표제지
국문 초록
ABSTRACT
목차
1장 서론 15
1.1. 단일 수준으로의 치료전환 보정 방법 17
1.1.1. Intention-to-treat(ITT) 17
1.1.2. Censoring switcher 17
1.1.3. Inverse probability Censoring weighting(IPCW) 18
1.1.4. Rank Preserving Structure Failure Time model(RPSFTM) 19
1.2. 다 수준으로의 치료 전환 보정 방법 21
1.2.1. Stratified RPSFTM(SRP) 23
1.2.2. Random- Forest-Based Prediction(RF) 25
2장 연구 방법 27
2.1. Iterative Parameter Estimation(IPE) 27
2.2. Stratified IPE(SIPE) 29
3장 모의실험 34
3.1. 생존시간 생성 34
3.2. 전환 메커니즘 36
3.3. 모의실험 설계 37
3.4. 평가 항목 38
4장 모의실험 결과 41
5장 고찰 63
6장 결론 65
참고문헌 67
Table 1. List of scenarios in simulation 38
Table 2. Result for scenarios with N=200, treatment effect=-0.3, censoring rate=25% 43
Table 3. Result for scenarios with N=200, treatment effect=-0.3, censoring rate=50% 44
Table 4. Result for scenarios with N=200, treatment effect=-0.3, censoring rate=75% 45
Table 5. Result for scenarios with N=200, treatment effect=-0.5, censoring rate=25% 46
Table 6. Result for scenarios with N=200,treatment effect=-0.5, censoring rate=50% 47
Table 7. Result for scenarios with N=200, treatment effect=-0.5, censoring rate=75% 48
Table 8. Result for scenarios with N=200, treatment effect=-0.8, censoring rate=25% 49
Table 9. Result for scenarios with treatment N=200, effect=-0.8, censoring rate=50% 50
Table 10. Result for scenarios with N=200, treatment effect=-0.8, censoring rate=75% 51
Table 11. Result for scenarios with N=400, treatment effect=-0.3, censoring rate=25% 52
Table 12. Result for scenarios with N=400, treatment effect=-0.3, censoring rate=50% 53
Table 13. Result for scenarios with N=400, treatment effect=-0.3, censoring rate=75% 54
Table 14. Result for scenarios with N=400, treatment effect=-0.5, censoring rate=25% 55
Table 15. Result for scenarios with N=400, treatment effect=-0.5, censoring rate=50% 56
Table 16. Result for scenarios with N=400, treatment effect=-0.5, censoring rate=75% 57
Table 17. Result for scenarios with N=400, treatment effect=-0.8, censoring rate=25% 58
Table 18. Result for scenarios with N=400, treatment effect=-0.8, censoring rate=50% 59
Table 19. Result for scenarios with N=400, treatment effect=-0.8, censoring rate=75% 60
Figure 1. Multi-treatment switching diagram 22
Figure 2. Stratified IPE algorithm flow chart 31
Figure 3. HR density plot 40
Figure 4. Bias plot (N=200, 400) 61
Figure 5. MSE plot (N=200, 400) 62