Keeping Youth Homelessness Evidence Living with Artificial Intelligence

A practical prototype for using AI-assisted review to keep intervention evidence easier to find, screen, and update.

LEARN MORE

From Static Reviews to Living Evidence

Welcome! If you are a youth homelessness researcher, service provider, student, or community partner, you know how important it is to have timely evidence about what interventions have been tested, what outcomes they improve, and where important gaps remain. But systematic reviews are often slow, labor-intensive, and difficult to update.

This website introduces a prototype for using artificial intelligence to support a living systematic review of youth homelessness interventions. The focus is on one early but important step in the review process: screening titles and abstracts.

Rather than replacing human reviewers, this prototype explores how AI can help make evidence review workflows faster, more transparent, and easier to revisit over time. It shows how AI-assisted screening can be designed, tested, compared with human screening decisions, and improved through prompt development.

The goal is to help researchers and community partners think practically about how AI can support evidence synthesis while keeping human judgment, accountability, and interpretation at the center.

Why Living Evidence Matters

Youth homelessness is complex. Interventions may focus on housing, outreach, case management, family support, mental health, education, employment, sexual health, or other areas of well-being. Because the field is broad, communities need evidence that is not only rigorous, but also accessible and current.

Systematic reviews help bring together findings across many studies. They can show what has been evaluated, what appears promising, and where evidence is still unclear. However, traditional reviews can take months or years to complete. New studies may appear after a review is finished, and updating the review can require repeating many labor-intensive steps.

That is the problem this prototype responds to:
How can we make evidence review more updateable without losing rigor?

Why AI-Assisted Abstract Screening?

Abstract screening is one of the most time-consuming parts of a systematic review. Reviewers must read large numbers of titles and abstracts to decide which studies may be relevant enough for full-text review.

This step matters because early screening decisions shape the whole review. If relevant studies are excluded too early, important evidence may be missed. If too many irrelevant studies are retained, the review becomes slower and harder to manage.

This prototype explores whether AI can assist with abstract screening as a human-guided triage tool. AI is used to support the workflow, but human reviewers remain responsible for validation, interpretation, and final decisions.

  • Understand why youth homelessness intervention evidence matters and why systematic reviews are important. Learn more.

  • See how abstracts move from database search to human screening, AI-assisted screening, merged decisions, and performance evaluation. Learn more.

  • Learn how different prompts changed AI screening performance and why prompt design matters. Learn more.

  • See what the prototype revealed about recall, precision, specificity, accuracy, F1, AUC, and agreement with human reviewers. Learn more.

  • Find the GitHub workflow and reproducibility materials for adapting the process. Learn more.

  • Living Notes is a space for short reflections, evidence updates, and field-facing commentary on youth homelessness interventions.

    The goal is to make evidence easier to discuss, question, and use. This page is for researchers, policy analysts, service providers, students, advocates, community partners, and others working to prevent and end youth homelessness. Learn more.

Explore the Prototype

On this website, you can:

Who We Are

We are a team of researchers and policy analysts at Chapin Hall Center for Children working to make evidence more useful for people who care about youth homelessness. Our goal is to help researchers, service providers, students, and community partners better understand how intervention evidence can be screened, updated, and translated into practice.

Work With Us

Meet the Team

  • Kay Chansiri, Ph.D.

    PRINCIPAL INVESTIGATOR

    Dr. Chansiri is a Senior Researcher at Chapin Hall and Co-Chair of the HCI & AI Affinity Group. Her research focuses on human-centered AI in decision support for social services. For this project, she led the reconceptualization of the living systematic review as an AI-assisted evidence review prototype and developed the end-to-end AI-assisted abstract screening workflow.

  • Sara Semborski, Ph.D., LCSW

    CO-PRINCIPAL INVESTIGATOR

    Dr. Sara Semborski is a Researcher at Chapin Hall whose work focuses on programs and policies designed to address disparities among young people experiencing homelessness. Her research uses quantitative and mixed methods to translate evidence into advocacy, policy, and practice. For this project, she led human validation of abstract screening decisions.

  • Xinyu Wei

    RESEARCHER

    Xinyu Wei is a Researcher at Chapin Hall with expertise in quantitative and qualitative research methods, policy analysis, data analysis, and visualization. For this project, she supported implementation of the AI-assisted workflow by assisting with code execution and workflow testing.

  • Brynne Townley

    ASSOCIATE POLICY ANALYST

    Brynne Townley is an Associate Policy Analyst at Chapin Hall with experience in policy analysis, survey development, project management, knowledge management, and translating research evidence into policy and practice. For this project, she provided human validation assistance for the abstract screening workflow.

  • Jade Winn, Ph.D., Ed.D.

    Dr. Winn is an Associate University Librarian at the University of Southern California, where her expertise includes evidence synthesis research methods, systematic and scoping reviews, information literacy, AI in education, and social science research methods. For this project, she led evidence synthesis and library research expertise and supported human validation of abstract screening decisions.

This project is supported by the Liberty Mutual Foundation. Their support helps advance work that makes youth homelessness intervention evidence more accessible, updateable, and useful for researchers, service providers, students, and community partners.

Project Support

Have a Question?

Send us a note and we’ll figure things out together.