mlsanityData to 7 Oct 2026

Multi-Large Language Model Orchestrated Severity Assessment of Clinical Records (MOSAIC)

arXiv 2607.05032 ↗First posted 6 Jul 2026cs.CLcited by 0

Not on the map. Why?

Abstract (arXiv)

Background: Disease severity is a multidimensional construct difficult to capture with rule-based approaches in Electronic Healthcare Records (EHR). Agentic large language model (LLM) systems could synthesise clinical evidence and reason over EHRs, but remain unevaluated for this task. Methods: MOSAIC is a two-phase agentic LLM framework for severity phenotyping, using type 2 diabetes (T2D) as a proof-of-concept. MOSAIC was evaluated on a synthetic cohort (SyntheticMass; open-weight N = 4,886; closed-weight N = 200) against three algorithmic ground truths (DCSI, DiSSCo, Cooper) and against all-cause mortality and incident complications. Open-weight (locally deployable) and proprietary pipelines were also compared. Results: The generated framework spanned domains absent from the comparators, including biomarker-based glycaemic staging, beta-cell function, and social determinants of health. Open-weight MOSAIC matched the proprietary pipeline (closed- vs open-weight weighted kappa = 0.773) and reached moderate agreement with Cooper (kappa = 0.597) and DCSI (kappa = 0.534) and fair agreement with DiSSCo (kappa = 0.320). Agent-based (Type 1) tiers showed significant separation of all-cause mortality (log-rank p < 0.001; crude hazard ratios 1.6-2.4 for non-Baseline tiers), with non-monotonic separation at the upper tiers, and an inverse gradient for incident complications (log-rank p < 0.001) consistent with depletion of susceptibles. Agentic classification also diverged from deterministic execution of the same rubric (MOSAIC Frozen; kappa = 0.428), indicating reasoning beyond fixed rules. Conclusion: MOSAIC shows agentic LLM systems can generate and apply clinically meaningful severity phenotypes from structured EHR data in T2D. Extending it to other diseases with similarly multidimensional severity warrants further research.

Key facts. No citing papers in the map yet (data through 7 Oct 2026). Builds on 50 of its 112 references that are in the map. First posted on arXiv 6 Jul 2026.

Builds on

50 of 112 references.

