COVID-19 Outcome Prediction and Monitoring Solution for Military Hospitals in South Korea: Development and Evaluation of an Application.

BACKGROUND: COVID-19 has officially been declared as a pandemic, and the spread of the virus is placing sustained demands on public health systems. There are speculations that the COVID-19 mortality differences between regions are due to the disparities in the availability of medical resources. Therefore, the selection of patients for diagnosis and treatment is essential in this situation. Military personnel are especially at risk for infectious diseases; thus, patient selection with an evidence-based prognostic model is critical for them. OBJECTIVE: This study aims to assess the usability of a novel platform used in the military hospitals in Korea to gather data and deploy patient selection solutions for COVID-19. METHODS: The platform's structure was developed to provide users with prediction results and to use the data to enhance the prediction models. Two applications were developed: a patient's application and a physician's application. The primary outcome was requiring an oxygen supplement. The outcome prediction model was developed with patients from four centers. A Cox proportional hazards model was developed. The outcome of the model for the patient's application was the length of time from the date of hospitalization to the date of the first oxygen supplement use. The demographic characteristics, past history, patient symptoms, social history, and body temperature were considered as risk factors. A usability study with the Post-Study System Usability Questionnaire (PSSUQ) was conducted on the physician's application on 50 physicians. RESULTS: The patient's application and physician's application were deployed on the web for wider availability. A total of 246 patients from four centers were used to develop the outcome prediction model. A small percentage (n=18, 7.32%) of the patients needed professional care. The variables included in the developed prediction model were age; body temperature; predisease physical status; history of cardiovascular disease; hypertension; visit to a region with an outbreak; and symptoms of chills, feverishness, dyspnea, and lethargy. The overall C statistic was 0.963 (95% CI 0.936-0.99), and the time-dependent area under the receiver operating characteristic curve ranged from 0.976 at day 3 to 0.979 at day 9. The usability of the physician's application was good, with an overall average of the responses to the PSSUQ being 2.2 (SD 1.1). CONCLUSIONS: The platform introduced in this study enables evidence-based patient selection in an effortless and timely manner, which is critical in the military. With a well-designed user experience and an accurate prediction model, this platform may help save lives and contain the spread of the novel virus, COVID-19.

A Patient Self-Checkup App for COVID-19: Development and Usage Pattern Analysis.

BACKGROUND: Clear guidelines for a patient with suspected COVID-19 infection are unavailable. Many countries rely on assessments through a national hotline or telecommunications, but this only adds to the burden of an already overwhelmed health care system. In this study, we developed an algorithm and a web application to help patients get screened. OBJECTIVE: This study aims to aid the general public by developing a web-based application that helps patients decide when to seek medical care during a novel disease outbreak. METHODS: The algorithm was developed via consultations with 6 physicians who directly screened, diagnosed, and/or treated patients with COVID-19. The algorithm mainly focused on when to test a patient in order to allocate limited resources more efficiently. The application was designed to be mobile-friendly and deployed on the web. We collected the application usage pattern data from March 1 to March 27, 2020. We evaluated the association between the usage pattern and the numbers of COVID-19 confirmed, screened, and mortality cases by access location and digital literacy by age group. RESULTS: The algorithm used epidemiological factors, presence of fever, and other symptoms. In total, 83,460 users accessed the application 105,508 times. Despite the lack of advertisement, almost half of the users accessed the application from outside of Korea. Even though the digital literacy of the 60+ years age group is half of that of individuals in their 50s, the number of users in both groups was similar for our application. CONCLUSIONS: We developed an expert-opinion-based algorithm and web-based application for screening patients. This innovation can be helpful in circumstances where information on a novel disease is insufficient and may facilitate efficient medical resource allocation.

An Easy-to-Use Machine Learning Model to Predict the Prognosis of Patients With COVID-19: Retrospective Cohort Study.

