIVATION: Time-evolving differential protein-protein interaction (PPI) networks are essential to understand serial activation of differentially regulated (up- or downregulated) cellular processes (DRPs) and their interplays over time. Despite developments in the network inference, current methods are still limited in identifying temporal transition of structures of PPI networks, DRPs associated with the structural transition and the interplays among the DRPs over time. RESULTS: Here, we present a probabilistic model for estimating Time-Evolving differential PPI networks with MultiPle Information (TEMPI). This model describes probabilistic relationships among network structures, time-course gene expression data and Gene Ontology biological processes (GOBPs). By maximizing the likelihood of the probabilistic model, TEMPI estimates jointly the time-evolving differential PPI networks (TDNs) describing temporal transition of PPI network structures together with serial activation of DRPs associated with transiting networks. This joint estimation enables us to interpret the TDNs in terms of temporal transition of the DRPs. To demonstrate the utility of TEMPI, we applied it to two time-course datasets. TEMPI identified the TDNs that correctly delineated temporal transition of DRPs and time-dependent associations between the DRPs. These TDNs provide hypotheses for mechanisms underlying serial activation of key DRPs and their temporal associations. AVAILABILITY AND IMPLEMENTATION: Source code and sample data files are available at http://sbm.postech.ac.kr/tempi/sources.zip. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Bayesian common spatial patterns for multi-subject EEG classification.

Multi-subject electroencephalography (EEG) classification involves algorithm development for automatically categorizing brain waves measured from multiple subjects who undergo the same mental task. Common spatial patterns (CSP) or its probabilistic counterpart, PCSP, is a popular discriminative feature extraction method for EEG classification. Models in CSP or PCSP are trained on a subject-by-subject basis so that inter-subject information is neglected. In the case of multi-subject EEG classification, however, it is desirable to capture inter-subject relatedness in learning a model. In this paper we present a nonparametric Bayesian model for a multi-subject extension of PCSP where subject relatedness is captured by assuming that spatial patterns across subjects share a latent subspace. Spatial patterns and the shared latent subspace are jointly learned by variational inference. We use an infinite latent feature model to automatically infer the dimension of the shared latent subspace, placing Indian Buffet process (IBP) priors on our model. Numerical experiments on BCI competition III IVa and IV 2a dataset demonstrate the high performance of our method, compared to PCSP and existing Bayesian multi-task CSP models.

Inference of dynamic networks using time-course data.

Cells execute their functions through dynamic operations of biological networks. Dynamic networks delineate the operation of biological networks in terms of temporal changes of abundances or activities of nodes (proteins and RNAs), as well as formation of new edges and disappearance of existing edges over time. Global genomic and proteomic technologies can be used to decode dynamic networks. However, using these experimental methods, it is still challenging to identify temporal transition of nodes and edges. Thus, several computational methods for estimating dynamic topological and functional characteristics of networks have been introduced. In this review, we summarize concepts and applications of these computational methods for inferring dynamic networks and further summarize methods for estimating spatial transition of biological networks.

GOM-Face: GKP, EOG, and EMG-based multimodal interface with application to humanoid robot control.

We present a novel human-machine interface, called GOM-Face , and its application to humanoid robot control. The GOM-Face bases its interfacing on three electric potentials measured on the face: 1) glossokinetic potential (GKP), which involves the tongue movement; 2) electrooculogram (EOG), which involves the eye movement; 3) electromyogram, which involves the teeth clenching. Each potential has been individually used for assistive interfacing to provide persons with limb motor disabilities or even complete quadriplegia an alternative communication channel. However, to the best of our knowledge, GOM-Face is the first interface that exploits all these potentials together. We resolved the interference between GKP and EOG by extracting discriminative features from two covariance matrices: a tongue-movement-only data matrix and eye-movement-only data matrix. With the feature extraction method, GOM-Face can detect four kinds of horizontal tongue or eye movements with an accuracy of 86.7% within 2.77 s. We demonstrated the applicability of the GOM-Face to humanoid robot control: users were able to communicate with the robot by selecting from a predefined menu using the eye and tongue movements.

Common spatial patterns for steady-state somatosensory evoked potentials.

