TreeText-CTS: Compact, Source-Traceable Tree-Path Evidence for Irregular Clinical Time-Series Prediction

2026-05-22 · Source: cs.LG updates on arXiv.org · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Data Science & Analytics, Health & Medical Research · Depth: Expert, extended

Summary

TreeText-CTS is a novel framework designed for irregular electronic health record (EHR) time-series prediction, converting complex trajectories into compact, human-readable, and source-traceable tree-path evidence. It routes multi-scale window summaries through frozen XGBoost models, verbalizing activated tree paths as deterministic threshold conditions. An evidence selector thensembles an informative subset, which a language-model encoder integrates for final prediction. Evaluated on PhysioNet 2012 mortality, MIMIC-III mortality, and PhysioNet 2019 sepsis-onset forecasting, TreeText-CTS achieved the best AUROC and AUPRC among text-based EHR interfaces, improving AUPRC by 6.0 to 9.7 absolute percentage points over prior methods, while remaining competitive with numerical time-series models. This approach ensures inspectable and traceable evidence for predictions.

Key takeaway

For Machine Learning Engineers developing clinical prediction models, TreeText-CTS offers a compelling approach to balance accuracy with interpretability. You should consider integrating tree-grounded language interfaces to generate compact, source-traceable evidence for predictions. This enhances auditability and provides human-readable insights, crucial for high-stakes healthcare applications, without sacrificing competitive performance against numerical models.

Key insights

TreeText-CTS bridges numerical and language models to provide compact, source-traceable evidence for clinical time-series predictions.

Principles

• Human-readable, source-traceable evidence enhances clinical prediction.
• Fixed tree models can extract prediction-relevant thresholds from EHR data.
• Compact, deterministic evidence units outperform raw serialization for language models.

Method

Multi-scale EHR window summaries are processed by frozen XGBoost models. Activated tree paths become deterministic predicate text, selected by CES, then composed by an LM encoder for prediction.

In practice

• Implement XGBoost to identify critical thresholds in irregular EHR data.
• Utilize a learned evidence selector for efficient language model input.
• Cache clinical glosses offline to augment deterministic predicates.

Topics

• Clinical Time-Series
• EHR Prediction
• XGBoost
• Language Models
• Interpretability
• Sepsis Forecasting
• Mortality Prediction

Best for: NLP Engineer, AI Scientist, Machine Learning Engineer, Research Scientist

Related on AIssential

• Text Knows What, Tables Know When: Clinical Timeline Reconstruction via Retrieval-Augmented Multimodal Alignment — Multimodal alignment of clinical text and EHR data significantly improves temporal precision in patient timeline reconstruction.
• Language Models as Interfaces, Not Oracles: A Hybrid LLM-ML System for Pediatric Appendicitis — Hybrid AI systems improve clinical decision support by using LLMs for feature extraction and validated ML models for reliable, auditable prediction.
• Before the Labels: How Dataset Construction Shapes Suicidality Detection in Clinical Text — EHR-based suicidality datasets encode specific operationalizations of suicidality, shaped by construction choices, not objective ground truth.
• ChatHealthAI: Aligning Electronic Health Record Representations with Large Language Models for Grounded Clinical Reasoning — ChatHealthAI integrates EHR representations with LLMs for grounded, interpretable clinical reasoning and accurate patient prediction.
• Possible or Definite? A Benchmark for Evaluating Diagnostic Uncertainty Preservation in Clinical Text — Large language models poorly preserve diagnostic uncertainty in clinical text, posing a significant safety risk.
• A Pipeline for Generating Longitudinal Synthetic Clinical Notes Using Large Language Models — A modular LLM-powered pipeline generates privacy-preserving, longitudinal synthetic clinical notes for AI development and evaluation.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by cs.LG updates on arXiv.org.