An App for Inspecting the Epidemic Trend of the Confirmed Cases on COVID-19 using the Online Rasch Model and Strength Coefficients(SC): An Observational Study (Preprint)
BACKGROUND When a novel coronavirus (e.g., COVID -19) starts to spread, one of the many questions asked is with regards to the trend of new confirmed cases increasing or decreasing during the on-going outbreak epidemic. Finding the turning point of the outbreak spread (i.e., from ascending to declining status) is continuously an urgent concern. Due to different weights (e.g.., an extremely high proportion of confirmed cases in Hubei, China) of the cases confirmed in countries/regions, using either the overall total number or a partially small portion of the infected-case regions to determine the turning point (e.g., the trend to decline) is problematic and unreliable. Rasch analysis used for examining individual performances of school students was thus considered as a tool to inspect the epidemic trend through the pattern of item (e.g., day in epidemic) difficulties over days. OBJECTIVE This study aims to (1) inspect the epidemic trend by performing Rasch model and observing the pattern of item difficulties over days, (2) develop an online algorithm to draw the trend plot, and (3) design an app for a better understanding of the outbreak situation on Google Maps. METHODS We downloaded the COVID-19 outbreak numbers from January 21 to February 27, 2020, from Github that contains information on confirmed cases in more than 30 Chinese locations and other countries/regions. Item (i.e., day) difficulties based on the recent 20 days were calibrated using the Rasch model. All responses were derived from the ordinal scores by using the logarithm function (i.e.., round(ln(confirmed cases),0) from 0 to 5). The epidemic trend was assessed by the correlation coefficients (CC) computed by the item difficulties along with the time points of days. The recent several CCs were plotted with a line chart. An online algorithm based on the Rasch model was built for displaying the outbreak trend on a daily basis. A strength coefficient(SC) was complemented to examine the outliers for each region in the recent three days. An app was developed to understand the daily epidemic trends on Google Maps. RESULTS The CCs measured by item(i.e., day) difficulties have been monotonously increased from -0.28(start from Feb. 8) to 0.36(till Feb. 27), indicating the epidemic trend has gradually declined. However, the trend out of China is increasing with the CC=-0.88. The SC was taken into consideration in three countries/regions: Italy(=0.87), Iran(0.59), and Shandong(China)(=0.58) on Feb. 21, 2020. A line chart was drawn online using item difficulties calibrated by the Rasch model for examining the epidemic trend. A dashboard was created to present the COVID-19 situation on Google Maps. CONCLUSIONS We created an online Rasch modeling algorithm that can calibrate daily item difficulties and then draw a line chart to analyze the epidemic trend. The SC is complemental to the trend observation. An app developed for displaying the epidemic trend helps us better understand the outbreak situation. CLINICALTRIAL Not available