Noriaki Horii

A New Modality Using Breath Sound Analysis to Evaluate the Control Level of Asthma

BACKGROUND Reliable symptom assessment is essential in asthma management. We developed new technology for analyzing breath sounds and assessed its clinical usefulness for monitoring asthmatic children. METHODS Eighty asthmatic children and 59 non-asthmatic children underwent breath sound analysis in an asymptomatic state. Their asthma control was assessed by the Asthma Control TestTM or Childhood ACTTM scores and divided into two groups, namely, well-controlled (perfect) (n = 19) and not well-controlled (not perfect) (n = 61). Breath sounds were recorded using two sensors, located on the right anterior chest and trachea. We calculated the acoustic transfer characteristics between the two points, which indicated the relationship between frequencies and attenuation during breath sound propagation. Two indices of sound parameters, the chest wall sound index (CWI) and the tracheal sound index (TRI), were calculated from the transfer characteristics and tracheal sounds. We also developed a new parameter, the breath sound index (BSI), on a 2-dimensional diagram of CWI and TRI and tried to determine whether BSI may clarify asthma control better than CWI or TRI alone. RESULTS There was a significant difference in TRI and BSI between asthmatic and non-asthmatic children (p = 0.007, p < 0.001). There was a significant difference in CWI and TRI between the well-controlled and not-wellcontrolled groups (p < 0.001). BSI discriminated between the two groups accurately (p < 0.001). The sensitivity and specificity of BSI for asthma control were 83.6% and 84.2%, respectively. CONCLUSIONS Asthma control could be evaluated using a new index calculated from breath sound analysis.BACKGROUND Reliable symptom assessment is essential in asthma management. We developed new technology for analyzing breath sounds and assessed its clinical usefulness for monitoring asthmatic children. METHODS Eighty asthmatic children and 59 non-asthmatic children underwent breath sound analysis in an asymptomatic state. Their asthma control was assessed by the Asthma Control TestTM or Childhood ACTTM scores and divided into two groups, namely, well-controlled (perfect) (n = 19) and not well-controlled (not perfect) (n = 61). Breath sounds were recorded using two sensors, located on the right anterior chest and trachea. We calculated the acoustic transfer characteristics between the two points, which indicated the relationship between frequencies and attenuation during breath sound propagation. Two indices of sound parameters, the chest wall sound index (CWI) and the tracheal sound index (TRI), were calculated from the transfer characteristics and tracheal sounds. We also developed a new parameter, the breath sound index (BSI), on a 2-dimensional diagram of CWI and TRI and tried to determine whether BSI may clarify asthma control better than CWI or TRI alone. RESULTS There was a significant difference in TRI and BSI between asthmatic and non-asthmatic children (p = 0.007, p < 0.001). There was a significant difference in CWI and TRI between the well-controlled and not-well-controlled groups (p < 0.001). BSI discriminated between the two groups accurately (p < 0.001). The sensitivity and specificity of BSI for asthma control were 83.6% and 84.2%, respectively. CONCLUSIONS Asthma control could be evaluated using a new index calculated from breath sound analysis.

Convolutional Neural Networks for Multi-topic Dialog State Tracking

The main task of the fourth Dialog State Tracking Challenge (DSTC4) is to track the dialog state by filling in various slots, each of which represents a major subject discussed in the dialog. In this article we focus on the ‘INFO’ slot that tracks the general information provided in a sub-dialog segment, and propose an approach for this slot-filling using convolutional neural networks (CNNs). Our CNN model is adapted to multi-topic dialog by including a convolutional layer with general and topic-specific filters. The evaluation on DSTC4 common test data shows that our approach outperforms all other submitted entries in terms of overall accuracy of the ‘INFO’ slot.

A multichannel convolutional neural network for cross-language dialog state tracking

The fifth Dialog State Tracking Challenge (DSTC5) introduces a new cross-language dialog state tracking scenario, where the participants are asked to build their trackers based on the English training corpus, while evaluating them with the unlabeled Chinese corpus. Although the computer-generated translations for both English and Chinese corpus are provided in the dataset, these translations contain errors and careless use of them can easily hurt the performance of the built trackers. To address this problem, we propose a multichannel Convolutional Neural Networks (CNN) architecture, in which we treat English and Chinese language as different input channels of one single CNN model. In the evaluation of DSTC5, we found that such multichannel architecture can effectively improve the robustness against translation errors. Additionally, our method for DSTC5 is purely machine learning based and requires no prior knowledge about the target language. We consider this a desirable property for building a tracker in the cross-language context, as not every developer will be familiar with both languages.

Treatment evaluation using lung sound analysis in asthmatic children

Non‐invasive assessment of treatment and prediction of attacks in asthmatic children do not yet exist. Lung sound analysis can non‐invasively evaluate airway obstruction. We used a recently developed technology for analysing lung sounds using ic700 (index of the chest wall at 700 Hz, sound intensity at 700 Hz) to evaluate response to inhaled corticosteroid (ICS) in asthmatic children.

Recurrent convolutional neural networks for structured speech act tagging

Spoken language understanding (SLU) is one of the important problem in natural language processing, and especially in dialog system. Fifth Dialog State Tracking Challenge (DSTC5) introduced a SLU challenge task, which is automatic tagging to speech utterances by two speaker roles with speech acts tag and semantic slots tag. In this paper, we focus on speech acts tagging. We propose local coactivate multi-task learning model for capturing structured speech acts, based on sentence features by recurrent convolutional neural networks. An experiment result, shows that our model outperformed all other submitted entries, and were able to capture coactivated local features of category and attribute, which are parts of speech act.