A Deep Learning-Based Chinese Semantic Parser for the Almond Virtual Assistant.

Almond is an extendible open-source virtual assistant designed to help people access Internet services and IoT (Internet of Things) devices. Both are referred to as skills here. Service providers can easily enable their devices for Almond by defining proper APIs (Application Programming Interfaces) for ThingTalk in Thingpedia. ThingTalk is a virtual assistant programming language, and Thingpedia is an application encyclopedia. Almond uses a large neural network to translate user commands in natural language into ThingTalk programs. To obtain enough data for the training of the neural network, Genie was developed to synthesize pairs of user commands and corresponding ThingTalk programs based on a natural language template approach. In this work, we extended Genie to support Chinese. For 107 devices and 261 functions registered in Thingpedia, 649 Chinese primitive templates and 292 Chinese construct templates were analyzed and developed. Two models, seq2seq (sequence-to-sequence) and MQAN (multiple question answer network), were trained to translate user commands in Chinese into ThingTalk programs. Both models were evaluated, and the experiment results showed that MQAN outperformed seq2seq. The exact match, BLEU, and F1 token accuracy of MQAN were 0.7, 0.82, and 0.88, respectively. As a result, users could use Chinese in Almond to access Internet services and IoT devices registered in Thingpedia.

Systemic lupus erythematosus is associated with poor outcome after acute myocardial infarction.

BACKGROUND: Systemic lupus erythematosus (SLE) is associated with a higher risk of cardiovascular disease. However, it is not clear whether or not SLE is associated with poor outcomes after acute myocardial infarction (AMI). METHODS AND RESULTS: Using the Taiwan National Health Insurance Database, we identified the SLE group as patients with AMI who have a concurrent discharge diagnosis of SLE. We also selected an age-, sex-, hospital level-, and admission calendar year-matched non-SLE group at a ratio of 1:3 from the total non-SLE group. One hundred fifty-one patients with SLE, 113,791 patients without SLE, and 453 matched patients without SLE were admitted with a diagnosis of AMI. Patients with SLE were significantly younger, predominantly female, and more likely to have chronic kidney disease than those without SLE. The in-hospital mortality rates were 12.6%, 9.0%, and 4.2% in the SLE, total non-SLE, and matched non-SLE groups, respectively. The in-hospital mortality was significantly higher in the SLE group than in the total non-SLE group (OR = 1.98; 95% CI = 1.2-3.26) and the matched non-SLE group (mortality OR = 2.20; 95% CI = 1.06-4.58). In addition, the SLE group was associated with a borderline significant risk of prolonged hospitalization when compared with the non-SLE group. CONCLUSION: SLE is associated with a higher risk of in-hospital mortality and a borderline significantly higher risk of prolonged hospitalization after AMI.

A high performance cloud computing platform for mRNA analysis.

Multiclass classification is an important technique to many complex bioinformatics problems. However, their performance is limited by the computation power. Based on the Apache Hadoop design framework, this study proposes a two layer architecture that exploits the inherent parallelism of GA-SVM classification to speed up the work. The performance evaluations on an mRNA benchmark cancer dataset have reduced 86.55% features and raised accuracy from 97.53% to 98.03%. With a user-friendly web interface, the system provides researchers an easy way to investigate the unrevealed secrets in the fast-growing repository of bioinformatics data.

Epilepsy analytic system with cloud computing.

Biomedical data analytic system has played an important role in doing the clinical diagnosis for several decades. Today, it is an emerging research area of analyzing these big data to make decision support for physicians. This paper presents a parallelized web-based tool with cloud computing service architecture to analyze the epilepsy. There are many modern analytic functions which are wavelet transform, genetic algorithm (GA), and support vector machine (SVM) cascaded in the system. To demonstrate the effectiveness of the system, it has been verified by two kinds of electroencephalography (EEG) data, which are short term EEG and long term EEG. The results reveal that our approach achieves the total classification accuracy higher than 90%. In addition, the entire training time accelerate about 4.66 times and prediction time is also meet requirements in real time.

Classification of schizophrenia using Genetic Algorithm-Support Vector Machine (GA-SVM).

