Inspiration

In Singapore’s rapidly ageing society, social isolation among elderly residents poses a critical challenge to community health and well-being. Many seniors living alone risk becoming disconnected from essential community support systems and healthcare networks. MONARCHX was conceived to empower social service agencies, community workers, and urban planners with data-driven insights to proactively identify vulnerable seniors and deploy targeted interventions before acute crises occur.

What it does

Our project leverages public demographic, spatial, and transport accessibility datasets to model and predict social isolation risk across different planning areas in Singapore. By combining geographic indicators with analytical modeling on Databricks, MONARCHX generates risk scores and interactive heatmaps. This enables care organizers to optimize resource allocation, prioritize outreach programs, and ensure support reaches the most isolated elderly populations efficiently.

How we built it

Data Processing & Storage: Ingested open datasets (SingStat, Data.gov.sg, LTA DataMall, and OneMap API) into Databricks Delta Lake, organizing pipelines into Bronze, Silver (geospatial-enriched), and Gold (risk-scored) layers.
Governance: Centralized data access control and lineage tracking using Databricks Unity Catalog.
Analytics & AI: Employed Databricks SQL for spatial queries and Databricks Genie to allow non-technical social workers to ask plain-English questions about isolation trends.
Machine Learning: Built predictive risk models using MLflow to track experiments and systematically calculate neighborhood-level isolation scores.
Visualization: Designed interactive Databricks Dashboards to display geographic risk maps and actionable priority metrics.

Challenges we ran into

Integrating disparate open datasets into a cohesive geospatial framework presented significant schema alignment and coordinate mapping challenges. Additionally, designing a machine learning model that accurately reflects social isolation risk without relying on invasive personal data required careful feature engineering using aggregated public proxies.

Accomplishments that we're proud of

Designing an end-to-end Databricks Lakehouse architecture that strictly adheres to ethical data privacy by using 100% open, publicly accessible data.
Enabling intuitive natural language querying via Databricks Genie so field workers can easily interact with complex data insights without writing code.

What we learned

We gained deep hands-on experience structuring lakehouse architectures on Databricks, enforcing data governance via Unity Catalog, and applying geospatial ML techniques to address real-world social impact challenges in Singapore.

What's next for MONARCHX ?

For Round 2 and beyond, we plan to refine our predictive models using real-time community feedback, integrate micro-mobility accessibility metrics, and build live alerting workflows that trigger outreach recommendations for local eldercare centers.