The J Healthcare Initiative seeks to develop a data-driven community health needs analysis that identifies underserved populations and high-priority geographic areas where our organization can direct programs, resources, and outreach for maximum impact. Using publicly available health and socioeconomic datasets, the student team will build predictive and clustering models that pinpoint where health needs are greatest, and create a lightweight no-code AI assistant that translates these technical findings into plain-language briefs our staff and funders can act on.

This project allows our startup nonprofit to make evidence-based decisions about where to expand, while generating the kind of quantitative impact metrics that strengthen grant applications and stakeholder reporting.

1. Build a clean, analysis-ready dataset from public health sources.

Aggregate and preprocess data from sources such as the CDC, County Health Rankings, U.S. Census, and Social Determinants of Health datasets using Python and Pandas, producing a documented, reusable dataset the organization can update over time.

2. Identify high-need communities through machine learning.

Apply clustering and/or predictive modeling to segment geographic areas (by county or zip code) according to health risk factors, access barriers, and demographic indicators, surfacing the communities most aligned with our mission.

3. Quantify and prioritize opportunities for impact.

Produce a ranked list of target areas with supporting metrics (e.g., number of high-need regions identified, population size reached, key risk indicators per area) that the organization can use for program planning and funder reporting.

4. Deliver a no-code AI assistant for ongoing use.

Build a lightweight assistant using Gemini, Notion AI, or Airtable that generates plain-language summaries of the analysis and answers staff questions about the findings, ensuring the work remains usable by non-technical team members after the project ends.

5. Communicate results to technical and non-technical stakeholders.

Deliver clear documentation, a final presentation, and a metrics summary report that explains the methodology, findings, and recommendations in accessible terms.

1. Cleaned, analysis-ready dataset

A documented dataset compiled from public sources (CDC, County Health Rankings, U.S. Census, Social Determinants of Health data), preprocessed in Python/Pandas, with a data dictionary explaining each variable and its source.

2. Machine learning model and analysis

Trained clustering and/or predictive models that segment geographic areas by health risk and access barriers, delivered as well-commented Python code (Jupyter notebooks or scripts) along with documentation of the methodology, model performance, and key assumptions.

3. Prioritized community needs report

A ranked list of high-need communities (by county or zip code) with supporting metrics for each, such as population size, key health risk indicators, and access gaps, presented in a format the organization can use directly for program planning and funder reporting.

4. Visual dashboard or summary visualizations

Charts, maps, or a lightweight dashboard that make the findings easy to interpret at a glance for non-technical stakeholders.

5. No-code AI assistant

A functional lightweight assistant (built with Gemini, Notion AI, or Airtable) that generates plain-language summaries of the findings and answers staff questions about the analysis, with a short user guide so our team can operate it independently.

6. Final documentation and presentation

A written project summary and a final presentation that explains the methodology, results, and recommended next steps in clear, accessible language, along with a metrics summary report capturing the project's measurable outcomes.