# Learning to Select Source-Traceable Evidence for Language-Model Prediction from Irregular Clinical Time Series

arXiv: https://arxiv.org/abs/2605.20292 · HTML: https://mlsanity.com/paper/2605.20292/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 27 of its 53 references that are in the map.
- First posted on arXiv 19 May 2026.

## Builds on (most cited first, 27 of 27)

- [Attention Is All You Need](https://mlsanity.com/paper/1706.03762/) (2017)
- [A Unified Approach to Interpreting Model Predictions](https://mlsanity.com/paper/1705.07874/) (2017)
- [Recurrent Neural Networks for Multivariate Time Series with Missing Values](https://mlsanity.com/paper/1606.01865/) (2016)
- [Self-critical Sequence Training for Image Captioning](https://mlsanity.com/paper/1612.00563/) (2016)
- [ERASER: A Benchmark to Evaluate Rationalized NLP Models](https://mlsanity.com/paper/1911.03429/) (2019)
- [Multitask learning and benchmarking with clinical time series data](https://mlsanity.com/paper/1703.07771/) (2017)
- [Rationalizing Neural Predictions](https://mlsanity.com/paper/1606.04155/) (2016)
- [Multi-Time Attention Networks for Irregularly Sampled Time Series](https://mlsanity.com/paper/2101.10318/) (2021)
- [Set Functions for Time Series](https://mlsanity.com/paper/1909.12064/) (2019)
- [Learning to Faithfully Rationalize by Construction](https://mlsanity.com/paper/2005.00115/) (2020)
- [Graph-Guided Network for Irregularly Sampled Multivariate Time Series](https://mlsanity.com/paper/2110.05357/) (2021)
- [Self-Supervised Transformer for Sparse and Irregularly Sampled Multivariate Clinical Time-Series](https://mlsanity.com/paper/2107.14293/) (2021)
- [Explaining Time Series Predictions with Dynamic Masks](https://mlsanity.com/paper/2106.05303/) (2021)
- [GraphCare: Enhancing Healthcare Predictions with Personalized Knowledge Graphs](https://mlsanity.com/paper/2305.12788/) (2023)
- [Making Pre-trained Language Models Great on Tabular Prediction](https://mlsanity.com/paper/2403.01841/) (2024)
- [Optimized Feature Generation for Tabular Data via LLMs with Decision Tree Reasoning](https://mlsanity.com/paper/2406.08527/) (2024)
- [Reasoning-Enhanced Healthcare Predictions with Knowledge Graph Community Retrieval](https://mlsanity.com/paper/2410.04585/) (2024)
- [Encoding Time-Series Explanations through Self-Supervised Model Behavior Consistency](https://mlsanity.com/paper/2306.02109/) (2023)
- [TimeCAP: Learning to Contextualize, Augment, and Predict Time Series Events with Large Language Model Agents](https://mlsanity.com/paper/2502.11418/) (2025)
- [ClinicalBench: Can LLMs Beat Traditional ML Models in Clinical Prediction?](https://mlsanity.com/paper/2411.06469/) (2024)
- [BioClinical ModernBERT: A State-of-the-Art Long-Context Encoder for Biomedical and Clinical NLP](https://mlsanity.com/paper/2506.10896/) (2025)
- [Prompting Large Language Models for Zero-Shot Clinical Prediction with Structured Longitudinal Electronic Health Record Data](https://mlsanity.com/paper/2402.01713/) (2024)
- [DuETT: Dual Event Time Transformer for Electronic Health Records](https://mlsanity.com/paper/2304.13017/) (2023)
- [General-Purpose Retrieval-Enhanced Medical Prediction Model Using Near-Infinite History](https://mlsanity.com/paper/2310.20204/) (2023)
- [When Raw Data Prevails: Are Large Language Model Embeddings Effective in Numerical Data Representation for Medical Machine Learning Applications?](https://mlsanity.com/paper/2408.11854/) (2024)
- [LLM Meeting Decision Trees on Tabular Data](https://mlsanity.com/paper/2505.17918/) (2025)
- [Can we generate portable representations for clinical time series data using LLMs?](https://mlsanity.com/paper/2603.23987/) (2026)

## Abstract (arXiv)

Numerical time-series models effectively process irregular electronic health record (EHR) trajectories, but do not expose which temporal patterns support each prediction as readable evidence. Existing text-based interfaces either serialize observations, preserving source traceability but offering limited clinical interpretation, or generate patient-level summaries that improve readability but can obscure links to source measurements. We introduce STEP-CTS (Source-Traceable Evidence for Prediction from Clinical Time-Series), which learns to select source-traceable text evidence for a language-model predictor. Multi-scale window statistics of each trajectory are verbalized as sets of deterministic threshold predicates, each set linked to its source window (e.g., "[5-7 h]: last temperature at least 38C"). An offline LLM, run once per unique predicate set without access to patient records, the prediction task, or outcome labels, attaches clinical concepts such as "fever" to their supporting predicates and abstains when none applies. Each predicate set, together with its concepts, forms an evidence unit. A learned Evidence Selector selects a fixed-size subset of the evidence units, which a pretrained clinical language-model encoder reads to make the prediction, with no patient-level text generation. Across three ICU benchmarks, STEP-CTS outperforms evaluated text-based baselines, improving AUPRC over the strongest by 5.1, 3.7, and 11.7 percentage points on P2012, MIMIC-III, and P2019, respectively, and is competitive with dedicated numerical time-series models. Ablations show that clinical concepts and learned evidence selection each contribute to predictive performance. In a blinded clinician study, the selected evidence is rated as traceable as deterministic serialization, near the ceiling of the scale, and above generated summaries on clinical interpretation.