MIT COVID-19 Challenge Datathon

New York City Vulnerable Neighborhoods Based on Community Health Conditions & Connections to COVID-19

A. Methodology Overview

1. Cluster vulnerable neighborhoods based on pre-existing health conditions
2. Identify relationships between clusters and COVID-19 cases per capita
3. identify neighborhood clusters that should be the focus of greater testing/preventative public health measures
4. Determine high-need neighborhoods based on the ratio between cases per capita and tests per capita
5. Prioritize the allocation of COVID-19 resources

B. Data

Community Health Condition Statistics

• http://a816-dohbesp.nyc.gov/IndicatorPublic/Subtopic.aspx
• Run by the NYC Department of Health and Mental Hygiene
• The following conditions were chosen because they have been demonstrated by the CDC to put people at higher risk for severe illness from COVID-19:
  • Heart Attack
  • Chronic obstructive pulmonary disease (COPD)
  • Asthma
  • Obesity
• Has neighborhood codes in UHF34 or UHF 42 standards

Reference: https://www.cdc.gov/coronavirus/2019-ncov/need-extra-precautions/people-at-higher-risk.html

NYC COVID-19 Data

• https://github.com/nychealth/coronavirus-data
• Updated daily
• Assembled by the NYC Department of Health and Mental Hygiene
• Has data per modified ZIP Code Tabulation Area (MODZCTA)

C. Findings

NYC Clusters based on Health Condition Statistics

All of Bronx and Staten Island neighborhoods are in Cluster 0. Most neighborhoods in Manhattan are in Cluster 1. Queens dominates Cluster 2.

Allocation of COVID-19 Resources for High Risk Clusters

Canerise - Flatlands and Southeast Queens in Cluster 0; Jamaica, Fordham - Bronx Park, and Rockaways in Cluster 2 seem to be under-tested.

D. Challenges & Limitations

• Finding datasets with consistent granularity
• Finding a large number of health indicator features in public datasets

E. Next Steps

• Build on our proof-of-concept by evaluating additional health conditions and additional cities

• Allocate resources to high risk areas

• Build regression models that can potentially predict COVID-19 case rate based on underlying health conditions

• Identify health factors that drive COVID-19 cases in certain neighborhoods

• Possible Expansion of Dataset:

  • 500 Cities: Local Data for Better Health: https://www.cdc.gov/500Cities/
  • HUD USPS Zip Code Crosswalk Files: https://www.huduser.gov/portal/datasets/usps_crosswalk.html