Sepsis Onset Detection Using Sparse Emergency Department Data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods fail to accurately stratify patients requiring ICU treatment from those suitable for emergency department care due to limited patient health information in emergency settings, leading to increased morbidity and mortality.
Innovation Solution
A method using machine learning models, specifically XGBoost, to determine sepsis onset based on limited medical parameters, including laboratory parameters like blood urea nitrogen, creatinine, and CRP, without relying on continuous data, and providing risk scores for sepsis detection.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If clinical criteria are used for sepsis recognition, then sepsis detection capability is improved, but accuracy of patient stratification deteriorates due to limited patient health information in emergency departments
Solution Approach 1:
The patent introduces an intermediary system comprising a processor that receives medical parameters from emergency department sources, determines presence of sepsis-specific parameters, and applies different machine learning models accordingly. This intermediary processing layer bridges the gap between limited emergency department data and accurate sepsis stratification needs, allowing the system to achieve both detection capability and stratification accuracy despite data limitations.
Solution Approach 2:
The patent changes the parameter processing approach by dynamically selecting different machine learning models based on the presence or absence of sepsis-specific parameters in the input data. When sepsis-specific parameters are present, a first machine learning model is applied; when absent, a second machine learning model is used. This parameter-based model selection resolves the contradiction by adapting the analysis method to the available data quality.
2Reliability
If more patient health information is collected, then patient stratification accuracy is improved, but time required for sepsis determination increases
Solution Approach 1:
The patent performs preliminary action by determining the presence or absence of sepsis-specific parameters in the received medical parameters before applying the machine learning model. This preliminary assessment allows the system to prepare and select the appropriate machine learning model in advance, streamlining the processing workflow and reducing overall determination time while maintaining accurate patient stratification.
3Measurement precision
If sepsis-specific parameters are obtained, then determination accuracy is improved, but complexity of the medical dataset increases
Solution Approach 1:
The patent applies dynamics by making the machine learning model selection dynamic rather than static. The system dynamically determines whether sepsis-specific parameters are present in the medical parameters and automatically selects the appropriate machine learning model accordingly. This dynamic adaptation simplifies the overall system architecture by using a single unified framework that handles both simple and complex datasets through intelligent model selection, rather than requiring separate fixed systems for each data type.
Data Source
AI summary
A method and a device for determining onset of sepsis are provided. In one aspect, the method includes receiving a medical dataset associated with the patient. Further, the method includes determining if the plurality of medical parameters includes at least one sepsis specific parameter. Additionally, the method includes determining a first output parameter if the plurality of medical parameters does not include at least one sepsis specific parameter. The method also includes determining a second output parameter indicative of onset of sepsis in the patient if the plurality of medical parameters include at least one sepsis specific parameter, wherein the medical parameters associated with the patient are obtained for at least one time instance.