USA-001

Los Angeles County Homelessness Prevention Unit (HPU) Predictive Analytics

Download PDF
United States North America High income Pilot / Controlled Trial Phase Confirmed

Los Angeles County Department of Health Services (Housing for Health division); LA County Chief Information Office

At a Glance

What it does Prediction (including forecasting) — Vulnerability, needs and risk assessment, including predictive analytics
Who runs it Los Angeles County Department of Health Services (Housing for Health division); LA County Chief Information Office
Programme Los Angeles County Homelessness Prevention Unit (HPU)
Confidence Confirmed
Deployment Status Pilot / Controlled Trial Phase
Key Risks Data-related risks
Key Outcomes Participants in the LA County HPU were 71% less likely to enter shelter or contact street outreach within 18 months; 86% of participants retained housing; enrolment increased from 21% to 35% after operational improvements.
Source Quality 8 sources — Report (government / official), Report (multilateral / development partner), Government website / press release, +1 more

The Los Angeles County Homelessness Prevention Unit (HPU) is a predictive-analytics-assisted homelessness prevention programme operating out of the Housing for Health division of the Los Angeles County Department of Health Services, in close collaboration with the LA County Chief Information Office and the Department of Mental Health. Launched in 2021, the HPU uses linked administrative data to identify people at heightened risk of first-time homelessness or returns to homelessness and then proactively offers support before shelter entry or street homelessness occurs.

The predictive model that powers the HPU was developed by the California Policy Lab (CPL) at UCLA. The model uses approximately 580 variables drawn from integrated administrative data across multiple county agencies, including the Department of Public Social Services, the Department of Mental Health, and the Department of Health Services. These variables include factors such as enrolment in health services, use of public benefits, emergency room visits, arrests, interactions with probation, and prior contacts with homeless services. People on the resulting high-risk list experience homelessness at a rate nearly 3.5 times higher than the broader eligible population.

Once the model identifies high-risk individuals, the HPU conducts proactive outreach through phone calls, mailed letters, and emails rather than waiting for people to request assistance. Case managers carry small caseloads and provide several months of personalised case management. Services include healthcare referrals, job training, mental health treatment, and practical support such as household items or technology needed to stabilise housing and employment. The intervention is therefore framed as preventative support allocation rather than automated denial, sanction, or exclusion.

The HPU pilot phase ran from May 2022 to February 2023. A California Policy Lab report found that participants in the HPU programme were 71 percent less likely to enter a homeless shelter or have contact with street outreach teams within 18 months than similar high-risk individuals who did not enrol. Participants also experienced lower rates of mental health crisis stabilisation events and criminal justice involvement. As of the most recent reporting, the HPU has served 1,498 people, and 86 percent of participants retained their housing upon completion of the programme. The enrolment rate rose from 21 percent to 35 percent after operational improvements including a dedicated outreach team and a standardised case review and discharge process. A formal randomised controlled trial evaluation is underway, with results anticipated in 2027. The integrated administrative data environment that supports the predictive model is governed by Los Angeles County data-sharing agreements and privacy controls that regulate how information flows between participating agencies.

The California Policy Lab also conducted a fairness evaluation of the predictive model, assessing whether it systematically excluded individuals from particular racial, ethnic, or gender groups. That analysis found no evidence of systematic exclusion, with similar false negative rates across groups and no statistically significant differences in performance across race, ethnicity, and gender.

The programme operates within a human-in-the-loop oversight model. Case managers review and validate AI-generated risk flags before outreach or referral actions are taken, and model outputs serve as decision support rather than automated determinations. The decision criticality is high because model outputs can influence access to prevention services and housing support. The HPU has also undergone fairness auditing and uses an integrated administrative data environment governed by county data-sharing and privacy controls.

Classifications follow the DCI AI Hub Taxonomy. Hover over field labels for definitions.

Social Protection Functions

Implementation/delivery chain
Assessment of needs/conditions + enrolment primaryCase management Outreach/communications/sensitisation
SP Pillar (Primary) The social protection branch: social assistance, social insurance, or labour market programmes. Social assistance
Programme Name Los Angeles County Homelessness Prevention Unit (HPU)
Programme Type The type of social protection programme, classified under social assistance, social insurance, or labour market programmes. View in glossary Other
System Level Where in the social protection system the AI is applied: policy level, programme design, or implementation/delivery chain. View in glossary Implementation/delivery chain
Programme Description The LA County Homelessness Prevention Unit (HPU), operated by the Department of Health Services Housing for Health division, uses a predictive model developed by the California Policy Lab at UCLA to identify and proactively reach individuals at highest risk of homelessness.
Implementation Type How the AI output is produced: Classical ML, Deep learning, Foundation model, or Hybrid. Affects validation, compute requirements, and governance profile. View in glossary Classical ML
Lifecycle Stage Current stage in the AI lifecycle, from problem identification through to monitoring, maintenance and decommissioning. View in glossary Integration and Deployment
Model Provenance Origin of the AI model: developed in-house, adapted from open-source, commercial/proprietary, or accessed via third-party API. View in glossary Developed in-house
Compute Environment Where the AI system runs: on-premise, government cloud, commercial cloud, or edge/device. View in glossary Not documented
Sovereignty Quadrant Classification of data and compute sovereignty: I (Sovereign), II (Federated/Hybrid), III (Cloud with safeguards), or IV (Shared Innovation Zone). View in glossary Not assessed
Data Residency Where the data used by the AI system is stored: domestic, regional, or international. View in glossary Not documented
Cross-Border Transfer Whether data crosses national borders, and if so, whether documented safeguards are in place. View in glossary Not documented
Decision Criticality The rights impact of the decision the AI supports. High criticality requires HITL oversight; moderate requires HOTL; low may operate HOOTL. View in glossary High
Human Oversight Type Level of human involvement: Human-in-the-Loop (active review), Human-on-the-Loop (monitoring), or Human-out-of-the-Loop (periodic audit). View in glossary HITL
Development Process Whether the AI system was developed fully in-house, through a mix of in-house and third-party, or fully by an external provider. View in glossary Mix of in-house and third-party
Highest Risk Category The most significant structural risk source identified: data, model, operational, governance, or market/sovereignty risks. View in glossary Data-related risks
Risk Assessment Status Whether a formal risk assessment, informal assessment, or independent audit has been conducted for this system. Formal assessment
  • Bias audit
  • DPIA/AIA conducted
  • Data minimisation controls
  • Human oversight protocol
  • Independent evaluation
CategorySensitivityCross-System LinkageAvailabilityKey Constraints
Administrative data from other sectorsSpecial categoryLinks data across multiple systemsCurrently available and usedLA County: 580 variables integrated across Department of Public Social Services, Department of Mental Health, Department of Health Services, including health service enrolment, benefits usage, ER visits, arrests, probation contacts, and homeless service contacts
Beneficiary registries and MISSensitiveLinks data across multiple systemsCurrently available and usedNYC CIDI: cash assistance and Medicaid records 2006-2015 linked to shelter applications, building characteristics, and neighbourhood data; requires data sharing across NYC DHS, DSS, HRA, HPD, and NYS Office of Court Administration
Social registriesSensitiveLinks data across multiple systemsCurrently available and usedHomeless shelter application and stay records used as both predictors and outcome variables in both systems

