This curve is generated using a Random Survival Forest (RSF) model. It depicts the predicted survival probability over time (in years), based on the patient’s clinical or risk-factor inputs. Each point on the curve reflects the model’s estimated probability that the patient remains alive up to that point in time.

• Patient-Specific Curve: The solid line shows this individual’s predicted survival probability at each time point.
• Cohort Curve (if visible): A dashed line may indicate the average survival for the broader population used to train/validate the model.

These estimates are probabilistic in nature and should be interpreted within the broader clinical context. Factors such as comorbidities, treatment variations, and unseen biological variables may affect actual outcomes.

DeepHit is a cutting-edge survival analysis model that uses a deep neural network to predict the likelihood of different outcomes over time. Unlike traditional methods, DeepHit:

• Does not assume a specific mathematical model (no strong parametric assumptions).
• Directly learns the distribution of survival times from the data.
• Handles competing risks—scenarios where multiple events (e.g., cancer-related death vs. death from other causes) are possible, and only one can occur.

These estimates are probabilistic in nature and should be interpreted within the broader clinical context. Factors such as comorbidities, treatment variations, and unseen biological variables may affect actual outcomes.