Application of Survival Analysis to Data from Discharge Abstract of Medical Record Department: Focused on Readniission.
- Author:
Kwisook CHOI
1
;
Joonhyun HONG
;
Jeonghwa LEE
Author Information
1. Graduate School of Health Science and Management, Yonsei University, Korea. kschoi03@hanmail.net
- Publication Type:Original Article
- Keywords:
Survival Analysis;
Data Mining;
Discharge Summary;
Readmission
- MeSH:
Congenital Abnormalities;
Information Storage and Retrieval;
Data Mining;
Death Certificates;
Humans;
Medical Records*;
Proportional Hazards Models;
Survival Analysis*
- From:Journal of Korean Society of Medical Informatics
2000;6(4):35-43
- CountryRepublic of Korea
- Language:Korean
-
Abstract:
Abundant data on patients have been accumulated in hospital since the introduction of the computerized system. Now data mining is required for the survival and growth of hospital. Cases of 19,558 patients were analyzed to investigate factors influencing readmission and repeated admissions, and to estimate probability of readmission with considering covariate effects. Techniques of Kaplan-Meier method, Cox proportional hazards model, and WLW method were applied to the analysis. The conclusions are as follows. The severity of disease, congenital defect and chronicity of disease are influencing readmission or repeated admissions of a patient. Patient s characteristics, such as gender, distance from residence and type of discharge are also related to them. The probability of readmission can be estimated for a patient with variety of conditions for certain period of time. It is suggestive that survival analysis is a good methodology for data mining works on computerized data in hospital. If death certificate data are connected with patients' data, we will be able to get a good data source to medical studies.