California Policy Lab (2024) The Homelessness Prevention Unit: A Proactive Approach to Preventing Homelessness. Los Angeles: CPL. Available at: https://capolicylab.org/wp-content/uploads/2024/12/Homelessness-Prevention-Unit-Report.pdf (Accessed: 31 October 2025).

View source Report (government / official)

California Policy Lab (2024) 'The Homelessness Prevention Unit: A Proactive Approach to Preventing Homelessness in Los Angeles County', capolicylab.org. Available at: https://capolicylab.org/the-homelessness-prevention-unit-a-proactive-approach-to-preventing-homelessness-in-los-angeles-county/ (Accessed: 27 March 2026).

View source Report (multilateral / development partner)

California Policy Lab (2025) 'Early Outcomes from the Los Angeles County Homelessness Prevention Unit', California Policy Lab. Available at: https://capolicylab.org/early-outcomes-from-the-los-angeles-county-homelessness-prevention-unit/ (Accessed: 24 March 2026).

View source Report (multilateral / development partner)

County of Los Angeles (2025) 'New report: Early signs of success from LA County's Homelessness Prevention Pilot', County of Los Angeles, 10 July. Available at: https://lacounty.gov/2025/07/10/new-report-early-signs-of-success-from-la-countys-homelessness-prevention-pilot/ (Accessed: 31 October 2025).

View source Government website / press release

NYC CIDI (2023) Homeless Prevention: At-Risk Students in NYC Schools. NYC.gov. Available at: https://www.nyc.gov/assets/cidi/downloads/pdfs/CIDI-Report-Homeless-Prevention-At-Risk-Students-in-NYC-Schools.pdf (Accessed: 31 October 2025).

View source Report (government / official)

NYC Center for Innovation through Data Intelligence (CIDI) (n.d.) 'Predicting Homeless Shelter Entry', NYC.gov. Available at: https://www.nyc.gov/site/cidi/projects/predicting-homeless-shelter-entry.page (Accessed: 24 March 2026).

View source Government website / press release

StateScoop (2025) 'LA County's new predictive model shows early success in homelessness prevention unit', StateScoop. Available at: https://statescoop.com/la-county-ai-predictive-model-reducing-homelessness/ (Accessed: 24 March 2026).

View source News article / media

UCLA Newsroom (2025) 'Homelessness Prevention Unit participants 71% less likely to enter a shelter, California Policy Lab at UCLA finds', UCLA Newsroom. Available at: https://newsroom.ucla.edu/stories/homeless-prevention-unit-helps-keep-people-off-streets-california-policy-lab-at-ucla (Accessed: 24 March 2026).

View source News article / media
Deployment Status How far the system has progressed into real-world operational use, from concept/exploration through to scaled and institutionalised. View in glossary Pilot / Controlled Trial Phase
Year Initiated The year the AI system was first initiated or development began. 2021
Scale / Coverage The scale and geographic or population coverage of the deployment. Los Angeles County HPU has served 1,498 people as of 2025
Technical Partners External technology vendors, academic partners, or development partners involved. California Policy Lab at UCLA (developed the predictive model); no commercial vendor identified in accessible sources
Outcomes / Results Participants in the LA County HPU were 71% less likely to enter shelter or contact street outreach within 18 months; 86% of participants retained housing; enrolment increased from 21% to 35% after operational improvements. Individuals on the high-risk list experience homelessness at 3.5x the rate of the broader eligible population.
Challenges Proactive outreach to high-risk individuals with complex health and mental health needs is the most challenging aspect of the programme. Initial low enrolment rates (21%) required significant operational improvements. Cold-calling vulnerable populations requires patience, persistence, and flexibility. Some technical model details remain undisclosed in public documentation.

How to Cite

DCI AI Hub (2026). 'Los Angeles County Homelessness Prevention Unit (HPU) Predictive Analytics', AI Hub AI Tracker, case USA-001. Digital Convergence Initiative. Available at: https://socialprotectionai.org/use-case/USA-001 [Accessed: 1 April 2026].

Change History

Created 30 Mar 2026, 08:41
by v2-import (import)