|
|
 |
| |
|
|
|
Statistical Modeling of Biomarker Time Series Using a Generalized Gamma-Like Distribution |
|
|
|
PP: 311-320 |
|
|
|
Author(s) |
|
|
|
Piyush Kumar Mishra,
Javid Gani Dar,
|
|
|
|
Abstract |
|
|
| In this study, a generalized gamma-like probability distribution with a flexible parameterization is introduced, and extensive properties of this probability distribution are discussed and demonstrated to provide more utility in using this probability distribution to model data containing skewed and heavy tails. Its cumulative distribution function (CDF), moments, entropy, and hazard rate functions are derived, and we estimate them using maximum likelihood methods. As a new application, we use this distribution to fit multivariate time-series data on medical biomarkers for oncology patients. The proposed model has the distributional properties of biological variables, namely non-normality and asymmetry. A comparative analysis reveals that the model outperforms traditional parametric methods in capturing the dynamics of the data. The work not only makes a theoretical contribution to the development of statistical distributions but also a practical one to real-world medical analytics, particularly in early-stage anomaly detection, disease progression modeling, and other applications. |
|
|
|
|
|
|
|
