Olivier Grisel - Predictive survival analysis with scikit-learn, scikit-survival and lifelines

Discover predictive survival analysis techniques with scikit-learn, scikit-survival, and lifelines libraries, covering time-to-event regression, Kaplan-Meier estimator, and more.

Key takeaways

Predictive survival analysis is used to estimate the survival function or the probability of survival over time, given a set of predictive features.
The concept of time-to-event regression is used to model the time until an event occurs, and is applicable to various fields such as medicine, insurance, and finance.
The Kaplan-Meier estimator is a measure of the survival function at a specific time point, and is widely used in survival analysis.
Censoring is a common problem in survival analysis, where some individuals do not experience the event of interest, and can be dealt with by using the Kaplan-Meier estimator.
The hazard rate is a measure of the speed of failure or the rate at which failures occur over time, and is used to model the time-to-event.
The concordance index is a measure of the quality of a model, and is used to evaluate the ability of a model to predict the correct order of events.
Non-linear feature engineering can be used to mitigate some of the limitations of traditional survival analysis methods.
The Brier score is a proper scoring rule for a classification model, and is used to evaluate the quality of a model.
Predictive survival analysis can be used in various fields such as medicine, insurance, and finance to model the likelihood of an event occurring over time.
The lifelines library is a Python library that provides a variety of tools for survival analysis, including the Kaplan-Meier estimator and the concordance index.
The scikit-survival library is a Python library that provides a variety of tools for survival analysis, including the Kaplan-Meier estimator and the concordance index.
The Cox proportional hazards model is a widely used model in survival analysis, and can be used to model the time-to-event while accounting for the effects of covariates.
Enriched feature sets can be used to improve the performance of a model, and can be created by combining different types of features.
Time-dependent ROC curves can be used to evaluate the performance of a model over time.
The Israel index is a measure of the quality of a model, and is used to evaluate the ability of a model to predict the correct order of events.

Olivier Grisel - Predictive survival analysis with scikit-learn, scikit-survival and lifelines

More talks