BACKGROUND: Prioritizing patients in need of intensive care is necessary to reduce the mortality rate during the COVID-19 pandemic. Although several scoring methods have been introduced, many require laboratory or radiographic findings that are not always easily available. OBJECTIVE: The purpose of this study was to develop a machine learning model that predicts the need for intensive care for patients with COVID-19 using easily obtainable characteristics-baseline demographics, comorbidities, and symptoms. METHODS: A retrospective study was performed using a nationwide cohort in South Korea. Patients admitted to 100 hospitals from January 25, 2020, to June 3, 2020, were included. Patient information was collected retrospectively by the attending physicians in each hospital and uploaded to an online case report form. Variables that could be easily provided were extracted. The variables were age, sex, smoking history, body temperature, comorbidities, activities of daily living, and symptoms. The primary outcome was the need for intensive care, defined as admission to the intensive care unit, use of extracorporeal life support, mechanical ventilation, vasopressors, or death within 30 days of hospitalization. Patients admitted until March 20, 2020, were included in the derivation group to develop prediction models using an automated machine learning technique. The models were externally validated in patients admitted after March 21, 2020. The machine learning model with the best discrimination performance was selected and compared against the CURB-65 (confusion, urea, respiratory rate, blood pressure, and 65 years of age or older) score using the area under the receiver operating characteristic curve (AUC). RESULTS: A total of 4787 patients were included in the analysis, of which 3294 were assigned to the derivation group and 1493 to the validation group. Among the 4787 patients, 460 (9.6%) patients needed intensive care. Of the 55 machine learning models developed, the XGBoost model revealed the highest discrimination performance. The AUC of the XGBoost model was 0.897 (95% CI 0.877-0.917) for the derivation group and 0.885 (95% CI 0.855-0.915) for the validation group. Both the AUCs were superior to those of CURB-65, which were 0.836 (95% CI 0.825-0.847) and 0.843 (95% CI 0.829-0.857), respectively. CONCLUSIONS: We developed a machine learning model comprising simple patient-provided characteristics, which can efficiently predict the need for intensive care among patients with COVID-19.

A case of early diagnosis of pulmonary capillary hemangiomatosis in a worker with exposure to silica.

BACKGROUND: Pulmonary capillary hemangiomatosis (PCH) is a progressive and refractory vascular disease in the lung. Pulmonary hypertension is frequently combined with PCH when capillary proliferation invades to nearby pulmonary vascular systems. It is difficult to differentiate PCH from other diseases such as pulmonary venoocclusive disease and pulmonary arterial hypertension that cause pulmonary hypertension as they frequently overlap. CASE PRESENTATION: A 29-year-old female who had worked at a bathtub factory presented with progressive exertional dyspnea for the past 2 years. Computed tomography revealed centrilobular, diffusely spreading ground-glass opacities sparing subpleural parenchyma with some cystic lesions and air-trapping in both lungs, suggesting a peculiar pattern of interstitial lung disease with airway involvement. There was not any evidence of right heart failure or pulmonary hypertension on echocardiogram, as well as radiography. Microscopic examination of the lung by thoracoscopic resection showed atypical proliferation of capillary channels within alveolar walls and interlobar septa, without invasion of large vessels. CONCLUSION: We experienced a pathologically diagnosed PCH in a young female complaining progressive dyspnea with prior exposure to occupational silica or organic solvent without elevated right ventricular systolic pressure (RVSP) who showed atypical pattern of radiologic findings.

Validation of Machine Learning Models to Predict Adverse Outcomes in Patients with COVID-19: A Prospective Pilot Study.

PURPOSE: We previously developed learning models for predicting the need for intensive care and oxygen among patients with coronavirus disease (COVID-19). Here, we aimed to prospectively validate the accuracy of these models. MATERIALS AND METHODS: Probabilities of the need for intensive care [intensive care unit (ICU) score] and oxygen (oxygen score) were calculated from information provided by hospitalized COVID-19 patients (n=44) via a web-based application. The performance of baseline scores to predict 30-day outcomes was assessed. RESULTS: Among 44 patients, 5 and 15 patients needed intensive care and oxygen, respectively. The area under the curve of ICU score and oxygen score to predict 30-day outcomes were 0.774 [95% confidence interval (CI): 0.614-0.934] and 0.728 (95% CI: 0.559-0.898), respectively. The ICU scores of patients needing intensive care increased daily by 0.71 points (95% CI: 0.20-1.22) after hospitalization and by 0.85 points (95% CI: 0.36-1.35) after symptom onset, which were significantly different from those in individuals not needing intensive care (p=0.002 and <0.001, respectively). Trends in daily oxygen scores overall were not markedly different; however, when the scores were evaluated within <7 days after symptom onset, the patients needing oxygen showed a higher daily increase in oxygen scores [1.81 (95% CI: 0.48-3.14) vs. -0.28 (95% CI: 1.00-0.43), p=0.007]. CONCLUSION: Our machine learning models showed good performance for predicting the outcomes of COVID-19 patients and could thus be useful for patient triage and monitoring.

Diagnostic index for acute eosinophilic pneumonia without bronchoscopy in military smokers.

