Section 3: Results (from DOI:10.14218/ERHM.2020.00023)

Navigation

Article	Trends and Prediction in Daily New Cases and Deaths of COVID-19 in the United States: An Internet Search-Interest Based Model
Sections in this Publication
Section	Section 1: Introduction (from DOI:10.14218/ERHM.2020.00023)
Section	Section 2: Methods (from DOI:10.14218/ERHM.2020.00023)
Section	Section 3: Results (from DOI:10.14218/ERHM.2020.00023)
Section	Section 4: Discussion (from DOI:10.14218/ERHM.2020.00023)
Section	Section 5: Future directions (from DOI:10.14218/ERHM.2020.00023)
Section	Section 6: Conclusions (from DOI:10.14218/ERHM.2020.00023)
Section	References (from DOI:10.14218/ERHM.2020.00023)
Named Entities in this Section
Entity	COVID-19 (disease - MeSH supplementary concept)
Entity	Cardiac Death (disease - MeSH descriptor)
Entity	unconventional SNARE in the ER 1 (gene)
Entity	2019 novel coronavirus (species)
Dataset	Pubtator Central BioC-JSON formatted article files

From publication: "Trends and Prediction in Daily New Cases and Deaths of COVID-19 in the United States: An Internet Search-Interest Based Model" published as Explor Res Hypothesis Med; 2020 Apr 18 ; 5 (2) 1-6. DOI: https://doi.org/10.14218/ERHM.2020.00023

Section 3: Results

The Johns Hopkins data repository and 1-point-3-acres.com provided slightly different estimates of COVID-19 daily new cases and deaths in the USA, although they claimed to share data. The data of a given date from 1-point-3-acres.com dataset varied by the release dates. Considering the data inconsistency, we chose the John Hopkins' data for modelling, and the 1-point-3-acres.com data for a sensitivity study. There were 636,282 new cases and,325 deaths of COVID-19 reported in the USA from March 1 to April 15, 2020, with a crude mortality of 4.45%. The daily new cases peaked at 35,098 cases on April 10, 2020 and the daily deaths peaked at 2,494 on April 15, 2020.

Google Trends search-interests had a 2-day delay in reporting (i.e. a search on April 9 yielded data up to April 7). COVID-19 has a much lower search interest score than COVID (Fig. 1), and was excluded from additional analysis also owing to its close relationship with COVID. As reported before, the correlation coefficients of search terms changed with lag time (Fig. 2). Among the nine terms we searched, COVID, "COVID pneumonia" and "COVID heart" had the top-3 correlation coefficients for the correlation with daily incidence and new deaths (Table 1). Our predicted COVID-19 daily new cases and new deaths would plateau for about 12 days (Fig. 3), suggesting a possible 12-day plateau of these epidemiologic parameters in the future.

The sensitivity study using 1-point-3-acres' data revealed the correlation coefficients that were similar to those produced using Johns Hopkins' data (Table 1). The 7-day follow-up with prospectively collected data showed no significant correlations of the observed data with the predicted daily new cases using search-interest of COVID, COVID heart and COVID pneumonia (p = 0.178, 0.480 and 0.094, respectively) nor with the predicted daily new deaths using search interest of COVID, COVID heart and COVID pneumonia (p = 0.267, 0.222 and 0.841, respectively).