{"id":"CU26327","slug":"ai-software-for-accurate--CU26327","source":{"id":"CU26327","dataset":"techtransfer","title":"AI software for accurate translation of clinical text to standardized medical codes","description_":"<p>This technology is a neural-symbolic AI system that translates ambiguous clinical text and abbreviations into standardized medical codes by combining language model reasoning with deterministic terminology matching to enable accurate data standardization. </p>\r\r<h2>Unmet Need: Reliable standardization of clinical text for health data interoperability</h2>\r\r<p>Clinical records and biomedical literature are full of ambiguous abbreviations and non-standard terms. For example, a shorthand like “RA” can mean rheumatoid arthritis, right atrium, or renal artery depending on context. Linking these free texts to standard codes is foundational for EHR interoperability, clinical trial matching, and health analytics. However, existing tools force a tradeoff between rule-based systems that handle common terms but break on abbreviations and rare concepts, while AI models handle ambiguity better but fail unpredictably on uncommon terminology and are not interpretable. Healthcare organizations need a solution that handles the full range of clinical languages while remaining accurate, scalable, and transparent enough for regulated clinical workflows. </p>\r\r<h2>The Technology: Autonomous AI for accurate medical coding from any clinical text</h2>\r\r<p>This technology is a neural-symbolic AI system called Medical Concept Mapping (MCM) that links free-text clinical mentions to standard terminology codes. Unlike a static pipeline that processes every term the same way, MCM operates as an AI agent. Specifically, it evaluates each incoming term, selects the appropriate processing path, executes it, and self-verifies the results before returning a code. When a term is clear, MCM uses fast, deterministic lookup against standard vocabularies. When a term is ambiguous, MCM calls a large language model to interpret the term in context, verify the interpretation preserves the original meaning, and then map the clarified term to the correct code. This selective, decision-driven approach keeps the system efficient while maintaining a full, auditable record of every step.</p>\r\r<p>In benchmarking across three public datasets, MCM outperformed leading tools by up to 24.8 points on ambiguous abbreviations, with independent reviewers validating 79.4% of AI-generated interpretations as correct.</p>\r\r<h2>Applications:</h2>\r\r<ul>\r<li>EHR data standardizations and harmonization across health systems</li>\r<li>Clinical NLP pipelines for concept extraction and coding</li>\r<li>Clinical trial eligibility and screening for cohort identification</li>\r<li>Evidence synthesis and systematic review automation</li>\r<li>Knowledge graph construction from clinical and research text</li>\r<li>Pharmacovigilance and adverse event detection from clinical narratives</li>\r<li>Research tool for benchmarking concept normalization systems</li>\r</ul>\r\r<h2>Advantages:</h2>\r\r<ul>\r<li>Operates as an autonomous agent, eliminating manual review</li>\r<li>Provides full decision auditability, meeting transparency requirements of regulated clinical workflows</li>\r<li>Utilizes rule based and AI based approach to tackle the problem </li>\r<li>Invokes AI selectively, reducing computational cost compared to end-to-end language model approaches</li>\r<li>Integrates into existing clinical NLP pipelines without infrastructure changes</li>\r</ul>\r\r<h2>Lead Inventor:</h2>\r\r<p><a href=\"https://www.dbmi.columbia.edu/profile/chunhua-weng/\">Chunhua Weng, Ph.D, FACMI, FIASHI</a></p>\r\r<h2>Related Publications:</h2>\r\r<ul>\r<li><a href=\"https://pubmed.ncbi.nlm.nih.gov/41935204/\">Zhang G, Fang Y, Chen F, Ta C, Hripcsak G, Ryan P, Peng Y, Weng C. “A neural-symbolic AI agent system for biomedical concept mapping” NPJ Digital Med. 2026 Apr. 4; 9(425).</a></li>\r</ul>\r\r<h2>Tech Ventures Reference:</h2>\r\r<ul>\r<li>Licensing Contact: <a href=\"mailto:techtransfer@columbia.edu\">Joan Martinez</a> </li>\r</ul>\r\r","tags":["Artificial intelligence","Audit","Automation","Benchmarking","Clinical trial","Concept map","High-throughput screening","Interoperability","Knowledge graph","Language model","Medical classification","Pharmacovigilance","Renal artery","Rheumatoid arthritis","Shorthand","Systematic review"],"file_number":"CU26327","collections":[],"meta_description":"Neural-symbolic AI maps ambiguous clinical text to standardized codes with auditability, efficiency, and context-aware accuracy.","apriori_judge_output":"{\"scores\":{\"novelty\":4.0,\"potential_impact\":4.0,\"readiness\":4.0,\"scalability\":3.0,\"timeliness\":4.0},\"weighted_score\":3.9,\"risks\":[\"Regulatory/regulatory approval uncertainty in regulated healthcare workflows\",\"Potential ambiguity in performance across diverse clinical corpora\",\"Dependence on LLMs may raise safety/compliance concerns\",\"Competition from established medical coding vendors\"],\"one_sentence_take\":\"Strong novelty and potential impact with solid readiness, but moderate scalability and some regulatory/edge-case risks.\"}","inventors":["Chunhua Weng","Gongbo Zhang"],"manager":"Joan Martinez","depts":["Biomedical Informatics"],"divs":["Columbia University Medical Center (CUMC)"],"date_released":"2026-07-24"},"highlight":{},"matched_queries":null,"score":0.0}