Predicting hospitalization of COVID-19 positive patients using clinician-guided machine learning methods.

OBJECTIVES: The coronavirus disease 2019 (COVID-19) is a resource-intensive global pandemic. It is important for healthcare systems to identify high-risk COVID-19-positive patients who need timely health care. This study was conducted to predict the hospitalization of older adults who have tested positive for COVID-19. METHODS: We screened all patients with COVID test records from 11 Mass General Brigham hospitals to identify the study population. A total of 1495 patients with age 65 and above from the outpatient setting were included in the final cohort, among which 459 patients were hospitalized. We conducted a clinician-guided, 3-stage feature selection, and phenotyping process using iterative combinations of literature review, clinician expert opinion, and electronic healthcare record data exploration. A list of 44 features, including temporal features, was generated from this process and used for model training. Four machine learning prediction models were developed, including regularized logistic regression, support vector machine, random forest, and neural network. RESULTS: All 4 models achieved area under the receiver operating characteristic curve (AUC) greater than 0.80. Random forest achieved the best predictive performance (AUC = 0.83). Albumin, an index for nutritional status, was found to have the strongest association with hospitalization among COVID positive older adults. CONCLUSIONS: In this study, we developed 4 machine learning models for predicting general hospitalization among COVID positive older adults. We identified important clinical factors associated with hospitalization and observed temporal patterns in our study cohort. Our modeling pipeline and algorithm could potentially be used to facilitate more accurate and efficient decision support for triaging COVID positive patients.

Sequential coupling of dry and wet COVID-19 screening to reduce the number of quarantined individuals.

INTRODUCTION: Currently, several countries are facing severe public health and policy challenges when designing their COVID-19 screening strategy. A quantitative analysis of the potential impact that combing the Rapid Antigen Test (RAT; Wet screening) and digital checker (Dry screening) can have on the healthcare system is lacking. METHOD: We created a hypothetical COVID-19 cohort for the analysis. The population size was set as 10 million with three levels of disease prevalence (10%, 1%, or 0.1%) under the assumption that a positive test result will lead to quarantine. A digital checker and two RATs are used for analysis. We further hypothesized two scenarios: RAT only and RAT plus digital checker. We then calculated the number of quarantined in both scenarios and compared the two to understand the benefits of sequential coupling of a digital checker with a RAT. RESULT: Sequential coupling of the digital checker and RAT can significantly reduce the number of individuals quarantined to 0.95-1.33M, 0.86-1.29M, and 0.86-1.29M, respectively, under the three different prevalence levels. CONCLUSION: Sequential coupling of digital checker and RAT at a population level for COVID-19 positive test to reduce the number of people who require quarantine and alleviating stress on the overburdened healthcare systems during the COVID-19 pandemic.

Estimating the time interval between transmission generations and the presymptomatic period by contact tracing surveillance data from 31 provinces in the mainland of China

The global pandemic of 2019 coronavirus disease (COVID-19) is a great assault to public health. Presymptomatic transmission cannot be controlled with measures designed for symptomatic persons, such as isolation. This study aimed to estimate the interval of the transmission generation (TG) and the presymptomatic period of COVID-19, and compare the fitting effects of TG and serial interval (SI) based on the SEIHR model incorporating the surveillance data of 3453 cases in 31 provinces. These data were allocated into three distributions and the value of AIC presented that the Weibull distribution fitted well. The mean of TG was 5.2 days (95% CI: 4.6-5.8). The mean of the presymptomatic period was 2.4 days (95% CI: 1.5-3.2). The dynamic model using TG as the generation time performed well. Eight provinces exhibited a basic reproduction number from 2.16 to 3.14. Measures should be taken to control presymptomatic transmission in the COVID-19 pandemic.

Derivation of a Clinical Risk Score to Predict 14-Day Occurrence of Hypoxia, ICU Admission, and Death Among Patients with Coronavirus Disease 2019.

