Novel model prediction time-to-event analysis: data validation and estimation of 200 million cases in the global COVID-19 epidemic.
BMJ Open
; 13(1): e065487, 2023 01 27.
Article
in English
| MEDLINE | ID: covidwho-2223665
ABSTRACT
OBJECTIVES:
Assessment of recuperation and death times of a population inflicted by an epidemic has only been feasible through studying a sample of individuals via time-to-event analysis, which requires identified participants. Therefore, we aimed to introduce an original model to estimate the average recovery/death times of infected population of contagious diseases without the need to undertake survival analysis and just through the data of unidentified infected, recovered and dead cases.DESIGN:
Cross-sectional study.SETTING:
An internet source that asserted from official sources of each government. The model includes two techniques-curve fitting and optimisation problems. First, in the curve fitting process, the data of the three classes are simultaneously fitted to functions with defined constraints to derive the average times. In the optimisation problems, data are directly fed to the technique to achieve the average times. Further, the model is applied to the available data of COVID-19 of 200 million people throughout the globe.RESULTS:
The average times obtained by the two techniques indicated conformity with one another showing p values of 0.69, 0.51, 0.48 and 0.13 with one, two, three and four surges in our timespan, respectively. Two types of irregularity are detectable in the data, significant difference between the infected population and the sum of the recovered and deceased population (discrepancy) and abrupt increase in the cumulative distributions (step). Two indices, discrepancy index (DI) and error of fit index (EI), are developed to quantify these irregularities and correlate them with the conformity of the time averages obtained by the two techniques. The correlations between DI and EI and the quantified conformity of the results were -0.74 and -0.93, respectively.CONCLUSION:
The results of statistical analyses point out that the proposed model is suitable to estimate the average times between recovery and death.Keywords
Full text:
Available
Collection:
International databases
Database:
MEDLINE
Main subject:
Epidemics
/
COVID-19
Type of study:
Observational study
/
Prognostic study
/
Randomized controlled trials
Limits:
Humans
Language:
English
Journal:
BMJ Open
Year:
2023
Document Type:
Article
Affiliation country:
Bmjopen-2022-065487
Similar
MEDLINE
...
LILACS
LIS