Benchmarking of eight recurrent neural network variants for breath phase and adventitious sound detection on a self-developed open-access lung sound database—HF_Lung_V1

A reliable, remote, and continuous real-time respiratory sound monitor with automated respiratory sound analysis ability is urgently required in many clinical scenarios—such as in monitoring disease progression of coronavirus disease 2019—to replace conventional auscultation with a handheld stethoscope. However, a robust computerized respiratory sound analysis algorithm for breath phase detection and adventitious sound detection at the recording level has not yet been validated in practical applications. In this study, we developed a lung sound database (HF_Lung_V1) comprising 9,765 audio files of lung sounds (duration of 15 s each), 34,095 inhalation labels, 18,349 exhalation labels, 13,883 continuous adventitious sound (CAS) labels (comprising 8,457 wheeze labels, 686 stridor labels, and 4,740 rhonchus labels), and 15,606 discontinuous adventitious sound labels (all crackles). We conducted benchmark tests using long short-term memory (LSTM), gated recurrent unit (GRU), bidirectional LSTM (BiLSTM), bidirectional GRU (BiGRU), convolutional neural network (CNN)-LSTM, CNN-GRU, CNN-BiLSTM, and CNN-BiGRU models for breath phase detection and adventitious sound detection. We also conducted a performance comparison between the LSTM-based and GRU-based models, between unidirectional and bidirectional models, and between models with and without a CNN. The results revealed that these models exhibited adequate performance in lung sound analysis. The GRU-based models outperformed, in terms of F1 scores and areas under the receiver operating characteristic curves, the LSTM-based models in most of the defined tasks. Furthermore, all bidirectional models outperformed their unidirectional counterparts. Finally, the addition of a CNN improved the accuracy of lung sound analysis, especially in the CAS detection tasks.

Associated Factors and Utility of Bed-side Assessment Tools to Avoid Antibiotic Overuse in an Urban ICU of a Developing Country

Antibiotic exposure in the intensive care unit (ICU) is very high, although 50% of all antibiotics may be unnecessary. We aimed to determine predicting factors, outcomes, and the utility of simple screening tools to avoid antibiotic overuse in the ICU. We analyzed 510 young children who did not receive antibiotics during ICU stay to those treated with antibiotics. The logistic regression analysis revealed that cases were more often older and independently associated with hypernatremia. Cases less often had severe underweight, altered mentation, age-specific fast breathing, lower chest wall in-drawing, adventitious sound on lung auscultation, abdominal distension, developmental delay, hyponatremia, hypocalcemia, and microscopic evidence of invasive diarrhea (for all, p<0.05). The case-fatality rate was significantly higher among the cases than the controls. For predicting 'no antibiotic approach,' the sensitivity of a negative quick Sequential Organ Failure Assessment (qSOFA) was similar to quick Pediatric Logistic Organ Dysfunction-2 (qPELOD-2) and higher than Systemic Inflammatory Response Syndrome (SIRS). In conclusion, 'No antibiotic approach' could be safely adopted in PICU using some simple clinical and laboratory characteristics, particularly in poor resource settings. A negative qSOFA or qPELOD-2 score calculated during PICU admission is superior to SIRS to avoid antibiotic overuse in under-five children.

Neonatal pneumonia in a rural primary care hospital in Bangladesh: prevalence, validation of clinical features and their outcome

Objective: To explore prevalence, validity of associated factors and their outcome of pneumonia in neonates.Methods: We retrospectively enrolled neonates admitted to a rural hospital in Bangladesh from January 2012-December 2014. Those with pneumonia constituted cases (n=142) and randomly selected three folds of cases from those without pneumonia formed the controls (n=426). Pneumonia was diagnosed by hospital physicians based on respiratory difficulty and/or abnormal auscultatory findings in lungs.Result: The deaths were significantly higher among the cases than the controls (p=0.025). In logistic regression analysis, fast breathing, lower chest wall in-drawing, adventitious sound and cough were independently associated with neonatal pneumonia (for all, p<0.01). However, best of all, sensitivity of fast breathing and lower chest wall in-drawing was 94% and 76% and specificity 81% and 82% respectively.Conclusion: The results underscore the importance of adherence to WHO defined clinical signs in diagnosing pneumonia in neonates especially in resource limited settings.Bangladesh Crit Care J September 2016; 4 (2): 74-78

Case Report: Hemoptysis localization – hearing with your eyes

Clinical Vignette: 45 year old was transferred from a peripheral facility for acute massive hemoptysis though maintained sufficient airway patency with no evidence of hemodynamic instability or respiratory failure. Thoracic auscultation revealed vesicular breathing with no adventitious sound. CXR from the peripheral site was normal.