AI-Powered Predictive Research Platform
Research and data to make the invisible visible for Lao Minnesotans.
The nation's first AI-enhanced data infrastructure for a refugee community. 4 ML models, 3 NLP engines, and real-time early warning — transforming reactive crisis response into proactive, culturally-grounded support for ~15,000 Lao Americans.
Read about our mission →All free: open access, community-governed, and ethically audited
24.0%
vs 5.0% state average
Key finding
Gambling addiction in the Lao community is 4.8x the state average
Disaggregated data reveals that 24% of Lao adults report problem gambling, compared to 5% statewide. This disparity is invisible when Lao data is aggregated under the "Asian" category.
Dr. Phouthakannha Nantharath
75/100
risk score — threshold exceeded
AI prediction
ML models predict rising elder isolation in Minneapolis North
Our Elder Isolation Index (86.2% accuracy) shows Minneapolis North at 75/100 risk score — exceeding the 65-point community threshold. Early warning triggered for proactive intervention.
Dr. Phouthakannha Nantharath
34.8%
vs 3.8% state average
Cross-domain
Linguistic isolation drives cascading health and economic disadvantage
34.8% of Lao households are linguistically isolated, creating barriers to healthcare access, job markets, and civic participation that compound across domains.
Dr. Phouthakannha Nantharath
Aggregated "Asian" category rates
Disparities are hidden when Lao data is grouped under "Asian."
Source: MN Community Survey 2024, Census ACS 5-yr estimates
Lao Community Data Portal
AI Pipeline — Live
6 sources (Census ACS, BLS, BRFSS, Community Survey)
36,371 processed
Data Insights
See all insightsThe "Asian" category hides critical disparities for Southeast Asian communities
When Lao data is separated from aggregate Asian statistics, a vastly different picture emerges. The Lao poverty rate is 18.7% — double the state average — yet the aggregate Asian poverty rate of 8.5% makes it appear Southeast Asian communities are thriving.
Temple networks serve as the primary social safety net for Lao elders
Analysis of 500+ community narratives reveals that Buddhist temples (wats) function as de facto social services — providing food, gathering space, and community organizing that formal systems miss.
Casino buses speak Lao — hospitals do not
For linguistically isolated elders, casinos provide one of the few accessible social environments: free transportation, Lao-speaking staff, familiar faces. Gambling is often a symptom of social isolation, not just addiction.
Second-generation outcomes improve dramatically but gaps persist
Cohort analysis shows college attainment jumping from 25% (1st gen) to 72% (2nd gen), yet the Lao community still trails the state average of 62.5% overall. Generational status remains a key predictor.
From Reactive to Predictive
Before: Reactive Crisis Response
Between data collection and policy response
Lao disparities hidden under pan-ethnic category
One-size-fits-all across all Asian communities
Problems identified only after they escalate
After: Predictive & Proactive
Continuous monitoring of 6 data sources + community narratives
First AI-powered portal built specifically for a refugee community
ML-recommended, community-informed, temple-based delivery
Early warning system predicts risk weeks before threshold breach
Why Data Disaggregation Matters
The approximately 15,000 Lao Americans in Minnesota represent the largest Lao population in any U.S. state, with roots in the refugee resettlement following the Secret War (1964–1975). Yet in nearly every state data system, they are categorized simply as “Asian” — a category that includes populations with vastly different socioeconomic profiles, health outcomes, and educational attainment levels.
This aggregation creates a statistical illusion. The “Asian” category, buoyed by high-income East Asian and South Asian populations, masks severe disparities: a gambling addiction rate 4.8× the national average, linguistic isolation at 34.8% (vs. 4.5% statewide), and 45.3% of elders with no formal schooling.
This portal uses AI-powered analysis, federal data integration (Census ACS, BLS, FRED, BEA), and community-governed research to make these hidden disparities visible — and actionable.
Source: Census ACS 2020-2024, MDH BRFSS, Community Survey 2025 · Lao Community Data Portal
Methodology & Rigor
Every prediction, visualization, and recommendation is built on transparent, reproducible methods that meet peer-review standards.
SHAP Explainability
Every prediction includes feature attribution scores — no black boxes
Fairness Auditing
Disparate impact ratio, demographic parity tested on every model run
Confidence Intervals
95% CI reported for all estimates. P-values and effect sizes disclosed
Model Cards
Standardized documentation: training data, limitations, intended use
Browse by topic
Interactive Dashboard
Real-time disaggregated data across health, economic, education, and housing domains with predictive trend forecasting.
ExploreAI Command Center
4 ML models, 3 NLP engines — monitor accuracy, fairness scores, and live prediction activity in real-time.
ExploreRisk Prediction Engine
Interactive risk explorer with feature importance, confidence intervals, and intervention recommendations.
ExploreResearch & Policy Tools
Cohort analysis, disparity indices, cross-tabulation, data export, and community survey tools for researchers.
ExploreStay informed
Receive data updates, policy briefs, and community insights from the Lao Community Data Portal. All analysis is free, open-access, and community-governed.
No spam. Unsubscribe any time. Community data delivered monthly.