Parallel Medical Record ML Models for Faster Information Extraction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing machine learning systems struggle to accurately and efficiently process medical records, failing to identify a wide variety of patterns such as comorbidity terms, ICD codes, body part information, and prescription information, and require significant computational time and complexity.
Innovation Solution
A multi-threaded machine learning system that executes multiple models in parallel using dedicated computational threads, employing combined convolutional neural networks and long short-term models, and ensemble learning to process nursing and other medical records, extracting relevant information including service provider, medical specializations, and dates of service.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If existing machine learning systems process medical records using traditional single-threaded approaches, then they can identify some information patterns, but the processing time is excessive and computational complexity is high
Solution Approach 1:
The patent segments the medical record processing task into multiple independent machine learning models, each dedicated to identifying specific information patterns (e.g., comorbidity terms, ICD codes, body part information, prescription information). These segmented models execute in parallel threads, transforming a single sequential processing bottleneck into multiple concurrent processing streams, thereby significantly improving information identification speed while reducing overall computational time.
Solution Approach 2:
The patent transitions from traditional single-threaded sequential processing to a multi-dimensional parallel processing architecture. By organizing multiple machine learning models across multiple computational threads running simultaneously, the system adds a temporal parallelism dimension to the processing workflow, enabling information identification to occur across multiple time slices rather than a single sequential timeline, thus dramatically reducing computational time.
2Adaptability or versatility
If existing machine learning systems use simple processing models, then computational time is reduced, but they fail to identify a wide variety of patterns in medical data
Solution Approach 1:
The patent creates a universal machine learning processing framework where multiple specialized models (comorbidity model, ICD model, body parts model, prescription model, provider name model) coexist within a single system. Each model is optimized for a specific information pattern but all operate under the same parallel processing architecture and data pipeline, enabling the system to handle diverse medical data patterns simultaneously without requiring separate processing systems for each pattern type.
Solution Approach 2:
The patent merges multiple specialized machine learning models into a unified parallel processing system. By combining comorbidity detection, ICD code identification, body part recognition, prescription analysis, and provider name extraction into a single integrated system that executes models concurrently, the patent achieves comprehensive pattern identification capability while managing system complexity through shared infrastructure and coordinated execution.
3Productivity
If existing machine learning systems process medical records sequentially, then model accuracy can be maintained, but processing efficiency is severely limited
Solution Approach 1:
The patent implements dynamic model execution where the selection and configuration of machine learning models adapt based on the specific medical record being processed. The system dynamically determines which models to activate and in what sequence, allowing flexible optimization of processing paths while maintaining accuracy. This dynamic approach enables the system to process different record types through appropriate model combinations rather than rigid sequential execution.
Solution Approach 2:
The patent incorporates feedback mechanisms where the output of one machine learning model can influence the execution of subsequent models. The system uses feedback from intermediate processing results to adjust processing parameters, refine predictions, and coordinate between parallel models, thereby maintaining information extraction accuracy while enabling concurrent processing. This feedback loop ensures that parallel execution does not compromise the reliability of the final extracted information.
Data Source
AI summary
Systems and methods for machine learning of medical records are provided. The system can execute multiple machine learning models on the medical records in parallel using multi-threaded approach wherein each machine learning model executes using its own, dedicated computational thread in order to significantly speed up the time with which relevant information can be identified from documents by the system. The multi-threaded machine learning models can include, but are not limited to, sentence classification models, comorbidity models, ICD models, body parts models, prescription models, and provider name models. The system can also utilize combined convolutional neural networks and long short-term models (CNN+LSTMs) as well as ensemble machine learning models to categorize sentences in medical records. The system can also extract service provider, medical specializations, and dates of service information from medical records.