A computer program for maximum likelihood estimation of survival probabilities under a logistic model

✍ Scribed by Sonja Johansen

• Publisher: Elsevier Science
• Year: 1973
• Weight: 365 KB
• Volume: 3
• Category: Article
• ISSN: 0010-468X
• DOI: 10.1016/0010-468x(73)90019-6

✦ Synopsis

Physical examinations made at annual or other fixed time intervals on each member of a group of healthy individuals yield sets of random variables measured over time, called time series. A computer program has been written which treats these measurements as predictor variables to be used in estimating the probability of occurrence of a medical event, such as a myocardial infarction or coronary heart death, during a specified time interval.

The statistical estimation procedure is maximum likelihood, and the probabilities are assumed to be multivariate logistic functions of the random variables. For each time interval, an individual is assumed to be classifiable into one of three mutually exclusive event categories (a trinomial model). These categories are: occurrence of (1) a non-lethal myocardial infarction, (2) a coronary heart death, and (3) neither of the above.