Deep learning diagnostic and risk-stratification pattern detection for COVID-19 in digital lung auscultations: clinical protocol for a case-control and prospective cohort study.

BACKGROUND: Lung auscultation is fundamental to the clinical diagnosis of respiratory disease. However, auscultation is a subjective practice and interpretations vary widely between users. The digitization of auscultation acquisition and interpretation is a particularly promising strategy for diagnosing and monitoring infectious diseases such as Coronavirus-19 disease (COVID-19) where automated analyses could help decentralise care and better inform decision-making in telemedicine. This protocol describes the standardised collection of lung auscultations in COVID-19 triage sites and a deep learning approach to diagnostic and prognostic modelling for future incorporation into an intelligent autonomous stethoscope benchmarked against human expert interpretation. METHODS: A total of 1000 consecutive, patients aged ≥ 16 years and meeting COVID-19 testing criteria will be recruited at screening sites and amongst inpatients of the internal medicine department at the Geneva University Hospitals, starting from October 2020. COVID-19 is diagnosed by RT-PCR on a nasopharyngeal swab and COVID-positive patients are followed up until outcome (i.e., discharge, hospitalisation, intubation and/or death). At inclusion, demographic and clinical data are collected, such as age, sex, medical history, and signs and symptoms of the current episode. Additionally, lung auscultation will be recorded with a digital stethoscope at 6 thoracic sites in each patient. A deep learning algorithm (DeepBreath) using a Convolutional Neural Network (CNN) and Support Vector Machine classifier will be trained on these audio recordings to derive an automated prediction of diagnostic (COVID positive vs negative) and risk stratification categories (mild to severe). The performance of this model will be compared to a human prediction baseline on a random subset of lung sounds, where blinded physicians are asked to classify the audios into the same categories. DISCUSSION: This approach has broad potential to standardise the evaluation of lung auscultation in COVID-19 at various levels of healthcare, especially in the context of decentralised triage and monitoring. TRIAL REGISTRATION: PB_2016-00500, SwissEthics. Registered on 6 April 2020.

Predictors of In-Hospital Mortality in Older Patients With COVID-19: The COVIDAge Study.

OBJECTIVE: To determine predictors of in-hospital mortality related to COVID-19 in older patients. DESIGN: Retrospective cohort study. SETTING AND PARTICIPANTS: Patients aged 65 years and older hospitalized for a diagnosis of COVID-19. METHODS: Data from hospital admission were collected from the electronic medical records. Logistic regression and Cox proportional hazard models were used to predict mortality, our primary outcome. Variables at hospital admission were categorized according to the following domains: demographics, clinical history, comorbidities, previous treatment, clinical status, vital signs, clinical scales and scores, routine laboratory analysis, and imaging results. RESULTS: Of a total of 235 Caucasian patients, 43% were male, with a mean age of 86 ± 6.5 years. Seventy-six patients (32%) died. Nonsurvivors had a shorter number of days from initial symptoms to hospitalization (P = .007) and the length of stay in acute wards than survivors (P < .001). Similarly, they had a higher prevalence of heart failure (P = .044), peripheral artery disease (P = .009), crackles at clinical status (P < .001), respiratory rate (P = .005), oxygen support needs (P < .001), C-reactive protein (P < .001), bilateral and peripheral infiltrates on chest radiographs (P = .001), and a lower prevalence of headache (P = .009). Furthermore, nonsurvivors were more often frail (P < .001), with worse functional status (P < .001), higher comorbidity burden (P < .001), and delirium at admission (P = .007). A multivariable Cox model showed that male sex (HR 4.00, 95% CI 2.08-7.71, P < .001), increased fraction of inspired oxygen (HR 1.06, 95% CI 1.03-1.09, P < .001), and crackles (HR 2.42, 95% CI 1.15-6.06, P = .019) were the best predictors of mortality, while better functional status was protective (HR 0.98, 95% CI 0.97-0.99, P = .001). CONCLUSIONS AND IMPLICATIONS: In older patients hospitalized for COVID-19, male sex, crackles, a higher fraction of inspired oxygen, and functionality were independent risk factors of mortality. These routine parameters, and not differences in age, should be used to evaluate prognosis in older patients.