Laboratory Test History Modeling with Denoise-Balance Filtering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Clinical laboratories face challenges in integrating and interpreting large volumes of laboratory data for accurate diagnostic insights, as human intuition is unreliable and conventional statistical approaches are limited, often missing key results or patterns, leading to inefficient diagnosis and management.
Innovation Solution
A computer-implemented method using a denoise-balance scheme to preprocess datasets, comprising index, test code, and feature filters, followed by training a machine learning model with an ensemble method to predict clinical diagnostic test results, enhancing accuracy and efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If manual interpretation of laboratory test results is used, then clinicians can integrate test results with clinical data and medical knowledge, but key results or patterns are easily overlooked due to human limitations in processing extensive data
Solution Approach 1:
The patent introduces computational approaches as an intermediary between laboratory test results and clinician interpretation. Machine learning models process hundreds or thousands of individual test results, extracting patterns and trends that would be difficult for clinicians to identify manually, while presenting synthesized insights rather than raw data volumes.
Solution Approach 2:
The patent replaces the manual mechanical process of clinician data review with automated computational systems. Machine learning algorithms systematically analyze laboratory data, eliminating human cognitive limitations in processing extensive datasets while maintaining diagnostic relevance through trained prediction models.
2Measurement precision
If computational approaches are used to analyze laboratory data, then diagnostic value is enhanced, but data processing complexity and model training requirements increase
Solution Approach 1:
The patent segments the complex computational task into distinct components: data preprocessing modules that clean and organize laboratory data, feature engineering steps that identify relevant predictors, and machine learning model training phases. This segmentation manages complexity by breaking down the overall system into manageable, specialized subsystems.
Solution Approach 2:
The patent performs preliminary data preprocessing and feature selection before model training. The system pre-processes laboratory data to ensure quality, selects relevant features in advance, and prepares training datasets, thereby reducing the computational burden during actual model training and deployment phases.
3Measurement precision
If machine learning models are trained on extensive historical data, then prediction accuracy improves, but training time and computational resources increase
Solution Approach 1:
The patent extracts and removes irrelevant or redundant features from the extensive historical laboratory data before training. By identifying and eliminating non-informative data elements, the system reduces the dimensionality of training datasets while preserving the signal needed for accurate predictions, thereby decreasing training time without sacrificing accuracy.
Solution Approach 2:
The patent optimizes model training parameters such as learning rates, batch sizes, and epoch counts based on the specific characteristics of laboratory data. These parameter adjustments enable efficient training on extensive historical datasets by balancing convergence speed with prediction accuracy, reducing overall training time while maintaining model performance.
Data Source
AI summary
The present disclosure relates to techniques for preprocessing samples and using preprocessed samples and machine learning models to predict clinical diagnostic tests for a patient from their historical laboratory testing data. Particularly, aspects are directed to obtaining datasets including features and/or historical laboratory test results for subjects, filtering the datasets based on a denoise-balance scheme to obtain filtered datasets, training a machine learning model using the filtered datasets to obtain a trained machine learning model, and providing the trained machine learning model. A candidate machine learning model may be an ensemble of classifiers implemented with a boosting algorithm, and the ensemble is trained by applying base machine learning algorithms on different distributions of the filtered datasets. The ensemble is then combined into a machine learning model having the set of learned model parameters for predicting results for clinical diagnostic tests.