ABSTRACT
TOPIC: Chest Infections TYPE: Original Investigations PURPOSE: COVID-19 is a major public health emergency with increasing mortality since the first case in December 2019. COVID-19 has multifaceted presentation and only about 15-20% required hospitalizations and a quarter of those required management in an intensive care unit (ICU). Herein, we reviewed the demographics, clinical outcomes, and identified the potential prognostic indicators in patients with COVID-19 who were admitted to our ICU. METHODS: A retrospective cohort study was conducted on patients ≥ 18 years old with confirmed COVID-19, who were admitted to our ICU between 03/17/2020 and 05/14/2020. Demographic, clinical, and laboratory data were reviewed and retrieved. Data was expressed as counts and percentages. T-test and chi-square (χ) were used for continuous and categorical variables respectively. Univariate analysis was also performed to identify the prognostic indicators associated with mortality. GraphPad PRISM was used for data analysis. RESULTS: Seventy-five patients were identified during the study period with an average age of 60.8 (ranged from 19-89), of which males were 99 (71%) and females were 41 (29%). Majority were Hispanic (45%) and African Americans (34%). The average body mass index was 30.6. Hypertension (47.9%) and chronic kidney disease (47.1%) were the most common comorbidities. Mechanical ventilation was required in 99 (70.7%) of the patients. Of those 99, 75 (76%) expired. There was no statistically significant difference between the survival and expired groups in term of age, gender, ethnicities and number of comorbidities. Need of mechanical ventilation (MV) and renal replacement therapy (RRT) were significantly associated with mortality (p<0.0001 for both parameters). Interestingly, use of therapeutic anticoagulation was associated with decrease risk of mortality (p-value 0.03865). When looking into the laboratory parameters, higher blood urea nitrogen (BUN) on Day 5 (p-value 0.0067), initial LDH (p<0.0001), initial CRP (p-value 0.0062), white blood cell counts (WBC) (p<0.0001), and high absolute neutrophil counts (ANC) (p-value 0.0002) were statistically associated with increased risk of mortality. SOFA scores did not predict mortality in these patients (p-value 0.9243) CONCLUSIONS: This retrospective cohort study on patients with COVID-19 who required ICU showed that need for Mechanical Ventilation and Renal Replacement Therapy were significantly associated with mortality. Surprisingly, use of therapeutic anticoagulation decreased the risk of mortality and initial LDH and CRP were the two inflammatory markers that may help predict mortality. This interesting finding need to be corroborated in a larger study. CLINICAL IMPLICATIONS: This study was done using the data during the initial two months of first wave of COVID-19 in the United States and the data represents our observations at that time. Several studies have been done on the use of Anticoagulation in COVID-19 patients thereafter, however there is no universally accepted protocol. This study helped us identify the differences in the outcome between first and second use with aggressive use of anticoagulation, Remdesivir, and dexamethasone during the second wave. DISCLOSURES: No relevant relationships by Sharath Bellary, source=Web Response No relevant relationships by Kok Hoe Chan, source=Web Response No relevant relationships by Joanna Crincoli, source=Web Response No relevant relationships by Claudia Komer, source=Web Response No relevant relationships by Sudha Lagudu, source=Web Response No relevant relationships by Richard Miller, source=Web Response No relevant relationships by Meenakshi Sindhuri Nali, source=Web Response No relevant relationships by Amy Paige, source=Web Response No relevant relationships by Rutwik Patel, source=Web Response No relevant relationships by Laxminarayan Prabhakar, source=Web Response No relevant relationships by Amr Ramahi, source=Web Response No relevant relationships by Divya Mounisha Thimmareddygari, source=Web Response
ABSTRACT
With over 127 million cases globally, the COVID-19 pandemic marks a sentinel event in global health. However, true case estimations have been elusive due to lack of testing and diagnostic capacity, asymptomatic cases, and individuals who do not get tested or seek care. Concomitantly, new digital surveillance tools to detect, characterize, and report COVID-19 cases are emerging, including using structured and unstructured data from users self-reporting COVID-19-related experiences on the Internet and social media platforms. In this study, we develop and evaluate a hybrid unsupervised and supervised machine learning approach to detect self-reported COVID-19-related symptoms on Twitter during the early stages of the pandemic. Tweets were collected from the public API stream from March 3rd-31st 2020, filtered for COVID-19-related terms. We used the biterm topic model to cluster tweets into theme-associated groups for the first 18 days of tweets, which were then extracted and manually annotated to identify users self-reporting suspected COVID-19 symptoms or status. Using this manually annotated data as a training set, we used an XLNet deep learning model for classifying symptom-related tweets from a larger corpus and also evaluated model performance. From 4, 492, 954 tweets collected, the unsupervised learning process yielded 3, 465 (<1%) symptom tweets used to form our ground-truth COVID-19 symptoms dataset (n = 11, 550). The XLNet text classifier achieved the highest accuracy (.91) and f1 (.62) compared to baseline models evaluated for classification. After re-training with adjusted loss function, we boosted the classifier's precision to 0.81 while maintaining a high f1 (0.66), resulting in identification of an additional 2, 622 symptom-related tweets when applied to an additional 11 days of tweets collected. Our study used a hybrid machine learning approach to enable high precision identification of Twitter user-generated COVID-19 symptom discussions. The model is a digital epidemiology tool that can identify social media users who self-report symptoms during the early periods of an outbreak. © 2021 IEEE.