A ReAct Agentic AI System for Natural Language Querying and Statistical Analysis of The Cancer Genome Atlas Clinical Data
- 1Northeastern University
Abstract
The Cancer Genome Atlas (TCGA) holds clinical data for over 11,000 patients across 33 cancer types, but access is hard because of complex file structures, heterogeneous formats, and the need for programming. We present an agentic system for natural language querying and statistical analysis of TCGA clinical data. The system uses a large language model (LLM) as an autonomous Reasoning and Acting (ReAct) agent that selects from eight computational tools, including data extraction, descriptive statistics, Kaplan-Meier survival analysis with log-rank tests, hypothesis testing, and verification against the curated TCGA Pan-Cancer Clinical Data Resource (CDR). The agent reasons about intermediate results, adapts its approach, and returns clinically contextualized responses with source attribution and auditable traces. We introduce TCGA-Agent-Bench, 440 queries across five difficulty tiers with ground truth from the independently curated CDR, evaluated with dual metrics for numerical accuracy and clinical completeness. The system achieves 93.4% overall accuracy, outperforming a fixed rule-based pipeline (87.1%), a single-pass LLM (81.8%), and retrieval-augmented generation (66.9% on a 50-query subset). Most of the benchmark is answerable from the CDR alone, so we locate the extraction layers value in fields the CDR lacks, such as drug treatments, TNM components, and biomarkers: on 26 queries targeting these, the full system answers 100% versus 3.8% for a CDR-only configuration. A tool-based agentic architecture enables accurate, auditable natural language analysis of clinical repositories, with value driven by tool design and recovered fields rather than model scale.
Related
- Beyond Billing Codes: An Agentic Large Language Model Pipeline for Extracting Glaucoma Procedural Data from Free-Text Clinical Notes
- Predicting One-Year Psychiatric Rehospitalization Using Natural Language Processing and Machine Learning Methods Based on Clinical Texts
- GLLaucoMed: A Secure Large Language Model-Powered Agentic Workflow for Automated Medication Extraction from Free-Text Glaucoma Clinical Notes.
- Extracting Symptoms of Psychotic Disorders from Clinical Notes using Natural Language Processing.
- Lung cancer research: how natural language processing enhances clinical data extraction
BibTeX
@misc{korutla2026react,
title = {{A ReAct Agentic AI System for Natural Language Querying and Statistical Analysis of The Cancer Genome Atlas Clinical Data}},
author = {R. Korutla and S. Amal},
howpublished = {medRxiv},
year = {2026},
month = jul,
doi = {10.64898/2026.07.15.26358188},
url = {https://www.medrxiv.org/content/10.64898/2026.07.15.26358188v2},
note = {CC BY},
}
RIS
TY - UNPB
TI - A ReAct Agentic AI System for Natural Language Querying and Statistical Analysis of The Cancer Genome Atlas Clinical Data
AU - R. Korutla
AU - S. Amal
PY - 2026
DA - 2026/07/15
PB - medRxiv
DO - 10.64898/2026.07.15.26358188
UR - https://www.medrxiv.org/content/10.64898/2026.07.15.26358188v2
AB - The Cancer Genome Atlas (TCGA) holds clinical data for over 11,000 patients across 33 cancer types, but access is hard because of complex file structures, heterogeneous formats, and the need for programming. We present an agentic system for natural language querying and statistical analysis of TCGA clinical data. The system uses a large language model (LLM) as an autonomous Reasoning and Acting (ReAct) agent that selects from eight computational tools, including data extraction, descriptive statistics, Kaplan-Meier survival analysis with log-rank tests, hypothesis testing, and verification against the curated TCGA Pan-Cancer Clinical Data Resource (CDR). The agent reasons about intermediate results, adapts its approach, and returns clinically contextualized responses with source attribution and auditable traces. We introduce TCGA-Agent-Bench, 440 queries across five difficulty tiers with ground truth from the independently curated CDR, evaluated with dual metrics for numerical accuracy and clinical completeness. The system achieves 93.4% overall accuracy, outperforming a fixed rule-based pipeline (87.1%), a single-pass LLM (81.8%), and retrieval-augmented generation (66.9% on a 50-query subset). Most of the benchmark is answerable from the CDR alone, so we locate the extraction layers value in fields the CDR lacks, such as drug treatments, TNM components, and biomarkers: on 26 queries targeting these, the full system answers 100% versus 3.8% for a CDR-only configuration. A tool-based agentic architecture enables accurate, auditable natural language analysis of clinical repositories, with value driven by tool design and recovered fields rather than model scale.
ER -