48.2%
5-Year Survival
4.9 years
Median Survival
3,000
Total Patients
2,480 (82.7%)
Deaths (Events)
Top Causal Factors
Most Protective: PR Positive
HR: 0.845 (p<0.001)
Most Harmful: Any Metastasis
HR: 1.485 (p<0.001)
Age Effect (per 10 years)
HR: 1.156 (p<0.001)
Survival by Molecular Subtype
Luminal B (HER2+)
51.1%
Luminal A/B
49.4%
Triple-negative
47.7%
HER2-enriched
44.9%
Methodology Overview
This dashboard employs causal inference techniques to identify factors that causally impact breast cancer survival. We use propensity score weighting (IPTW) combined with Cox proportional hazards models to estimate Average Treatment Effects (ATE). Causal graphs are learned using domain knowledge and validated through diagnostic tests.
Survival Statistics
Log-rank test p-value
< 0.001
Median Survival (Overall)
4.9 years (95% CI: 4.7-5.1)
Forest Plot of Causal Effects
Effect Sizes Table
| Factor | HR (95% CI) | P-value | Interpretation |
|---|---|---|---|
| PR Positive | 0.845 (0.789-0.905) | < 0.001 | Strongly Protective |
| ER Positive | 0.928 (0.865-0.995) | 0.078 | Protective |
| HER2 Positive | 1.123 (1.045-1.207) | 0.002 | Moderate Risk |
| Age (per 10y) | 1.156 (1.089-1.227) | < 0.001 | Risk Factor |
| Any Metastasis | 1.485 (1.342-1.644) | < 0.001 | High Risk |
| Lymph Nodes+ | 1.234 (1.156-1.318) | < 0.001 | Risk Factor |
Simplified Causal DAG
Age
ER Status
PR Status
HER2 Status
Metastasis
Survival
Patient Profile
Predicted Outcomes
5-Year Survival Probability
52.3%
Risk Score
Moderate
Median Survival
5.2 years
Treatment Effects by Subgroup
Subgroup Statistics
Sample Size
3,000
Events
2,480
Mortality Rate
82.7%
Propensity Score Overlap
Covariate Balance
Model Assumptions
Proportional Hazards (Schoenfeld)
✓ p = 0.342
Linearity (Martingale)
✓ p = 0.156
Overlap (Common Support)
✓ 98.7% overlap
Balance (SMD < 0.1)
✓ All covariates
Data Quality
Missing Data
2.3%
Complete Cases
2,931 (97.7%)
Follow-up Time
Median 4.2 years
Event Rate
82.7%