BACKGROUND: Uncertainty surrounding COVID-19 regarding rapid progression to acute respiratory distress syndrome and unusual clinical characteristics make discharge from a monitored setting challenging. A clinical risk score to predict 14-day occurrence of hypoxia, ICU admission, and death is unavailable. OBJECTIVE: Derive and validate a risk score to predict suitability for discharge from a monitored setting among an early cohort of patients with COVID-19. DESIGN: Model derivation and validation in a retrospective cohort. We built a manual forward stepwise logistic regression model to identify variables associated with suitability for discharge and assigned points to each variable. Event-free patients were included after at least 14 days of follow-up. PARTICIPANTS: All adult patients with a COVID-19 diagnosis between March 1, 2020, and April 12, 2020, in 10 hospitals in Massachusetts, USA. MAIN MEASURES: Fourteen-day composite predicting hypoxia, ICU admission, and death. We calculated a risk score for each patient as a predictor of suitability for discharge evaluated by area under the curve. KEY RESULTS: Of 2059 patients with COVID-19, 1326 met inclusion. The 1014-patient training cohort had a mean age of 58 years, was 56% female, and 65% had at least one comorbidity. A total of 255 (25%) patients were suitable for discharge. Variables associated with suitability for discharge were age, oxygen saturation, and albumin level, yielding a risk score between 0 and 55. At a cut point of 30, the score had a sensitivity of 83% and specificity of 82%. The respective c-statistic for the derivation and validation cohorts were 0.8939 (95% CI, 0.8687 to 0.9192) and 0.8685 (95% CI, 0.8095 to 0.9275). The score performed similarly for inpatients and emergency department patients. CONCLUSIONS: A 3-item risk score for patients with COVID-19 consisting of age, oxygen saturation, and an acute phase reactant (albumin) using point of care data predicts suitability for discharge and may optimize scarce resources.

Massive migration promotes the early spread of COVID-19 in China: a study based on a scale-free network.

BACKGROUND: The coronavirus disease 2019 (COVID-19) epidemic met coincidentally with massive migration before Lunar New Year in China in early 2020. This study is to investigate the relationship between the massive migration and the coronavirus disease 2019 (COVID-19) epidemic in China. METHODS: The epidemic data between January 25th and February 15th and migration data between Jan 1st and Jan 24th were collected from the official websites. Using the R package WGCNA, we established a scale-free network of the selected cities. Correlation analysis was applied to describe the correlation between the Spring Migration and COVID-19 epidemic. RESULTS: The epidemic seriousness in Hubei (except the city of Wuhan) was closely correlated with the migration from Wuhan between January 10 and January 24, 2020. The epidemic seriousness in the other provinces, municipalities and autonomous regions was largely affected by the immigration from Wuhan. By establishing a scale-free network of the regions, we divided the regions into two modules. The regions in the brown module consisted of three municipalities, nine provincial capitals and other 12 cities. The COVID-19 epidemics in these regions were more likely to be aggravated by migration. CONCLUSIONS: The migration from Wuhan could partly explain the epidemic seriousness in Hubei Province and other regions. The scale-free network we have established can better evaluate the epidemic. Three municipalities (Beijing, Shanghai and Tianjin), eight provincial capitals (including Nanjing, Changsha et al.) and 12 other cities (including Qingdao, Zhongshan, Shenzhen et al.) were hub cities in the spread of COVID-19 in China.

Linking key intervention timings to rapid declining effective reproduction number to quantify lessons against COVID-19.

Coronavirus disease 2019 (COVID-19) is currently under a global pandemic trend. The efficiency of containment measures and epidemic tendency of typical countries should be assessed. In this study, the efficiency of prevention and control measures in China, Italy, Iran, South Korea, and Japan was assessed, and the COVID-19 epidemic tendency among these countries was compared. Results showed that the effective reproduction number(Re) in Wuhan, China increased almost exponentially, reaching a maximum of 3.98 before a lockdown and rapidly decreased to below 1 due to containment and mitigation strategies of the Chinese government. The Re in Italy declined at a slower pace than that in China after the implementation of prevention and control measures. The Re in Iran showed a certain decline after the establishment of a national epidemic control command, and an evident stationary phase occurred because the best window period for the prevention and control of the epidemic was missed. The epidemic in Japan and South Korea reoccurred several times with the Re fluctuating greatly. The epidemic has hardly rebounded in China due to the implementation of prevention and control strategies and the effective enforcement of policies. Other countries suffering from the epidemic could learn from the Chinese experience in containing COVID-19.