Landmark Analysis Explorer

Reset the survival clock at a landmark time to see how conditioning on survival-to-date changes the predicted curve and its eligible population.

  • landmark analysis
  • conditional survival
  • dynamic prediction
  • survival analysis

Baseline vs. landmark survival prediction

s = 1.0 yr01234567800.20.40.60.81Years since surgerySurvival probability
Baseline (from t=0)Landmark (conditional on T > s)

Risk set size over time

Eligible: 223012345678050100150200250Years since surgeryNumber at risk

Eligibility at the landmark

Total patients
260.000
Eligible at s
223.000
Ineligible (event/censored before s)
37.000
S(s)
0.858

Prediction formula

SymbolMeaningCurrent value
sLandmark time1.0 yr
τPrediction window2.0 yr
EstimandP(T > s + τ | T > s, H(s))
Conditioning changes the question, not just the curve
The baseline curve answers "survival from the moment of surgery." The landmark curve, reset to 1.0 at s and computed only from patients who survived to s, answers a different and often more relevant question for a patient sitting in front of you at their s-year follow-up: survival conditional on having already made it this far.