CACSurv: Concordance-Aligned Comparative Learning with Large Language Models for Cancer Survival Prediction
2026-08-17 • Artificial Intelligence
Artificial Intelligence
AI summaryⓘ
The authors focus on predicting cancer patient survival using detailed medical reports, which include pathology and clinical information. They found existing methods struggle because traditional approaches predict exact survival times, which doesn't work well when some patients' outcomes are unknown (censored). To fix this, the authors created CACSurv, a new method that compares patients to each other rather than predicting exact times, improving how censored data is handled. Their approach uses a special reward system to train large language models to rank patients by survival risk accurately and outperforms previous models on a new cancer survival dataset.
Cancer survival predictionLarge language models (LLMs)Censored dataTime-to-event regressionConcordance index (C-index)Right censoringComparative reasoningTCGA (The Cancer Genome Atlas)Survival analysisRanking methods
Authors
Tianqi Xiang, Qixiang Zhang, Xinpeng Ding, Yi Li, Xiaomeng Li
Abstract
Cancer survival prediction supports treatment planning, risk stratification, and follow-up management. Existing methods use structured clinical variables, whole-slide images, genomic profiles, or multimodal inputs, while patient reports remain underexplored. We study report-centric survival prediction using reports that organize pathological, clinical, and molecular evidence. Large language models (LLMs) can reason over such reports, but case-wise time regression introduces two mismatches. First, a formulation mismatch arises because survival evaluation depends on ordering comparable patients, whereas independent time predictions do not enforce ranking consistency. Second, a supervision mismatch arises because a censored patient's observed time indicates survival beyond that point and cannot serve as an exact regression target, although it still implies orderings relative to patients who died earlier. To address these mismatches, we propose CACSurv, a Concordance-Aligned Comparative framework for report-centric survival prediction. CACSurv reformulates survival modeling as mini-cohort comparative reasoning, where an LLM predicts relative prognostic orderings. We introduce concordance-aligned rewards derived from comparable relations under right censoring, enabling censored outcomes to provide ranking supervision without exact event-time targets. At inference, Monte Carlo Reference Aggregation compares each patient with sampled references and aggregates positions into a cohort-level ranking. We establish TCGA-SurvReport, a benchmark covering six TCGA cancer cohorts. CACSurv achieves the highest C-index on all six cohorts and an average C-index of 0.722, outperforming the strongest published survival model by 6.5 percentage points and the strongest LLM time-regression baseline by 4.2 percentage points. Our code, models, and dataset will be available at https://github.com/xmed-lab/CACSurv.