Machine Learning Albuminuria Screening from Routine Patient Data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing albuminuria screening methods, relying on urine samples, fail to detect the condition in many patients, leading to delayed diagnosis and progression of chronic kidney disease (CKD), as albuminuria testing is not commonly ordered for patients without diabetes or other kidney-related conditions.
Innovation Solution
A machine learning model trained on patient demographics, vital signs, and blood tests, utilizing gradient boosting technology, predicts urine albumin-to-creatinine ratio (UACR) levels to identify patients with undiagnosed albuminuria, enabling early detection and intervention.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If urine albumin testing is performed using traditional methods, then albuminuria can be detected with high accuracy, but the screening coverage is limited and many patients remain undiagnosed
Solution Approach 1:
The system performs preliminary identification of patients at risk for albuminuria using machine learning models that analyze electronic health record data before actual urine testing. This preliminary screening identifies high-risk patients who should be prioritized for urine albumin testing, thereby expanding screening coverage while maintaining diagnostic accuracy.
Solution Approach 2:
The patent introduces an intermediary machine learning prediction system that bridges the gap between limited urine testing resources and the need for widespread albuminuria screening. The ML model acts as a mediator by predicting which patients are likely to have albuminuria based on available EHR data, enabling targeted testing expansion without requiring universal urine screening.
2Productivity
If urine albumin testing is expanded to all patients, then screening coverage increases, but the complexity of the screening system and resource requirements increase significantly
Solution Approach 1:
The system performs preliminary risk stratification using machine learning models that analyze electronic health record data before actual urine testing. This preliminary screening identifies high-risk patients who should be prioritized for urine albumin testing, thereby expanding screening coverage while maintaining diagnostic accuracy.
Solution Approach 2:
The machine learning model automatically analyzes existing electronic health record data to identify patients at risk for albuminuria without requiring additional manual assessment or complex screening infrastructure. The system leverages already-collected patient data (demographics, vital signs, blood tests) to perform self-service risk identification.
3Productivity
If machine learning prediction is used to identify at-risk patients, then screening efficiency improves and resources are optimized, but false predictions may lead to missed diagnoses or unnecessary testing
Solution Approach 1:
The system implements feedback loops where urine albumin test results are fed back into the machine learning model to continuously refine and improve prediction accuracy. This allows the system to learn from actual test outcomes and adjust its risk predictions, reducing false positives and false negatives over time while maintaining high screening efficiency.
Solution Approach 2:
The system performs preliminary identification of patients at risk for albuminuria using machine learning models that analyze electronic health record data before actual urine testing. This preliminary screening identifies high-risk patients who should be prioritized for urine albumin testing, thereby expanding screening coverage while maintaining diagnostic accuracy.
Data Source
AI summary
An example embodiment may involve obtaining, by a computing system, an observation of demographic values of an individual, vital sign values of the individual, and blood test values of the individual: applying, by the computing system, a machine learning model to the observation, wherein the machine learning model was trained with a training data set, wherein the training data set contained observations of corresponding demographic values, vital sign values, blood test values, and either urine albumin-to-creatinine ratio (UACR) values or urine protein-to-creatinine ratio (UPR) values for a plurality of individuals, and wherein the machine learning model is configured to provide predictions of whether further observations are indicative of undiagnosed albuminuria or proteinuria; and providing, by the computing system, a prediction of whether the individual exhibits undiagnosed albuminuria or proteinuria based on the observation.