Machine-Learning-Based COVID-19 and Dyspnoea Prediction Systems for the Emergency Department

The COVID-19 pandemic highlighted an urgent need for reliable diagnostic tools to minimize viral spreading. It is mandatory to avoid cross-contamination between patients and detect COVID-19 positive individuals to cluster people by prognosis and manage the emergency department's resources. Fondazione IRCCS Policlinico San Matteo Hospital's Emergency Department (ED) of Pavia let us evaluate the exploitation of machine learning algorithms on a clinical dataset gathered from laboratory-confirmed rRT-PCR test patients, collected from 1 March to 30 June 2020. Physicians examined routine blood tests, clinical history, symptoms, arterial blood gas (ABG) analysis, and lung ultrasound quantitative examination. We developed two diagnostic tools for COVID-19 detection and oxygen therapy prediction, namely, the need for ventilation support due to lung involvement. We obtained promising classification results with F1 score levels meeting 92%, and we also engineered a user-friendly interface for healthcare providers during daily screening operations. This research proved machine learning models as a potential screening methodology during contingency times.

Deep learning and lung ultrasound for Covid-19 pneumonia detection and severity classification.

The Covid-19 European outbreak in February 2020 has challenged the world's health systems, eliciting an urgent need for effective and highly reliable diagnostic instruments to help medical personnel. Deep learning (DL) has been demonstrated to be useful for diagnosis using both computed tomography (CT) scans and chest X-rays (CXR), whereby the former typically yields more accurate results. However, the pivoting function of a CT scan during the pandemic presents several drawbacks, including high cost and cross-contamination problems. Radiation-free lung ultrasound (LUS) imaging, which requires high expertise and is thus being underutilised, has demonstrated a strong correlation with CT scan results and a high reliability in pneumonia detection even in the early stages. In this study, we developed a system based on modern DL methodologies in close collaboration with Fondazione IRCCS Policlinico San Matteo's Emergency Department (ED) of Pavia. Using a reliable dataset comprising ultrasound clips originating from linear and convex probes in 2908 frames from 450 hospitalised patients, we conducted an investigation into detecting Covid-19 patterns and ranking them considering two severity scales. This study differs from other research projects by its novel approach involving four and seven classes. Patients admitted to the ED underwent 12 LUS examinations in different chest parts, each evaluated according to standardised severity scales. We adopted residual convolutional neural networks (CNNs), transfer learning, and data augmentation techniques. Hence, employing methodological hyperparameter tuning, we produced state-of-the-art results meeting F1 score levels, averaged over the number of classes considered, exceeding 98%, and thereby manifesting stable measurements over precision and recall.

Can Alveolar-Arterial Difference and Lung Ultrasound Help the Clinical Decision Making in Patients with COVID-19?

BACKGROUND: COVID-19 is an emerging infectious disease, that is heavily challenging health systems worldwide. Admission Arterial Blood Gas (ABG) and Lung Ultrasound (LUS) can be of great help in clinical decision making, especially during the current pandemic and the consequent overcrowding of the Emergency Department (ED). The aim of the study was to demonstrate the capability of alveolar-to-arterial oxygen difference (AaDO2) in predicting the need for subsequent oxygen support and survival in patients with COVID-19 infection, especially in the presence of baseline normal PaO2/FiO2 ratio (P/F) values. METHODS: A cohort of 223 swab-confirmed COVID-19 patients underwent clinical evaluation, blood tests, ABG and LUS in the ED. LUS score was derived from 12 ultrasound lung windows. AaDO2 was derived as AaDO2 = ((FiO2) (Atmospheric pressure - H2O pressure) - (PaCO2/R)) - PaO2. Endpoints were subsequent oxygen support need and survival. RESULTS: A close relationship between AaDO2 and P/F and between AaDO2 and LUS score was observed (R2 = 0.88 and R2 = 0.67, respectively; p < 0.001 for both). In the subgroup of patients with P/F between 300 and 400, 94.7% (n = 107) had high AaDO2 values, and 51.4% (n = 55) received oxygen support, with 2 ICU admissions and 10 deaths. According to ROC analysis, AaDO2 > 39.4 had 83.6% sensitivity and 90.5% specificity (AUC 0.936; p < 0.001) in predicting subsequent oxygen support, whereas a LUS score > 6 showed 89.7% sensitivity and 75.0% specificity (AUC 0.896; p < 0.001). Kaplan-Meier curves showed different mortality in the AaDO2 subgroups (p = 0.0025). CONCLUSIONS: LUS and AaDO2 are easy and effective tools, which allow bedside risk stratification in patients with COVID-19, especially when P/F values, signs, and symptoms are not indicative of severe lung dysfunction.

