Abstract
Clinical decision support systems using large language models can face challenges in reliability and explainability, particularly in adhering to validated clinical guidelines when processing unstructured and incomplete patient electronic health records. A described system can use an agentic pipeline to structure a patient's health data into a temporal, sequential history. This patient history can then be compared against a knowledge graph of predefined, multi-step clinical pathways using a non-contradiction matching algorithm, which considers a pathway a potential match if no patient data point directly contradicts a guideline condition. The system can then generate a ranked list of probable clinical pathways, recommend next actions, and identify data gaps. This approach is designed to improve the likelihood of providing explainable and guideline-adherent decision support for medical professionals when working with complex patient information.
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.
Recommended Citation
Kuligin, Leonid and Tschochohei, Max, "The Agentic Clinical Decision Support via Non-Contradiction Matching Against Temporal Pathways", Technical Disclosure Commons, ()
https://www.tdcommons.org/dpubs_series/11770