Artikel
The growth curve and the death rate of SARS
Suche in Medline nach
Autoren
Veröffentlicht: | 26. Mai 2004 |
---|
Gliederung
Text
The epidemic analysis of SARS differs from the lifetime analysis, but the probabilistic growth curves fitted to the infected, fatal, and cured cases of SARS may similarly be treated as the lifetime analysis. Using the truncated data models to the infected, fatal, and cured cases with some censoring time, we can estimate the underlying probability distribution function. In the case in Hong Kong, the lognormal distribution function is best fitted to these data cases among the Weibull, gamma, and lognormal models.
Since SARS shows a severe death rate, our next concern is to know such a death rate as soon as possible after a similar outbreak begins. Using, again, the truncated data models to the infected and fatal cases, we can estimate the total (or final) numbers of the patients and deaths, and the death rate may be estimated using these two numbers. We may also estimate the death rate using the numbers of the patients and recoveries, but this estimate differs from that using the numbers of the patients and deaths, especially when the censoring time is located at early stages.
To circumvent this inconsistency, and to obtain much more reliable estimates, we propose a mixed trunsored model which can use the data of the patients, deaths, and recoveries simultaneously; the trunsored model is a new incomplete data model regarded as a unified model of the censored and truncated models in lifetime analysis [[Ref. 1]]. The estimate of the death rate and its error are easily and stably obtained in a numerical sense. In the case in Hong Kong, the death rate is estimated to be about 17%. The estimated death rates in Canada, Hong Kong, Taiwan, Singapore, and Viet Nam are regarded as the similar rate, but the rate in China is very different from those in other districts.
References
- 1.
- H. Hirose, The Trunsored model and its applications to lifetime analysis: a unification of the censored and truncated models, IEEE Transactions on Reliability, to appear.