BACKGROUND/AIMS: Acute eosinophilic pneumonia (AEP) is common among military smokers; however, bronchoscopy is required for the diagnosis. We aimed to derive and validate a scoring system to diagnose AEP without bronchoscopy. METHODS: We conducted a retrospective study including patients diagnosed with AEP or any other pneumonia among military smokers hospitalized in the Armed Forces Capital Hospital from 15 November 2016 through 25 December 2019. The patients were divided into derivation and validation groups according to their admission day. Patient symptoms, laboratory findings, and computed tomography findings were candidate variables. Least absolute shrinkage and selection operator (LASSO) regression was used to calculate the scores for each variable. RESULTS: Among 414 patients, AEP was confirmed in 54 of 279 patients (19.4%) in the derivation group and in 18 of 135 patients (13.3%) in the validation group. Ten variables were selected using LASSO regression: new-onset or a recently increased smoking (≤ 4 weeks) (8 points), interlobular septal thickening (5 points), absence of sputum (3 points), ground glass opacity (3 points), acute onset (≤ 3 days) (2 points), dyspnea (2 points), chest pain (2 points), leukocytosis (2 points), bronchovascular bundle thickening (2 points), and bilateral involvement (2 points). The area under the receiver-operating characteristic curve of the score to diagnose AEP was 0.997 (95% confidence interval, 0.992 to 1.000) in the derivation group and 0.985 (95% confidence interval, 0.965 to 1.000) in the validation group. CONCLUSION: We introduce a scoring system that can distinguish AEP from other types of pneumonia in military smokers without the need for bronchoscopy.

Development of a multipotent diagnostic tool for chest X-rays by multi-object detection method.

The computer-aided diagnosis (CAD) for chest X-rays was developed more than 50 years ago. However, there are still unmet needs for its versatile use in our medical fields. We planned this study to develop a multipotent CAD model suitable for general use including in primary care areas. We planned this study to solve the problem by using computed tomography (CT) scan with its one-to-one matched chest X-ray dataset. The data was extracted and preprocessed by pulmonology experts by using the bounding boxes to locate lesions of interest. For detecting multiple lesions, multi-object detection by faster R-CNN and by RetinaNet was adopted and compared. A total of twelve diagnostic labels were defined as the followings: pleural effusion, atelectasis, pulmonary nodule, cardiomegaly, consolidation, emphysema, pneumothorax, chemo-port, bronchial wall thickening, reticular opacity, pleural thickening, and bronchiectasis. The Faster R-CNN model showed higher overall sensitivity than RetinaNet, nevertheless the values of specificity were opposite. Some values such as cardiomegaly and chemo-port showed excellent sensitivity (100.0%, both). Others showed that the unique results such as bronchial wall thickening, reticular opacity, and pleural thickening can be described in the chest area. As far as we know, this is the first study to develop an object detection model for chest X-rays based on chest area defined by CT scans in one-to-one matched manner, preprocessed and conducted by a group of experts in pulmonology. Our model can be a potential tool for detecting the whole chest area with multiple diagnoses from a simple X-ray that is routinely taken in most clinics and hospitals on daily basis.

Prediction of patients requiring intensive care for COVID-19: development and validation of an integer-based score using data from Centers for Disease Control and Prevention of South Korea.

BACKGROUND: Unavailability or saturation of the intensive care unit may be associated with the fatality of COVID-19. Prioritizing the patients for hospitalization and intensive care may be critical for reducing the fatality of COVID-19. This study aimed to develop and validate a new integer-based scoring system for predicting patients with COVID-19 requiring intensive care, using only the predictors available upon triage. METHODS: This is a retrospective study using cohort data from the Korean Centers for Disease Control and Prevention that included all admitted patients with COVID-19 between January 19 and June 3, 2020, in South Korea. The primary outcome was patients requiring intensive care defined as actual admission to the intensive care unit; at any time use of an extracorporeal life support device, mechanical ventilation, or vasopressors; and death. Patients admitted until March 20 were included for the training dataset to develop the prediction models and externally validated for the patients admitted afterward. Two logistic regression models were developed with different predictors and the predictive performance was compared: one with patient-provided variables and the other with added radiologic and laboratory variables. An integer-based scoring system was developed based on the developed logistic regression model. RESULTS: A total of 5193 patients were considered, with 4663 patients included after excluding patients with age under 18 or insufficient data. For the training dataset, 3238 patients were included. Of the included patients, 444 (9.5%) patients required intensive care. The model developed with only the clinical variables showed an area under the curve of 0.884 for the validation set. The performance did not differ when radiologic and laboratory variables were added. Seven variables were selected for developing an integer-based scoring system: age, sex, initial body temperature, dyspnea, hemoptysis, history of chronic kidney disease, and activities of daily living. The area under the curve of the scoring system was 0.880. CONCLUSIONS: An integer-based scoring system was developed for predicting patients with COVID-19 requiring intensive care, with high performance. This system may aid decision support for prioritizing the patient for hospitalization and intensive care, particularly in a situation with limited medical resources.

Burden and clinical characteristics of high grade tuberculosis destroyed lung: a nationwide study.

