Data Bridge Project To Help Health and Social Entities Safely Share Data

After a visit to the emergency department, simply contacting an unhoused patient with test results and a prescription or referral can be a challenge.

The lack of adequate systems to connect these medical providers to outreach workers and housing navigators results in missed opportunities for better, coordinated, and ongoing help. Agencies struggle to identify shared clients, enable comprehensive care, and connect people experiencing homelessness to health care, social services, and housing. For the last mile of care required to reach unhoused patients, existing solutions are too complicated, too expensive, or both.

There is a critical need to implement low-cost, simple, and safe data sharing systems across the social safety net to promote collaboration and improve care for people experiencing homelessness.

Data Bridge, a project of the Levitt Center and the Georgetown Massive Data Institute, will enable entities to share essential data to better serve unhoused individuals in terms of operations and administrative priorities. In the past few years, approaches to data sharing that overcome long standing privacy hurdles have become vastly more practical. Data Bridge will catalyze data sharing by providing open source easy to use implementations of these approaches, training materials, and implementation services. Our vision is to reduce the costs and complexity of data sharing, thereby expanding access and reducing burdens on clients and staff.

Achieving our vision requires shaping the ecosystem. We have partnered with Homebase, a national nonprofit with over three decades of experience as a facilitator, community partner, and technical assistance provider. Homebase has deep knowledge of the realities of homelessness as well as programs and systems that help people regain housing, health, and dignity.

In collaboration with Homebase, we are 1) developing free and open source software built around operational needs, 2) providing at-cost integration services, 3) creating materials and provide assistance to public interest technology organizations to customize and deploy the software, and 4) partnering with established organizations that provide education, expertise, and leadership to show about what is possible and to set benchmarks that raise expectations for speed and cost. This strategy enables us to address the four key barriers that experts have identified as blockers to widespread use of modern privacy preserving tools in public services: awareness, adoption, integration, and usability (White House, 2023).

Pilot Software

Data Bridge has developed pilot software that enables collaboration with encrypted data (Table 1). Our software can be deployed locally in ten minutes, can be used by someone with basic technical skills, is free and open source, and provides the highest level of privacy protection available.

The figure below illustrates two use cases in the context of an example collaboration between a Managed Care Provider and a County Homeless Management Information System Lead.

Data Bridge Enables Collaboration via Encrypted Data

In the above example figure, data from each entity are encrypted within that entity and are never decrypted outside of the originating entity.

Importantly, this process does not require any new hardware or services. All the required components can be deployed via a software container in about 10 minutes on any system (Windows, Mac, or Linux). The two entities only need an agreed-upon place to put encrypted data that will be retrieved by each partner agency, such as a Secure File Transfer Protocol (SFTP) server. The server itself will not be able to access any information, as all the computations will happen on the agency machines running the script, with each agency retaining its own private key and never being able to decrypt the data of the other. The script itself is simple to execute and should only need a few minutes of time from a business or IT analyst to start it in a terminal or command prompt.

Table 1 recaps how our approach addresses the needs in the field.

Table 1: Our working software addresses key data sharing challenges

Challenge	How it’s addressed
Agencies are uncertain of the value of joining a data sharing consortium Leaders want to know the overlap of clients across multiple agencies without complex legal agreements	Multiple agencies can find out how many people they have in common, while no one can learn who those people are. This enables agencies to determine counts and percentages without signing a DSA for individual level data. (projects with Homebase, Massive Data Institute, and NYC Office of Technology and Innovation)
Small CBOs need access to agency or local health information exchange data, but do not have the same level of IT security	Private set intersection can guarantee that CBOs will only be able to access data about individuals in their roster, while enabling a governance policy about what data travels in what direction. (projects with Homebase)
The ongoing cost of licenses and services makes ambitious efforts like local information exchanges difficult to sustain	Our software suite will be easy-to-configure, free, and open source, setting a benchmark that agencies can use to push their vendors.

‍

Core Team

Eric Giannella, PhD. Eric is a Senior Scientist at the Levitt Center for Social Emergency Medicine and an Associate Research Professor at the Georgetown Better Government Lab. He has initiated and contributed to data sharing projects spanning the safety net, justice, and health. Previously, he was Senior Director of Data Science at Code for America. Eric has a BA and MS from Stanford and a PhD from UC Berkeley.

Vince Dorie, PhD. Vince is a statistician and Senior Fellow at Georgetown Massive Data Institute. Previously, he was a Principal Data Scientist at Code for America and has also worked at Columbia University, NYU, and CA DOJ. Vince has written three privacy-preserving sharing tools from scratch in C++, Python, and R. Vince has a BS and MS from Stanford and a PhD from Columbia.