Clinical Speech Processing Service for Intent Recognition
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current natural language processing (NLP) technologies struggle to accurately interpret clinical voice inputs due to the vast and complex medical vocabulary, lacking integration with electronic health records (EHRs) and failing to capture the semantic meaning of medical terms, making it difficult to provide accurate and structured data for patient care.
Innovation Solution
A speech processing and mapping service (SPMS) that coordinates with NLP engines, semantic mappers, and EHR systems to identify intents and map voice inputs to standard terminologies like SNOMED CT and LOINC, enabling the extraction of relevant patient information and providing actionable insights.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If manual addition of medical terms is used to train NLP models, then the model can be trained with clinical terminology, but the process is extremely difficult and time-consuming
Solution Approach 1:
The system performs preliminary action by pre-processing and mapping medical terms from various sources (EHR systems, terminologies like SNOMED CT, LOINC) before they are needed for NLP training. The terminology management system pre-organizes medical terms, relationships, and verbatim phrases in a structured format that can be directly imported into NLP models, eliminating the need for manual term addition during training.
Solution Approach 2:
The patent introduces an intermediary terminology management system that acts as a mediator between medical terminology sources and NLP models. This intermediary layer handles the complex task of mapping, normalizing, and storing medical terms in a format suitable for NLP training, thereby separating the terminology management complexity from the NLP model training process.
2Reliability
If all medical terms are loaded into NLP models to ensure accurate intent recognition, then the model becomes more accurate, but the complexity of managing and loading these terms increases
Solution Approach 1:
The system segments the terminology management function into a separate, dedicated terminology management system that operates independently from the NLP model. This segmentation allows the complex task of managing hundreds of thousands of medical terms to be handled by a specialized system with structured data stores and mapping mechanisms, rather than embedding this complexity within the NLP model itself.
Solution Approach 2:
The terminology management system serves multiple functions: it stores medical terms from various sources (SNOMED CT, LOINC, EHR systems), maps between different terminology systems, provides verbatim phrases for NLP training, and enables bulk loading of terminology data. This multi-functionality consolidates what would otherwise be multiple separate systems into a single unified terminology management platform.
3Reliability
If medical terms are manually loaded into NLP models, then the model can process clinical terminology, but the terms cannot be reused across different NLP models for different applications
Solution Approach 1:
The terminology management system is designed as a universal platform that serves multiple NLP models and applications. It maintains a centralized repository of medical terms and mappings that can be exported to any NLP model, enabling terminology reuse across different applications without requiring manual re-loading for each model or application.
Solution Approach 2:
The system enables copying of terminology data from the centralized management system to multiple NLP models and applications. Instead of manually adding terms to each model, the terminology can be bulk copied or exported in various formats, allowing the same terminology foundation to support multiple different NLP applications simultaneously.
Data Source
AI summary
Methods and systems for processing of voice input to identify intents and mapped standard terminologies are provided. Using natural language processing, an intent of a voice input is identified. The intent is utilized to identify a standard terminology that maps to the intent. The standard terminology is utilized to identify information relevant to the standard terminology in a patient's electronic health record.