Probabilistic Model for Medical Transcript Analysis

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Automated analysis of medical transcripts is challenging due to their unstructured nature and varying medical terminology, making it difficult to accurately infer patient conditions without manual expertise.

Innovation Solution

A probabilistic model is trained using labeled text passages to identify key words and phrases, constructing a graphical model that associates words with medical concepts, allowing for the inference of patient states based on probabilistic associations.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Extent of automation

If phrase spotting methods are used to search for key terms in medical transcripts, then automated analysis can be performed, but the accuracy of inferring patient conditions deteriorates due to varying medical terminology and unstructured text

Engineering Contradiction:

Improveautomated analysisVSAvoidaccuracy of inferring patient conditions

Core Design Contradiction:

Extent of automationVSMeasurement precision

Solution Approach 1:

A probabilistic model serves as an intermediary between the unstructured medical transcript text and the structured patient condition inference. The model processes the ambiguous, varying terminology in the text and transforms it into accurate condition classifications, resolving the contradiction between automated processing capability and inference accuracy.

Inventive Principle:

Principle #24Intermediary (Mediator)

Solution Approach 2:

The system changes the approach from binary phrase matching to probabilistic parameter estimation. Instead of searching for exact key terms, the model assigns probabilistic weights to different terms and phrases, allowing accurate inference even when terminology varies across different medical contexts and practitioners.

Inventive Principle:

Principle #35Parameter changes

2Measurement precision

If manual expert review is used to analyze medical transcripts, then accuracy of patient condition identification is maintained, but productivity and time efficiency deteriorate

Engineering Contradiction:

Improveaccuracy of patient condition identificationVSAvoidtime efficiency

Core Design Contradiction:

Measurement precisionVSProductivity

Solution Approach 1:

The probabilistic model enables the system to perform self-service analysis of medical transcripts without requiring manual expert intervention. The model independently processes unstructured text, identifies medical concepts, and infers patient conditions with accuracy comparable to expert review, thereby maintaining both precision and productivity.

Inventive Principle:

Principle #25Self-service

3Device complexity

If simple keyword searching is used, then device complexity is reduced, but reliability of medical concept identification deteriorates due to inability to handle implied conditions and contextual variations

Engineering Contradiction:

Improveanalysis system complexityVSAvoidreliability of medical concept identification

Core Design Contradiction:

Device complexityVSReliability

Solution Approach 1:

The system replaces simple mechanical keyword searching with a probabilistic modeling approach. This substitution enables the system to handle implied conditions, contextual variations, and varying terminology while maintaining manageable complexity through statistical methods rather than complex rule-based systems.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Data Source

PatentUS7840511B2Learning or inferring medical concepts from medical transcripts using probabilistic models with words or phrases identification

Publication Date: 2010.11.23 CERNER INNOVATION INC

AI summary

A medical concept is learned about or inferred from a medical transcript. A probabilistic model is trained from medical transcripts. For example, the problem is treated as a graphical model. Discrimitive or generative learning is used to train the probabilistic model. A mutual information criterion can be employed to identify a discrete set of words or phrases to be used in the probabilistic model. The model is based on the types of medical transcripts, focusing on this source of data to output the most probable state of a patient in the medical field or domain. The learned model may be used to infer a state of a medical concept for a patient.