Lung ultrasound for the early diagnosis of COVID-19 pneumonia: an international multicenter study.

PURPOSE: To analyze the application of a lung ultrasound (LUS)-based diagnostic approach to patients suspected of COVID-19, combining the LUS likelihood of COVID-19 pneumonia with patient's symptoms and clinical history. METHODS: This is an international multicenter observational study in 20 US and European hospitals. Patients suspected of COVID-19 were tested with reverse transcription-polymerase chain reaction (RT-PCR) swab test and had an LUS examination. We identified three clinical phenotypes based on pre-existing chronic diseases (mixed phenotype), and on the presence (severe phenotype) or absence (mild phenotype) of signs and/or symptoms of respiratory failure at presentation. We defined the LUS likelihood of COVID-19 pneumonia according to four different patterns: high (HighLUS), intermediate (IntLUS), alternative (AltLUS), and low (LowLUS) probability. The combination of patterns and phenotypes with RT-PCR results was described and analyzed. RESULTS: We studied 1462 patients, classified in mild (n = 400), severe (n = 727), and mixed (n = 335) phenotypes. HighLUS and IntLUS showed an overall sensitivity of 90.2% (95% CI 88.23-91.97%) in identifying patients with positive RT-PCR, with higher values in the mixed (94.7%) and severe phenotype (97.1%), and even higher in those patients with objective respiratory failure (99.3%). The HighLUS showed a specificity of 88.8% (CI 85.55-91.65%) that was higher in the mild phenotype (94.4%; CI 90.0-97.0%). At multivariate analysis, the HighLUS was a strong independent predictor of RT-PCR positivity (odds ratio 4.2, confidence interval 2.6-6.7, p < 0.0001). CONCLUSION: Combining LUS patterns of probability with clinical phenotypes at presentation can rapidly identify those patients with or without COVID-19 pneumonia at bedside. This approach could support and expedite patients' management during a pandemic surge.

Lung ultrasound presentation of COVID-19 patients: phenotypes and correlations.

Bedside lung ultrasound (LUS) can play a role in the setting of the SarsCoV2 pneumonia pandemic. To evaluate the clinical and LUS features of COVID-19 in the ED and their potential prognostic role, a cohort of laboratory-confirmed COVID-19 patients underwent LUS upon admission in the ED. LUS score was derived from 12 fields. A prevalent LUS pattern was assigned depending on the presence of interstitial syndrome only (Interstitial Pattern), or evidence of subpleural consolidations in at least two fields (Consolidation Pattern). The endpoint was 30-day mortality. The relationship between hemogasanalysis parameters and LUS score was also evaluated. Out of 312 patients, only 36 (11.5%) did not present lung involvment, as defined by LUS score < 1. The majority of patients were admitted either in a general ward (53.8%) or in intensive care unit (9.6%), whereas 106 patients (33.9%) were discharged from the ED. In-hospital mortality was 25.3%, and 30-day survival was 67.6%. A LUS score > 13 had a 77.2% sensitivity and a 71.5% specificity (AUC 0.814; p < 0.001) in predicting mortality. LUS alterations were more frequent (64%) in the posterior lower fields. LUS score was related with P/F (R2 0.68; p < 0.0001) and P/F at FiO2 = 21% (R2 0.59; p < 0.0001). The correlation between LUS score and P/F was not influenced by the prevalent ultrasound pattern. LUS represents an effective tool in both defining diagnosis and stratifying prognosis of COVID-19 pneumonia. The correlation between LUS and hemogasanalysis parameters underscores its role in evaluating lung structure and function.

