Supporting long-term condition management: a workflow framework for the co-development and operationalization of...

Area of Science:

• Health Informatics
• Machine Learning in Healthcare
• Clinical Data Science

Background:

• Rising prevalence of chronic diseases like cardiovascular disease, COPD, asthma, and diabetes mellitus leads to increased mortality and healthcare costs.
• Machine learning (ML) shows promise for earlier diagnosis, triage, and treatment selection in chronic disease management.
• Limited translation of ML research into clinical practice due to lack of clinical involvement and planning beyond model development.

Purpose of the Study:

• To present a multistage workflow framework for the co-development and operationalization of ML models using routine clinical data.
• To facilitate a coordinated and collaborative process from concept to clinical use.
• To address the gap in translating ML research into live clinical environments.

Main Methods:

• Development of a novel multistage workflow framework for ML model co-development and operationalization.
• Utilization of routine clinical data from electronic health records.
• Informed by multidisciplinary team experience in developing and operationalizing COPD risk prediction models.

Main Results:

• Detailed overview of the proposed workflow framework.
• Case studies demonstrating the development and operationalization of two risk-prediction models for COPD.
• Successful application of the framework in a real-world clinical setting (NHS Greater Glasgow & Clyde).

Conclusions:

• The presented framework supports the effective co-development and operationalization of ML models using electronic health record data.
• This approach enhances the translation of ML research into practical clinical tools for chronic disease management.
• Collaborative, multidisciplinary efforts are crucial for successful implementation of data-driven healthcare solutions.