MODELLING PROGRESSION IN STAGE III COLON CANCER USING SURVIVAL ANALYSIS

Iseh, Matthew Joshua, John, Eme Eseme , Matthew, Michael Aloysius — Fri, 10 Apr 2026 00:00:00 +0000

Stage III colon cancer poses a significant challenge for the treatment and management. This study aimed to model the progression of stage III colon cancer using survival analysis and identify key predictors of survival and disease recurrence. A retrospective cohort study was conducted using data from 929 patients with stage III colon cancer. Kaplan-Meier analysis and Cox proportional hazards regression were used to estimate survival rates and identify significant predictors of overall survival. The results show that using the Kaplan Meier approach the probability of survival started with 67% in 1-36 months and increases with time to 88% in 37-72 months, then increases at its peak in 73-108 months to approximately 91%. The result further shows a typical Kaplan Meier property of stepwise decrement from state 1 to state 2 in each of the groups. Also, the log-rank test demonstrates an unadjusted comparison between the groups and reveals that the survival probabilities in each group are the same. In the overall evaluation, the result reveals that two parameters (number of positive lymph nodes and extent of local spread) considered in this study are significant and contribute immensely to the survival of patients.to

OPTIMAL WEIGHT ADJUSTMENT WITH SUBSAMPLING THE NONRESPONDENTS IN LONGITUDINAL SURVEY FOR SMALL AREA ESTIMATION

Iseh, Matthew Joshua Bassey, Mbuotidem Okon , Etebong P. Clemen — Fri, 10 Apr 2026 00:00:00 +0000

Efficient estimation of population parameters in small area estimation (SAE) is crucial, especially under stratified random sampling when nonresponse occurs. Several estimators have been developed to improve estimation accuracy. However, some of these methods often fail to fully adjust for nonresponse bias, leading to unreliable domain mean estimates. This study proposes a new calibration-based estimator, by integrating the Kullback-Leibler distance function-based weight adjustments. The estimator is formulated under two conditions: when nonresponse affects only the study variable, and when nonresponse affects both study and auxiliary variables. Properties of the estimators have been derived and the results confirm that the proposed estimator provides greater efficiency and lower error rates than existing methods. Empirical validation is conducted using data from a longitudinal survey (before, during, and after COVID-19) from Household Finance & Consumption Survey (HFCS) and Integrated Household Survey (IHS) for 2019-2021 from the Department of Statistic of the Central Bank of Nigeria. Comparative performance analysis using variance and mean square error calculations demonstrates that the proposed estimator consistently outperforms the existing domain estimators in handling nonresponse across different domains. The study concludes that calibration techniques enhance estimation accuracy in stratified sampling and offers recommendations for further research on alternative calibration functions in small area estimation.

Academic Journal of Statistic and Mathematics

MODELLING PROGRESSION IN STAGE III COLON CANCER USING SURVIVAL ANALYSIS

OPTIMAL WEIGHT ADJUSTMENT WITH SUBSAMPLING THE NONRESPONDENTS IN LONGITUDINAL SURVEY FOR SMALL AREA ESTIMATION