Lung Ultrasound in Covid-19: A useful diagnostic tool

Coronavirus disease 2019 (Covid-19), caused by a novel enveloped RNA betacoronavirus, has recently been declared a public health emergency by the World Health Organization (WHO). The lack of knowledge at the beginning of the pandemics, associated with the inherent risk of infective spreading, makes initial recognition and management particularly complex, in terms of defining effective diagnostic and therapeutic protocols. In the Emergency setting, Lung Ultrasound (LUS) can play an important role in the management of patients with SARS-CoV2-related pneumonia, expanding from the initial diagnosis to the subsequent monitoring and follow-up. Among many other potential advantages (such has the absence of ionizing radiation, its inherent cost-effectiveness, and bedside repeatability), LUS provides immediate diagnostic response and might prevent the risk of spreading the infection by moving the patient from the Emergency Room to the Radiology facilities. Aim of this short review is to define the potential role of lung ultrasound in Covid-19 patients, according to the evidence in the medical literature.

Emergency Department and Out-of-Hospital Emergency System (112-AREU 118) integrated response to Coronavirus Disease 2019 in a Northern Italy centre.

Since December 2019, the world has been facing the life-threatening disease, named Coronavirus disease-19 (COVID-19), recognized as a pandemic by the World Health Organization. The response of the Emergency Medicine network, integrating "out-of-hospital" and "hospital" activation, is crucial whenever the health system has to face a medical emergency, being caused by natural or human-derived disasters as well as by a rapidly spreading epidemic outbreak. We here report the Pavia Emergency Medicine network response to the COVID-19 outbreak. The "out-of-hospital" response was analysed in terms of calls, rescues and missions, whereas the "hospital" response was detailed as number of admitted patients and subsequent hospitalisation or discharge. The data in the first 5 weeks of the Covid-19 outbreak (February 21-March 26, 2020) were compared with a reference time window referring to the previous 5 weeks (January 17-February 20, 2020) and with the corresponding historical average data from the previous 5 years (February 21-March 26). Since February 21, 2020, a sudden and sustained increase in the calls to the AREU 112 system was noted (+ 440%). After 5 weeks, the number of calls and missions was still higher as compared to both the reference pre-Covid-19 period (+ 48% and + 10%, respectively) and the historical control (+ 53% and + 22%, respectively). Owing to the overflow from the neighbouring hospitals, which rapidly became overwhelmed and had to temporarily close patient access, the population served by the Pavia system more than doubled (from 547.251 to 1.135.977 inhabitants, + 108%). To minimize the possibility of intra-hospital spreading of the infection, a separate "Emergency Department-Infective Disease" was created, which evaluated 1241 patients with suspected infection (38% of total ED admissions). Out of these 1241 patients, 58.0% (n = 720) were admitted in general wards (n = 629) or intensive care unit (n = 91). To allow this massive number of admissions, the hospital reshaped many general ward Units, which became Covid-19 Units (up to 270 beds) and increased the intensive care unit beds from 32 to 60. In the setting of a long-standing continuing emergency like the present Covid-19 outbreak, the integration, interaction and team work of the "out-of-hospital" and "in-hospital" systems have a pivotal role. The present study reports how the rapid and coordinated reorganization of both might help in facing such a disaster. AREU-112 and the Emergency Department should be ready to finely tune their usual cooperation to respond to a sudden and overwhelming increase in the healthcare needs brought about by a pandemia like the current one. This lesson should shape and reinforce the future.