BACKGROUND: Little clinical information on high grade tuberculosis destroyed lung (TDL) is available. The aim of this study was to investigate the characteristics and healthcare utilization of high grade TDL patients, and compared the differences between acute exacerbation and healthcare burden according to inhaler compliance. METHODS: This was an observational retrospective cohort study using the Korean Health Insurance Review and Assessment (HIRA) service database (2011-2015). Patients diagnosed with high grade TDL in 2011 were enrolled and reviewed for 5 years. The patients were further divided into adherent and non-adherent groups. Their socioeconomic outcomes according to treatment adherence in 2012 were analyzed. RESULTS: Among the 13,346 patients diagnosed with high grade TDL, 3,637 were assigned to the adherent group and 9,709 to the non-adherent group. Overall, 65.91% of the enrolled patients were male and the mean age of the study population was 64.68±10.06 years. All patients visited a tertiary hospital, but 99.04% and 69.74% also visited primary and secondary hospitals, respectively. The mean number of hospital admissions per year was 1.38±2.03 times per patient. The average total annual per-patient cost was US$4,140.95±3,715.01 and each patient spent a total of 56.21±45.28 days per year using hospital services. The majority of the patients in the adherent group were male (80.09% vs. 60.60%, P<0.01), and were of older age (65.71±9.35 vs. 64.29±10.28, P<0.01) than the non-adherent group. The frequencies of visiting a tertiary hospital (96.87 vs. 90.12%, P<0.01), the total mean healthcare utilization costs (US$4,151.77±4,084.76 vs. US$3,592.54±4,229.93, P<0.01), and the frequencies of exacerbations (0.72±2.03 vs. 0.46±1.51, P<0.01) were higher in the adherent group. However, healthcare services were used on significantly fewer days in the adherent group (52.96±50.87 vs. 56.67±50.81, P<0.01). CONCLUSIONS: High grade TDL imposes a high socioeconomic burden in Korea. Estimated medical costs and exacerbation event rate were higher in the adherent group whereas number of days of healthcare usage was significantly lower.

Incidence of Adverse Effects and Discontinuation Rate between Patients Receiving 250 Micrograms and 500 Micrograms of Roflumilast: A Comparative Study.

BACKGROUND: Roflumilast is the only approved oral phosphodiesterase-4 inhibitor for the treatment of severe chronic obstructive pulmonary disease (COPD) in patients with chronic bronchitis and a history of frequent exacerbations. The purpose of this study was to examine the incidence of adverse effects associated with roflumilast treatment in a real-world setting. Further, we compared the incidence of adverse effects and the discontinuation rate among patients receiving different doses. METHODS: We identified all outpatients diagnosed with COPD at Seoul St. Mary's Hospital between May 2011 and September 2016 and retrospectively reviewed their medical records. Roflumilast was prescribed to patients in doses of 500 µg and 250 µg. RESULTS: A total of 269 COPD patients were prescribed roflumilast in our hospital during the study period. Among them, 178 patients were treated with 500 µg and 91 patients were treated with 250 µg. The incidence of adverse effects was 38.2% in the 500 µg group and 25.3% in the 250 µg group (p=0.034). The discontinuation rate of roflumilast was 41.6% (n=74) in the 500 µg group and 23.1% (n=21) in the 250 µg group (p=0.003). When adjusted by age, sex, smoking status, and lung function, 500 µg dose was significantly associated with the discontinuation of roflumilast (odds ratio, 2.87; p<0.001). CONCLUSION: There was a lower incidence of adverse effects and discontinuation among patients treated with 250 µg compared with 500 µg dose. Further studies regarding the optimal dose of roflumilast are required.

Heterogeneity of asthma-COPD overlap syndrome.

Many patients suffering from asthma or COPD have overlapping features of both diseases. However, a phenotypical approach for evaluating asthma-COPD overlap syndrome (ACOS) has not been established. In this report, we examined the phenotypes in patients with ACOS. Patients diagnosed with ACOS between 2011 and 2015 were identified and classified into four phenotype groups. Group A was composed of patients who smoked <10 pack years and had blood eosinophil counts ≥300. Group B was composed of patients who smoked <10 pack years and had blood eosinophil counts <300. Group C was composed of patients who smoked ≥10 pack years and had blood eosinophil counts ≥300. Group D was composed of patients who smoked <10 pack years and had blood eosinophil counts <300. Clinical characteristics were analyzed and compared among groups. Comparisons were made among 103 ACOS patients. Patients in group D were oldest, while patients in group A were youngest. There were relatively more female patients in groups A and B; the majority of patients in groups C and D were male. The degree of airflow obstruction was most severe in group C. The rate of being free of severe exacerbation was significantly lower in group C than in the other groups. In this study, each ACOS phenotype showed different characteristics. The proportion of patients free of severe exacerbation differed significantly among groups. At this time, further studies on the phenotypes of ACOS are required.