Two-sided conformalized survival analysis

2024-10-31

AI Summary

This paper addresses individualized survival time prediction for right-censored data. We propose the first conformal prediction-based method for constructing prediction intervals that provides rigorous finite-sample coverage guarantees under the sole assumption of independent and identically distributed (i.i.d.) observations. The method adaptively distinguishes individual similarity: it outputs two-sided intervals for highly similar individuals and degenerates to one-sided lower bounds for dissimilar ones. Coverage accuracy is achieved via similarity-weighted truncation and nonparametric calibration. Unlike Cox regression or random forest quantile methods, our approach imposes no distributional assumptions. Empirical evaluation on both synthetic and real-world datasets demonstrates that it consistently attains nominal coverage levels (e.g., ≥89.7% observed at the 90% target), substantially improving statistical reliability and practical utility.

Technology Category

• Machine Learning: Calibration & Uncertainty Quantification
• Reasoning under Uncertainty: Causality
• Data Mining & Knowledge Management: Conversational Systems for Recommendation & Retrieval

Application Category

• User Modeling, Personalization and Recommendation: User privacy protection in personalized systems
• Security and Privacy: Large-scale security measurements
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

This paper presents a novel method using conformal prediction to generate two-sided or one-sided prediction intervals for survival times. Specifically, the method provides both lower and upper predictive bounds for individuals deemed sufficiently similar to the non-censored population, while returning only a lower bound for others. The prediction intervals offer finite-sample coverage guarantees, requiring no distributional assumptions other than the sampled data points are independent and identically distributed. The performance of the procedure is assessed using both synthetic and real-world datasets.

Problem

• Predict survival times with two-sided intervals
• Handle right censoring in survival data
• Ensure finite-sample coverage guarantees

Innovation

• Two-sided conformal prediction for survival times
• Finite-sample coverage guarantees without distributional assumptions
• Handles right censoring with similarity-based predictive bounds