Automated Classifier Generation for Medical Record Coding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The healthcare industry faces challenges in consolidating and analyzing large amounts of unstructured medical information, leading to unreliability, uncertainty in diagnosis, and increased costs due to the lack of efficient tools for data management and big data platforms.
Innovation Solution
The development of a system and method for automatic classifier generation and classification of health records using labeled datasets, employing techniques like John Lewis Z-score and mutual information maximization to identify predictive features, which are combined into dictionaries for evaluating medical records and improving the identification of code-able events.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If automated classifier generation is implemented, then identification accuracy of code-able events is improved, but system complexity increases
Solution Approach 1:
The system segments the complex classification task into multiple independent classifier dictionaries, each trained on specific labeled datasets for different code-able events. These dictionaries can be generated, stored, and applied separately, reducing the complexity of managing a single large classifier system while maintaining high identification accuracy through specialized models.
Solution Approach 2:
The system performs preliminary action by pre-training and storing multiple classifier dictionaries in advance using labeled datasets. These pre-generated dictionaries are then directly applied during runtime without requiring complex real-time training, simplifying the operational system while ensuring high identification accuracy through pre-optimized classifiers.
2Reliability
If multiple classifier dictionaries are trained and stored, then identification reliability is improved, but data storage requirements increase
Solution Approach 1:
The system extracts and stores only the essential classifier dictionaries that are most relevant to specific code-able events identified in the medical records. By selectively taking out and storing only necessary classifiers rather than all possible ones, the system maintains high identification reliability for critical events while minimizing data storage requirements.
3Measurement precision
If feature selection techniques are applied, then classification precision is improved, but processing time increases
Solution Approach 1:
The system performs feature selection and classifier training as preliminary actions during the offline dictionary generation phase. By completing computationally intensive feature selection and precision optimization in advance, the system achieves high classification precision while keeping runtime processing time minimal when applying the pre-trained dictionaries to actual medical records.
Data Source
AI summary
Systems and methods to automatically generate classifiers are provided. A labeled dataset is initially received. The dataset may be for a positive class, or may be a negative for a class, or a false positive class. N features that are predictive for the class (or false positive or the negative class) are identified. These features are combined within a classifier dictionary. Medical records received may be processed in order to be machine readable. Features within the medical records are identified and are compared against the dictionary of classifiers. Matches indicate classes within the medical record. The classifier dictionary may be periodically updated in response to insufficient classification accuracy, or when new data becomes available.