Automated Diagnosis Code Selection Based On Clinical Notes

Technology Description

Unmet Need: U.S. medical coding market size was valued at USD 18.2 billion in 2022 and is expected to grow at a compound annual growth rate (CAGR) of 9.85% from 2023 to 2030.

Product: ClinicalCLIP is a cutting-edge artificial intelligence model to predict diagnosis codes. This novel technology leverages natural language supervision by pairing diagnosis codes with their text descriptions to train clinical note encoders, enabling the model to predict diagnosis codes from unstructured clinical notes with exceptional accuracy.

Stage: Currently in proof-of-concept development, ClinicalCLIP has demonstrated an impressive 92.71% accuracy in automating diagnosis code selection—marking a significant improvement over existing AI models like ClinicalBERT.

Competitive Advantage: What sets ClinicalCLIP apart is its innovative approach to training, which eliminates the need for manual data labeling and supports zero-shot transfer, allowing the model to adapt to new tasks with minimal additional training. By addressing documentation inefficiencies, ClinicalCLIP has the potential to reduce provider workload, combat burnout, and restore valuable patient-doctor interactions while optimizing healthcare delivery. This groundbreaking technology represents a pivotal step in advancing AI applications in medicine.

Patent Status

Patent Pending