Steady-state somatosensory evoked potential (SSSEP) is a recently developing brain-computer interface (BCI) paradigm where the brain response to tactile stimulation of a specific frequency is used. Thus far, spatial information was not examined in depth in SSSEP BCI, because frequency information was regarded as the main concern of SSSEP analysis. However, given that the somatosensory cortex areas, each of which correspond to a different body part, are well clustered, we can assume that the spatial information could be beneficial for SSSEP analysis. Based on this assumption, we apply the common spatial pattern (CSP) method, which is the spatial feature extraction method most widely used for the motor imagery BCI paradigm, to SSSEP BCI. Experimental results show that our approach, where two CSP methods are applied to the signal of each frequency band, has a performance improvement from 70% to 75%.

Parameter learning for alpha integration.

In pattern recognition, data integration is an important issue, and when properly done, it can lead to improved performance. Also, data integration can be used to help model and understand multimodal processing in the brain. Amari proposed α-integration as a principled way of blending multiple positive measures (e.g., stochastic models in the form of probability distributions), enabling an optimal integration in the sense of minimizing the α-divergence. It also encompasses existing integration methods as its special case, for example, a weighted average and an exponential mixture. The parameter α determines integration characteristics, and the weight vector w assigns the degree of importance to each measure. In most work, however, α and w are given in advance rather than learned. In this letter, we present a parameter learning algorithm for learning α and ω from data when multiple integrated target values are available. Numerical experiments on synthetic as well as real-world data demonstrate the effectiveness of the proposed method.

Identification of co-occurring transcription factor binding sites from DNA sequence using clustered position weight matrices.

Accurate prediction of transcription factor binding sites (TFBSs) is a prerequisite for identifying cis-regulatory modules that underlie transcriptional regulatory circuits encoded in the genome. Here, we present a computational framework for detecting TFBSs, when multiple position weight matrices (PWMs) for a transcription factor are available. Grouping multiple PWMs of a transcription factor (TF) based on their sequence similarity improves the specificity of TFBS prediction, which was evaluated using multiple genome-wide ChIP-Seq data sets from 26 TFs. The Z-scores of the area under a receiver operating characteristic curve (AUC) values of 368 TFs were calculated and used to statistically identify co-occurring regulatory motifs in the TF bound ChIP loci. Motifs that are co-occurring along with the empirical bindings of E2F, JUN or MYC have been evaluated, in the basal or stimulated condition. Results prove our method can be useful to systematically identify the co-occurring motifs of the TF for the given conditions.

Tongue-rudder: a glossokinetic-potential-based tongue-machine interface.

Glossokinetic potentials (GKPs) are electric potential responses generated by tongue movement. In this study, we use these GKPs to automatically detect and estimate tongue positions, and develop a tongue-machine interface. We show that a specific configuration of electrode placement yields discriminative GKPs that vary depending on the direction of the tongue. We develop a linear model to determine the direction of tongue from GKPs, where we seek linear features that are robust to a baseline drift problem by maximizing the ratio of intertask covariance to intersession covariance. We apply our method to the task of wheelchair control, developing a tongue-machine interface for wheelchair control, referred to as tongue-rudder. A teeth clenching detection system, using electromyography, was also implemented in the system in order to assign teeth clenching as the stop command. Experiments on off-line cursor control and online wheelchair control confirm the unique advantages of our method, such as: 1) noninvasiveness, 2) fine controllability, and 3) ability to integrate with other EEG-based interface systems.

Probabilistic models for semisupervised discriminative motif discovery in DNA sequences.

Methods for discriminative motif discovery in DNA sequences identify transcription factor binding sites (TFBSs), searching only for patterns that differentiate two sets (positive and negative sets) of sequences. On one hand, discriminative methods increase the sensitivity and specificity of motif discovery, compared to generative models. On the other hand, generative models can easily exploit unlabeled sequences to better detect functional motifs when labeled training samples are limited. In this paper, we develop a hybrid generative/discriminative model which enables us to make use of unlabeled sequences in the framework of discriminative motif discovery, leading to semisupervised discriminative motif discovery. Numerical experiments on yeast ChIP-chip data for discovering DNA motifs demonstrate that the best performance is obtained between the purely-generative and the purely-discriminative and the semisupervised learning improves the performance when labeled sequences are limited.