References · 50

  1. 2023Large Language Models for Software Engineering: A Systematic Literature Reviewcited by 389
  2. 2023Software Testing with Large Language Models: Survey, Landscape, and Visioncited by 163
  3. 2024SWT-Bench: Testing and Validating Real-World Bug-Fixes with Code Agentscited by 129
  4. 2021TOGA: A Neural Method for Test Oracle Generationcited by 57
  5. 2024Enchanting Program Specification Synthesis by Large Language Models using Static Analysis and Program Verificationcited by 55
  6. 2023Can Large Language Models Transform Natural Language Intent into Formal Method Postconditions?cited by 53
  7. 2023A3Test: Assertion-Augmented Automated Test Case Generationcited by 33
  8. 2023Nuances are the Key: Unlocking ChatGPT to Find Failure-Inducing Tests with Differential Promptingcited by 23
  9. 2024Large Language Models and Video Games: A Preliminary Scoping Reviewcited by 23
  10. 2024TOGLL: Correct and Strong Test Oracle Generation with LLMscited by 23
  11. 2024A Large-scale Empirical Study on Fine-tuning Large Language Models for Unit Testingcited by 15
  12. 2024LLM-Powered Test Case Generation for Detecting Bugs in Plausible Programscited by 14
38 more
  1. 2024Test Oracle Automation in the era of LLMscited by 11
  2. 2025ClassInvGen: Class Invariant Synthesis using Large Language Modelscited by 11
  3. 2024Doc2OracLL: Investigating the Impact of Documentation on LLM-based Test Oracle Generationcited by 10
  4. 2023Neural-Based Test Oracle Generation: A Large-scale Evaluation and Lessons Learnedcited by 9
  5. 2023Towards More Realistic Evaluation for Neural Test Oracle Generationcited by 8
  6. 2025Hallucination to Consensus: Multi-Agent LLMs for End-to-End JUnit Test Generationcited by 8
  7. 2023Assessing Evaluation Metrics for Neural Test Oracle Generationcited by 7
  8. 2024Towards Generating Executable Metamorphic Relations Using Large Language Modelscited by 7
  9. 2025AsserT5: Test Assertion Generation Using a Fine-Tuned Code Language Modelcited by 7
  10. 2025A Tale of 1001 LoC: Potential Runtime Error-Guided Specification Synthesis for Verifying Large-Scale Programscited by 6
  11. 2025Are Autonomous Web Agents Good Testers?cited by 5
  12. 2024Enhancing Differential Testing With LLMs For Testing Deep Learning Librariescited by 4
  13. 2025AugmenTest: Enhancing Tests with LLM-Driven Oraclescited by 4
  14. 2025Improving Deep Assertion Generation via Fine-Tuning Retrieval-Augmented Pre-trained Language Modelscited by 4
  15. 2026WebTestPilot: Agentic End-to-End Web Testing against Natural Language Specification by Inferring Oracles with Symbolized GUI Elementscited by 4
  16. 2024Toward Automated Validation of Language Model Synthesized Test Cases using Semantic Entropycited by 3
  17. 2025METAMON: Finding Inconsistencies between Program Documentation and Behavior using Metamorphic LLM Queriescited by 3
  18. 2025Acceptance Test Generation with Large Language Models: An Industrial Case Studycited by 3
  19. 2025AI-Driven Tools in Modern Software Quality Assurance: An Assessment of Benefits, Challenges, and Future Directionscited by 3
  20. 2025Generative AI for Testing of Autonomous Driving Systems: A Surveycited by 3
  21. 2024A Study of Using Multimodal LLMs for Non-Crash Functional Bug Detection in Android Appscited by 2
  22. 2025Improving Retrieval-Augmented Deep Assertion Generation via Joint Trainingcited by 2
  23. 2025Tratto: A Neuro-Symbolic Approach to Deriving Axiomatic Test Oraclescited by 2
  24. 2025Automated Testing of the GUI of a Real-Life Engineering Software using Large Language Modelscited by 2
  25. 2025Automated Discovery of Test Oracles for Database Management Systems Using LLMscited by 2
  26. 2026Understanding LLM-Driven Test Oracle Generationcited by 2
  27. 2026Consistency Meets Verification: Enhancing Test Generation Quality in Large Language Models Without Ground-Truth Solutionscited by 2
  28. 2026SpecMind: Cognitively Inspired, Interactive Multi-Turn Framework for Postcondition Inferencecited by 2
  29. 2026CASCADE: Detecting Inconsistencies between Code and Documentation with Automatic Test Generationcited by 2
  30. 2023SAGA: Summarization-Guided Assert Statement Generationcited by 1
  31. 2025LLM-based Property-based Test Generation for Guardrailing Cyber-Physical Systemscited by 1
  32. 2025Private GPTs for LLM-driven testing in software development and machine learningcited by 1
  33. 2025SATORI: Static Test Oracle Generation for REST APIscited by 1
  34. 2025Understanding the Characteristics of LLM-Generated Property-Based Tests in Exploring Edge Casescited by 1
  35. 2026Just-in-Time Catching Test Generation at Metacited by 1
  36. 2026MR-Coupler: Automated Metamorphic Test Generation via Functional Coupling Analysiscited by 1
  37. 2026Improving Dynamic Specification Inference with LLM-Generated Counterexamplescited by 1
  38. 2026POSTCONDBENCH: Benchmarking Correctness and Completeness in Formal Postcondition Inferencecited by 1

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2025Exploring Large Language Models in Healthcare: Insights into Corpora Sources, Customization Strategies, and Evaluation Metrics
  2. 2026DIASENTINEL: An Auditable Multi-Agent System for Guideline-Grounded Diabetes Risk Screening
  3. 2026A primer on evaluation methods for large language models in healthcare
  4. 2025Evaluating Hierarchical Clinical Document Classification Using Reasoning-Based LLMs
  5. 2024Exploring LLM Multi-Agents for ICD Coding
  6. 2024Towards Automatic Evaluation for LLMs' Clinical Capabilities: Metric, Data, and Algorithm
  7. 2025BRIDGE: Benchmarking Large Language Models for Understanding Real-world Clinical Practice Text
  8. 2026First, Do No Harm (With LLMs): Mitigating Racial Bias via Agentic Workflows