Recently, Event-Related Potential (ERP) has being the most popular method in evaluating brain waves of schizophrenia patients. ERP is one of the electroencephalography (EEG), which is measured the change of brain waves after giving patients certain stimulations instead of resting state. However, with traditional statistical analysis method, both P50 and MMN showed significant difference between controls and patients but not in Gamma band. Gamma band is a 30-50 Hz auditory stimulation which had been suggested may be abnormal in schizophrenia patients. Our data are recruited from 5 schizophrenia patients and 5 controls in National Taiwan University Hospital have been tested with this platform. The results showed that detection rate is 88.24% and we also analyzed the importance of features, including Standard Deviation (SD) and Total Variation (TotalVar) in different stage of wavelet transform. Therefore, this proposed methodology could serve as a valuable clinical decision support for physiologists in evaluating schizophrenia.

A physiology-based seizure detection system for multichannel EEG.

BACKGROUND: Epilepsy is a common chronic neurological disorder characterized by recurrent unprovoked seizures. Electroencephalogram (EEG) signals play a critical role in the diagnosis of epilepsy. Multichannel EEGs contain more information than do single-channel EEGs. Automatic detection algorithms for spikes or seizures have traditionally been implemented on single-channel EEG, and algorithms for multichannel EEG are unavailable. METHODOLOGY: This study proposes a physiology-based detection system for epileptic seizures that uses multichannel EEG signals. The proposed technique was tested on two EEG data sets acquired from 18 patients. Both unipolar and bipolar EEG signals were analyzed. We employed sample entropy (SampEn), statistical values, and concepts used in clinical neurophysiology (e.g., phase reversals and potential fields of a bipolar EEG) to extract the features. We further tested the performance of a genetic algorithm cascaded with a support vector machine and post-classification spike matching. PRINCIPAL FINDINGS: We obtained 86.69% spike detection and 99.77% seizure detection for Data Set I. The detection system was further validated using the model trained by Data Set I on Data Set II. The system again showed high performance, with 91.18% detection of spikes and 99.22% seizure detection. CONCLUSION: We report a de novo EEG classification system for seizure and spike detection on multichannel EEG that includes physiology-based knowledge to enhance the performance of this type of system.

Measurement of natural background radiation intensity on a train.

This work aims to measure different components of natural background radiation on a train. A radiation measurement system consisting of four types of radiation detectors, namely, a Berkeley Lab cosmic-ray detector, moderated (3)He detector, high-pressure ionisation chamber and NaI(Tl) spectrometer, associated with a global positioning system unit was established for this purpose. For the commissioning of the system, a test measurement on a train along the railway around the northern Taiwan coast from Hsinchu to Hualien with a distance of ∼ 275 km was carried out. No significant variation of the intensities of the different components of natural background radiation was observed, except when the train went underground or in the tunnels. The average external dose rate received by the crew of the train was estimated to be 62 nSv h(-1).

Web-based healthcare hand drawing management system.

The paper addresses Medical Hand Drawing Management System architecture and implementation. In the system, we developed four modules: hand drawing management module; patient medical records query module; hand drawing editing and upload module; hand drawing query module. The system adapts windows-based applications and encompasses web pages by ASP.NET hosting mechanism under web services platforms. The hand drawings implemented as files are stored in a FTP server. The file names with associated data, e.g. patient identification, drawing physician, access rights, etc. are reposited in a database. The modules can be conveniently embedded, integrated into any system. Therefore, the system possesses the hand drawing features to support daily medical operations, effectively improve healthcare qualities as well. Moreover, the system includes the printing capability to achieve a complete, computerized medical document process. In summary, the system allows web-based applications to facilitate the graphic processes for healthcare operations.

Bio-signal analysis system design with support vector machines based on cloud computing service architecture.

Today, many bio-signals such as Electroencephalography (EEG) are recorded in digital format. It is an emerging research area of analyzing these digital bio-signals to extract useful health information in biomedical engineering. In this paper, a bio-signal analyzing cloud computing architecture, called BACCA, is proposed. The system has been designed with the purpose of seamless integration into the National Taiwan University Health Information System. Based on the concept of. NET Service Oriented Architecture, the system integrates heterogeneous platforms, protocols, as well as applications. In this system, we add modern analytic functions such as approximated entropy and adaptive support vector machine (SVM). It is shown that the overall accuracy of EEG bio-signal analysis has increased to nearly 98% for different data sets, including open-source and clinical data sets.