November 17, 2020

Testing the readiness of (big) health data in India for advanced analytics and AI - Final takeaways from 2019 Burke Fellow Satchit Balsari

Satchit Balsari, MBBS, MPH is an Assistant Professor in Emergency Medicine at Harvard Medical School (Beth Israel Deaconess Medical Center), and in Global Health and Population at Harvard Chan. Balsari’s research is focused on the application of mobile and cloud-based technologies to protect the health of populations marginalized by disasters and war. His signature initiatives include EMcounter, a portable digital surveillance tool deployed at the world’s largest mass gatherings; Voices, a crowd-sourced, online disaster response analysis tool; and the Hurricane Maria Mortality Study (co-led with Professor Caroline Buckee), that refuted the US government’s low death toll in post-hurricane Puerto Rico. Balsari directs the India Digital Health Network (IDHN), an interdisciplinary faculty initiative at Harvard’s Lakshmi Mittal South Asia Institute, advancing digital health implementation science in India.. Dr. Balsari has also developed a range of domestic and international courses in global health and mobile technology, taught in India, Sri Lanka, South Africa, and the US. At Harvard, he co-directs the university-wide course, “Entrepreneurial Solutions to Intractable Social Problems,” and “Societal Response to Disasters” at HSPH. Dr. Balsari is faculty at the Harvard FXB Center for Health and Human Rights. He is a recipient of the Dr B.C. Roy National Award presented by the President of India; an Aspen Ideas Scholar, and an Asia 21 Fellow of the Asia Society.

In 2019, Satchit Balsari was one of three scholars awarded the Burke Global Health Fellowship. He was selected for his project titled Testing the readiness of (big) health data in India for advanced analytics and AI. In his application, Balsari argued that billions are being invested in big health data analytics and AI to develop decision making and prediction tools that aim to improve how, when, where and why we seek care. However, fundamental questions about how health data may be used, collated, anonymized, exchanged, sold, traded or analyzed are yet to be answered. International jurisprudence around these issues is still developing: empiric testing is now required to find the right balance between the desire to leverage vast amounts of data for secondary use with the need to protect an individual’s privacy and safety. 

During his year as a Burke Fellow, Balsari sought to test the “AI readiness” of healthcare data in India.. In 2019 and 2020 there were significant milestones in the development of India’s health data ecosystem. The pandemic precluded Balsari’s travel to India, and he subsequently modified the work plan while continuing to work towards strengthening the foundation of health data exchange to support the advancement of AI in healthcare. As Balsari reflected on his time as Burke Global Health fellow he stated “the support provided by the Burke Fellowship allowed me to apply my training and skills to do my bit to contribute to…addressing the pandemic, as well as leveraging technology to advance health care in resource-constrained settings. I am deeply grateful for the trust and encouragement.” The Burke Fellowship gave Dr. Balsari concentrated time to think, synthesize, and write several manuscripts; he describes that time as an immeasurable gift. 

Satchit Balsari (center, top) and teammates on a home visit in tribal Melghat region, conducting a needs assessment and examining potential implementation challenges while designing digital health interventions.

As a 2019 Burke Fellow, Dr. Balsari achieved significant goals, and he  continues to contribute meaningfully to the field of AI in global health. Key accomplishments from his time as a Burke fellow include:

1. Introducing regulatory sandboxes (from fin-tech) to the digital health data ecosystem 

In July 2019, the government of India released a prototype for the nation’s health data exchange. It reflected the key principles Balsari’s team had laid out in a 2017 paper,  “Reimagining Health Data Exchange: An Application Programming Interface–Enabled Roadmap for India,” In response to the open call for public review, Balsari (along with colleagues in India and the US) submitted a detailed response, “Response to the Invitation for Public Comment:  National Digital Health Blueprint,” which is accessible online

2. Response to the Joint Parliamentary Committee (India) on Personal Data Protection Legislation

In January 2020, the government of India tabled a draft of the proposed Personal Data Protection Bill, inviting public comment. Balsari convened a team of experts from Harvard and from India to submit a detailed analysis of the impact of the bill on health data, with a particular focus on health data exchange necessary for future applications like machine learning and AI. This document, Responses to the Government of India’s Joint Parliamentary Committee’s Consultation on The Personal Data Protection Bill, 2019,” was submitted to parliament in February 2020.

3. AI, big data, and the pandemic 

The pandemic resulted in an explosion of data in the public domain — data that until now were typically neither easily nor publically available. In particular, human mobility data from cell phones, ad tech companies, and social media corporations were used to monitor and inform physical distancing measures, with the goal of applying machine learning tools to control population flow. While arguably useful, Balsari felt this was a flawed undertaking with significant limitations in methodology, analysis, and translational capabilities. With colleagues from Harvard and the humanitarian aid organization Direct Relief, Balsari launched the COVID-19 Mobility Data Network (CMDN) covid19mobility.org, a consortium of 70+ research scientists from around the world that serves as a trusted intermediary between technology companies and local city and state governments from California to Dhaka. This work will now be expanded to a larger interdisciplinary university-wide platform scheduled to be launched in the fall, which Balsari will co-chair with colleagues from HSPH and Direct Relief.  

4. Applying differential privacy to big data

The proposed prototyping of large health data sets for anonymization and access will begin once travel restrictions in India are relaxed. Meanwhile, Balsari pivoted the work plan to apply the same principles to human mobility data from Facebook, and subsequently applied and received an intramural grant from the Harvard Data Science Initiative. Working with faculty from the Harvard John A. Paulson School of Engineering and Applied Sciences (SEAS), the Institute for Quantitative Social Science (IQSS) and HSPH, this new research will examine the application of OpenDP (developed by Gary King and Salil Vadhan) on Facebook Data for Good datasets.