Patient Risk Intervention Prioritization With Continuous ML Updates
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional healthcare management systems are incompatible with modern, value-based approaches and fail to support patient populations like Medicaid recipients, particularly in rural or remote locations, lacking predictive power for individual patients and failing to incorporate diverse data sources effectively.
Innovation Solution
A system utilizing a machine learning model trained on enriched patient data, including EHRs, insurance claims, and patient engagement data, to generate personalized intervention lists and continuously update them based on feedback, enhancing predictive power for individual patients.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If traditional fee-for-service software tools are used, then the system supports traditional healthcare delivery models, but it cannot support value-based healthcare delivery models
Solution Approach 1:
The system dynamically adapts between different healthcare delivery models by configuring software tools and workflows based on the selected model (fee-for-service vs. value-based). This allows the same system to support multiple delivery models without requiring separate systems, resolving the contradiction between adaptability and complexity.
Solution Approach 2:
The system is designed with universal capabilities to support both traditional fee-for-service and modern value-based healthcare delivery models through a single platform. By integrating multiple functions and models within one system, it eliminates the need for separate systems while maintaining support for diverse delivery approaches.
2Measurement precision
If traditional population models are used, then the system can predict cost for aggregate populations, but it lacks predictive power for individual patients
Solution Approach 1:
The system applies different levels of predictive modeling to different patient contexts: aggregate population models for general trends and individualized predictive models for specific patients. This local differentiation allows precise individual predictions where needed while using broader models for population-level insights, resolving the contradiction between individual precision and data requirements.
3Adaptability or versatility
If tools are designed for traditional healthcare delivery, then they work well for conventional settings, but they fail to support rural or remote locations with limited connectivity
Solution Approach 1:
The system dynamically adjusts its operational mode based on connectivity availability. In high-connectivity environments, it uses cloud-based processing and real-time updates. In low-connectivity rural or remote settings, it switches to offline-capable modes with local processing and asynchronous synchronization, maintaining ease of operation across diverse environments while expanding adaptability.
4Measurement precision
If continuous model re-training is performed, then the predictive accuracy improves over time, but the computational resources and time required increase
Solution Approach 1:
The system implements periodic re-training of predictive models at scheduled intervals rather than continuous re-training. This periodic approach maintains predictive accuracy over time while managing computational resource consumption and training time, resolving the contradiction between improving precision and reducing time loss.
Solution Approach 2:
The system maintains continuous predictive capability through incremental learning and online updating mechanisms that allow the model to adapt to new data without requiring complete re-training. This continuous useful action preserves predictive accuracy while minimizing interruptions and computational overhead associated with full model re-training cycles.
Data Source
AI summary
Techniques for operationalizing predicted changes in risk based on interventions are disclosed. In an example method, a computing system stores information about a plurality of patients. The computing system receives, from an ensemble machine learning model, intervention information for a patient including a prediction of risk and change in risk for certain interventions. The computing system generates a prioritized intervention list and provides it to a client device. The computing system receives, from the client device, updated patient engagement data for the patient and adds it to the patient engagement data. The computing system receives, from the ensemble machine learning model that is re-trained using the updated data, updated predictions. The computing system generates an updated prioritized intervention list and provides it to the client device to cause a graphical user interface (“GUI”) to be automatically refreshed with the updated list.