A novel data-driven approach for Personas validation in healthcare using self-supervised machine learning.

Area of Science:

• Data Science
• Machine Learning
• Healthcare Analytics

Background:

• Persona validation traditionally relies on expensive external methods.
• Existing data during persona creation is often underutilized for validation.

Purpose of the Study:

• To develop a novel, cost-effective persona validation method.
• To leverage readily available data for persona validation.
• To utilize self-supervised machine learning for persona validation.

Main Methods:

• A self-supervised machine learning (SSML) approach was developed.
• Data was split into training (80%) and testing (20%) sets.
• 5-fold cross-validation identified optimal models, with majority voting for final predictions.

Main Results:

• The SSML method was tested on two distinct healthcare datasets.
• High performance metrics were achieved: weighted accuracy (up to 94.12%), precision (up to 92.83%), recall (up to 91.67%), and F1 score (up to 91.76%).
• The method demonstrated effective differentiation of persona clusters.

Conclusions:

• The proposed SSML method offers strong generalization capabilities for persona validation.
• It successfully validates the ability of personas to stratify target populations.
• This approach significantly lowers the cost of persona validation compared to existing methods.