public health
From patterns to careful interpretation
Responsible youth-persona work at WDGPH, described through public methods and safeguards.
- Role
- AI4PH Data and Analytics Intern · author
- Status
- Ongoing internship · public methods articles
Context and real problem
Public-health teams need ways to examine how youth needs, strengths, risks, and supports overlap. Exploratory clustering can reveal patterns, but it cannot decide what those patterns mean or whether they should become personas.
Andres’s role
As an AI4PH Data and Analytics Intern with Wellington-Dufferin-Guelph Public Health, I have worked on exploratory analytics and written public explanations of its responsible-use boundaries. My contribution connects technical choices with documentation and communication.
Collaborators and setting
This work sits within WDGPH’s public-health and analytics setting, through the AI4PH training platform. Survey design, epidemiology, privacy, and public-health judgment matter alongside technical methods. This summary does not imply sole ownership of the project or its data.
Methods and tools
The public account describes survey-routing review, distinctions between missing and not-asked responses, dimensionality reduction, comparison of clustering approaches, and checks for stability and interpretability. Python, Jupyter, Kubeflow, and version control support the analytical workflow.
Responsibility and safety
Personas are provisional descriptions of response patterns, not identities or diagnoses. Privacy, confidentiality, equity, and human review constrain what can be interpreted or shared. The project is not a basis for ranking schools, identifying students, predicting individual futures, or automatically assigning services.
Outputs and outcomes
Two Phiz articles explain the methodological choices and safeguards. They make the reasoning accessible while preserving the confidentiality of the underlying analysis.
What I learned
Finding a pattern and understanding it are separate tasks. A useful workflow leaves room to question the data preparation, revise the interpretation, and decide that a candidate should remain private or should not become a persona at all.
Current status and links
The articles were published in August 2026, and the internship continues as of October 3, 2026. The links provide the methods account; survey records and internal results remain within their approved setting.