The National Digital Health Mission – announced by the Prime Minister on Independence Day — aims to develop the backbone needed for the integrated digital health infrastructure of India. As the COVID-19 pandemic has shown, a well-functioning personal health data infrastructure can play a key role in public health management. Developing countries like India, with significant health challenges, perhaps need such an infrastructure the most. This can help not only with diagnostics and management of health episodes, but also with broader public health monitoring, socio-economic studies, epidemiology, research, prioritising resource allocation and policy interventions. Digitisation can’t be a substitute for the fundamentals — for example, investment in nutrition and welfare, primary healthcare services and healthcare professionals – but it can certainly make healthcare more organised, efficient, and effective.
However, before we start designing databases and APIs and drafting laws, we must be mindful of certain considerations for design choices and policies to achieve the desired social objectives.
First, the theory of pathways to “public good” needs to be carefully developed. This should not be based on mere claims and opinions but on rigorous scrutiny and peer review. There must be a careful examination of how exactly digitisation may facilitate better diagnosis and management, and an understanding of the data structures required for effective epidemiology. We must articulate how we may use digitisation and data to understand and alleviate health problems such as malnutrition and child stunting, the precise data we require to better understand crucial maternal- and childcare-related problems, how digitisation and precise data might help better manage and plan our vaccination programmes or the healthcare system in general, and how implementation programmes may affect different communities — especially marginalised ones — in different ways.
Second, the potential tensions between public good and individual rights must be examined, as must the suitable ways to navigate them. Any data infrastructure endeavour that fails to avoid possible exclusions and hardships due to digitisation, or effectively address privacy concerns and incorporate ex-ante privacy protection, is bound to get mired in controversies and endless litigations. Moreover, for the balancing to be sound and for determining the level of due diligence required, it is imperative to clearly define the operational standards for privacy management. Privacy cannot be protected by mere forceful proclamations of privacy-by-design. Conflating privacy with security, as is typical in careless approaches, will invariably lead to problematic solutions. In fact, most attempts at building health data infrastructures worldwide — including in the UK, Sweden, Australia, the US and several other countries — have led to serious privacy-related controversies and have not yet been completely successful.
Third, and relating to the previous point, is digital identity. Even if we define and use a sector-specific identity, the question of when and how to link it with that of other sectors remains — for example, with banking or insurance for financial transactions, or with welfare and education for transactions and analytics. Indiscriminate linking may break silos and create a digital panopticon, whereas not linking at all will result in not realising the full powers of data analytics and inference.
Fourth, we need to work out the operational requirements of the data infrastructure in ways that are informed by, and consonant with, the previous points. In other words, the design of the operationalisation elements must follow the deliberations on above points, and not run ahead of them. This requires identifying the diverse data sources and their complexity — which may include immunisation records, birth and death records, informal health care workers, dispensaries, primary health care centres, anganwadis and schools, government and private hospitals at the primary, secondary and tertiary levels, imaging and other diagnostic centres, personal health equipment, self-declared records of lifestyle indicators and habits, and perhaps also genetic data along with profiling information. It also requires an understanding of their frequency of generation, error models, access rights, interoperability, sharing and other operational requirements. There also are the complex issues of research and non-profit uses of data, and of data economics for private sector medical research.
Finally, “due process” has always been a weak point in India, particularly for technological interventions. Building an effective system that can engender people’s trust not only requires managing the floor of the Parliament and passing a just and proportional law, but also building a transparent process of design and refinement through openness and public consultations. It will be imperative to consider all concerns, avoid “crony expertism” and reject half-baked and poorly-conceived designs. In particular, technologists and technocrats should take care to not define “public good” as what they can conveniently deliver, and instead understand what is actually required. While we can understand the urge to move forward quickly, given the urgent need to improve health outcomes in the country, deliberate care is needed.
Developing a comprehensive understanding of the considerations related to health data infrastructure may also inform the general concerns of e-governance and administrative digitisation in India, which have not been all smooth sailing.
This article first appeared in the print edition on March 8, 2021 under the title ‘In digital health, equity first’. Banerjee is professor, computer science and engineering, IIT Delhi; Sagar is founding head, School of Public Policy, IIT Delhi
Developing a comprehensive understanding of the considerations related to health data infrastructure may also inform the general concerns of e-governance and administrative digitisation in India, which have not been all smooth sailing. 
Share
