This paper presents a novel diagnosis cloud framework that enables the extraction and transfer of knowledge from one network to another. It also presents use cases and requirements. We present the implementation details of the diagnosis cloud framework for two specific types of models: topic models and Markov Logic Networks (MLNs).
Conference Paper
Mitigating the effects of non-stationary unseen noises on language recognition performance
We introduce a new dataset for the study of the effect of highly non-stationary noises on language recognition (LR) performance.
Development of an Agonist of the TGF-Beta Signaling Pathway to Treat Alzheimer’s Disease
Alzheimer’s disease (AD) is a neurodegenerative disorder that leads to progressive cognitive dysfunction. Current knowledge of the processes leading to AD is still limited, and no effective treatments are available. Because neurodegeneration is associated with injury and activation of innate immune responses in the brain, drugs that could mimic the beneficial aspects of this response are potential therapeutic candidates. The cytokine transforming growth factor (TGF)-?1 is an organizer of the brain’s response to injury and has been shown to have neuroprotective effects in models of brain injury and degeneration. Recombinant TGF-?1 has been used to treat various forms of brain injury in vivo but delivery is not suitable for human use. Studies from our lab have demonstrated that TGF-?1 can reduce the overall accumulation of A?, a key factor in AD pathogenesis, in mouse models for AD and in cell culture. Numerous studies have also demonstrated that TGF-?1 is a potent neurotrophic factor, although high-level chronic TGF-?1 production can also be detrimental. Recently, we reported that reduced TGF-?1 expression in vivo or in cultured neurons increases neurodegeneration. Additional studies show that reducing TGF-? signaling in neurons of a mouse model for AD increases A? accumulation and neurodegeneration and that TGF-? receptor expression is reduced in human AD brains. We have identified bioactive small molecule chemical compounds that can activate the TGF-? signaling pathway in hippocampal neurons of mice and that pass the blood-brain barrier. With reporter cell lines for the TGF-? signaling pathway we screened a diverse small molecule drug library and identified several compounds that are able to activate the reporter system in vitro and in TGF-? reporter mice in vivo. The compounds induce specific TGF-?-responsive genes in cell culture consistent with Smad dependent activation of the TGF-? pathway. These chemicals share common properties from which we propose here to derive a lead compound within 5 years. This project includes structure activity relationship analysis of identified active compounds, medicinal chemistry, toxicology and pharmacology in a subcontract with SRI International. Compounds will be tested in neuroprotection and neurotoxicity assays in cell culture and in TGF-? reporter mice in vivo. The two most promising compounds will then be tested in an in vivo model of neurodegeneration and in a mouse model for AD. Part of the in vivo analysis on neurodegeneration will be done in collaboration with researchers at UCSD. At the end of our studies we propose to have for the first time a novel neuroprotective and amyloid reducing investigational new drug based on the TGF-? signaling pathway for testing in patients with AD.
Using Analytics for Improving Implementation Fidelity in a Large Scale Efficacy Trial
Highly Accurate Phonetic Segmentation Using Boundary Correction Models and System Fusion
Accurate phone-level segmentation of speech remains an important task for many subfields of speech research. We investigate techniques for boosting the accuracy of automatic phonetic segmentation based on HMM acoustic-phonetic models. In prior work we were able to improve on state-of-the-art alignment accuracy by employing special phone boundary HMM models, trained on phonetically segmented training data, in conjunction with a simple boundary-time correction model. Here we present further improved results by using more powerful statistical models for boundary correction that are conditioned on phonetic context and duration features. Furthermore, we find that combining multiple acoustic front-ends gives additional gains in accuracy, and that conditioning the combiner on phonetic context and side information helps. Overall, we reduce segmentation errors on the TIMIT corpus by almost one half, from 93.9% to 96.8% boundary accuracy with a 20-ms tolerance.
Mechanism of Action of Flufirvitide, a Peptide Inhibitor of Influenza Virus Infection
Influenza is an infectious disease typically transmitted through the air. It is responsible for seasonal epidemics affecting millions of people, and sporadic global pandemics. Influenza infection is a membrane fusion‐dependant process, occurring in the endosome of the host cell after viral binding and endocytosis. The virus‐host membrane fusion process is mediated by hemagglutinin (HA), a viral surface glycoprotein. Studies show that when the virus is subjected to low pH in the endosome, the HA protein partially unfolds and changes conformation, exposing the fusion initiation region (FIR). A 16 amino acid peptide sequence (Flufirvitide) derived from the fusion initiation region of the HA protein has shown effective inhibition of influenza virus infection. It is hypothesized that there is an interaction between the peptide and the FIR which inhibits fusion of the virus to the host cell. Plaque inhibition assays and animal studies show high efficacy of the peptide against the virus. We are currently developing biochemical and biophysical assays to study the interaction between Flufirvitide and HA. Circular Dichroism studies show that the peptide has a random coil conformation at pH 7 and higher. To elucidate the mechanism of fusion inhibition, the interaction between peptide and HA is being investigated with immunodetection, immunoprecipitation, and florescence techniques. Additionally, binding and interaction of the peptide with the intact virus is being studied by using Cryo‐electron microscopy.
Varenicline Markedly Decreases Antipsychotic-Induced Tardive Dyskinesia in a Rodent Model
Tardive dyskinesia is a potentially irreversible drug-induced movement disorder that arises as a side effect of antipsychotic therapy. Antipsychotics form the mainstay of treatment for schizophrenia and bipolar disorder and, in addition, are increasingly being prescribed for major depressive disorder, autism, attention deficit hyperactivity disorder, obsessive compulsive disorder and post-traumatic stress disorder. There is therefore a clear need for therapies to reduce tardive dyskinesia. Our recent studies showed that nicotine administration decreased haloperidol-induced vacuous chewing movements (VCMs) in a rat model of tardive dyskinesia. The present experiments demonstrate that nicotine (300 µg/ml in drinking water) also reduced VCMs (50%) in mice whether haloperidol was given via constant infusion (subcutaneous pellet) or daily injection. The nicotine-mediated decline is thus observed across species using various haloperidol treatment regimens. We then tested the effect of varenicline, an agonist that interacts with multiple nicotinic receptor subtypes. Low dose varenicline (0.1 mg/kg) decreased haloperidol-induced VCMs to a much greater extent (~90%) than nicotine (~50%). Since varenicline also acts at serotonergic receptors, we tested the effect of nicotine in combination with a selective serotonergic receptor drug 8-OH-DPAT (0.3 mg/kg). Nicotine or 8-OH-DPAT treatment alone decreased haloperidol-induced VCMs by ~60%, while combined administration reduced VCMs to a significantly greater extent than either drug alone (83%). These data are the first to show that drugs acting at both nicotinic and serotonergic receptors result in a pronounced decline in antipsychotic-induced VCMs. Drugs such as varenicline that act at both receptors may thus represent a novel therapy for reducing tardive dyskinesia.