SURVIVAL MODEL AND ESTIMATION FOR LUNG CANCER PATIENTS - Quantitative Medical Data Analysis Using Mathematical Tools and Statistical Techniques - page 217

Fig. 5. Estimated Baseline Functions.

A scatter plot for “score vs. survival time” is shown in Figure 6. Notice
that time assumes a negative value if it is censored (patient is still alive.)

Fig. 6. Scores vs. Time to Death or Censoring.

Figure 6 shows that when a patient scores negative or very small value,
he or she tends to survive; the lower the score is, the longer he or she will
live. On the other hand, a high positive score means death. This proves that
proportional hazard regression is a beneficial way to estimate β coecients.
Final Remarks: 1. In this study, we set up a survival model for lung

cancer patients. This was achieved by three steps: using proportional hazard
regression to estimate the coe cients for five covariates, using non-linear
least square fit to estimate the exponential baseline hazard function, and
using a neural network to exam the survival model. The analysis tools used