Profile analysis and an application in nuclear medicine dosimeter data

Multivariate analysis of variance (MANOVA) examines whether there is a difference between the levels of the relevant factor (group) or factors in terms of the means of multiple response variables. In multivariate analysis of variance, if the considered variable measured at different times (minutes, hours, days, weeks, months, years), time is called the dependent factor for this variable, and the levels of this dependent factor are expressed as periods. The analysis of experimental designs with this structure is generally known as repeated measures analysis of variance. Profile analysis, which is related to repeated measures analysis of variance and is a partially simplified form of this analysis, is briefly used to test the similarity of values between individuals in terms of the characteristic of interest, or in other words, to determine whether two or more groups have similar profiles. Profile analysis is performed by showing the levels of the dependent factor on the x-axis and the levels of the independent factor on the y-axis, and then showing the averages for the considered variable. This allows us to test whether the profiles are parallel and whether the averages are the same (changing) at the levels of the independent factor. In this study, the general characteristics of profile analysis are explained, and an application was performed with using nuclear medicine dosimetry data. The application material consisted of data taken from the body and skin surface of 23 individuals working in a nuclear medicine unit over 7 periods. Then, the variables of weight, height, and age were divided into two categories: below and above the mean. Profile analysis was performed to determine the change in dosimetry values taken from the body and skin surface over time, in relation to weight, height, age, marital status, education level, and occupation variables. The results of the analysis showed that parallelism tests for both body and skin measurements were statistically insignificant for all variables, and it was determined that the groups exhibited similar patterns over time. When level equality tests were examined, a statistically significant difference was found between the groups only in the occupation variable in terms of body measurements; in all other variables, there was no significant difference between the overall mean levels of the groups. In skin measurements, level equality tests were found insignificant for all variables. In contrast, flatness tests were found statistically significant for all variables in both body and skin measurements, revealing that dosimeter measurements did not show a constant pattern over time, but exhibited significant changes depending on the periods. It was observed that profile analysis is simpler and easier to interpret than MANOVA, and thus usable in the applications.