Linear Regression
The classic regression analysis is probably the most popular method for predicting disease occurrence. The task of the regression is to find the unknown parameters and form functional relationships between the incidence and determining factors. There are two models of regression. Linear regression is the statistical method predicting values of the dependent variable Y from the values of the independent variable X (Steyerberg et al., 2010). The linear regression is fast and easy to obtain. Moreover, it is transparent and clear to analyze. The obtained regression coefficients can be judged on how this or that factor affects the result to make on this basis more useful conclusions.
Logistic Regression
The logistic regression is similar to the linear one. It is used when there is interest in a binary outcome, in other words, the presence or absence of a symptom or a subject and several predictors. From a logistic regression equation, one can determine which predictors influence the outcome and evaluate the likelihood of a particular outcome using the values of the predictors of the patient. For example, complications or the effectiveness of the treatment might be predicted with some degree of confidence.
Multiple Regression
Most often, several factors affect the prediction. Consequently, the multiple regression model with several factors is useful while predicting the appearance of the disease. In particular, the effect of each factor separately as well as their total impact on the simulated index might be calculated. The most common is the incremental method, which has the same structure as the direct method, but after each step, the variables are investigated over in a reverse method. In this method, blocks of independent variables can be specified. In this case, predetermined blocks are processed in one step together.
Survival Analysis
The survival analysis is mainly applied to the same statistical problems as other methods, but its peculiarity is that it is associated with incomplete data. It should be noted that the survival function represents the probability that the patient would live a longer period. The construction of life tables, survival distribution fitting, estimation of survival function using the Kaplan-Meier procedure is descriptive research methods for incomplete data (Jewell, Kimber, Lee, & Whitmore, 2013). Some of these methods allow reflecting survival in different groups. In addition, survival analysis contains regression models to evaluate multidimensional relationships between permanent constants such as lifetime and others. The most natural way to describe the survival rate is the tabulation of the lifetime. This method is one of the oldest methods for the analysis of survival data. Such a table can be considered as the extended frequency table. The area of possible times of occurrence of critical events such as death, failure, and others is divided into several intervals. For each interval, the proportion of patients who are alive along with the proportion of those who died should be calculated.
Conclusion
In conclusion, it was stated that statistical prediction models might be used to estimate the nature of the potential disease and its occurrence probability. The introduction of statistical prediction models would enable physicians to assess the likelihood of the disease objectively
References
Jewell, N. P., Kimber, A. C., Lee, M. T., & Whitmore, G. A. (2013). Lifetime data: Models in reliability and survival analysis. New York, NY: Springer.
Steyerberg, E. W., Vickers, A. J., Cook, N. R., Gerds, T., Gonen, M., Obuchowski, N.,… Kattan, M. W. (2010). Assessing the Performance of Prediction Models. Epidemiology, 21(1), 128-138.