Predicting the status of 35 sustainable development goal indicators in Indian villages: a semi-supervised machine learning approach for precision public policy.

Abstract

National and district-level monitoring of the Sustainable Development Goals (SDGs) in India obscures substantial inequalities at finer geographic resolutions, such as villages. While villages are central to service delivery and local governance, village-level estimates remain limited, hindering precision targeting and accountability. We estimated village-level prevalence for 35 SDG indicators in 2021, assessed their status relative to SDG targets, and identified Required Rates of Improvement (RRI) to achieve targets by 2030. We used cross-sectional data from the National Family Health Survey 2019-21 and the 2011 Census. We estimated cluster-level indicator values using multilevel modeling and linked them to census villages through a semi-supervised learning framework to generate predictions for 597,603 villages. Substantial village-level heterogeneity was observed across all SDG indicators. As of 2021, none of the villages had achieved several SDG targets, including health insurance coverage, access to basic services, women's bank account ownership, and internet use. In contrast, 99.9% of villages had achieved the target for adolescent pregnancy among girls aged 10-14 years. The largest mean RRI values were observed for access to basic services (7.61 percentage points/year), health insurance coverage among women (6.74), health insurance coverage among men (6.39), and clean fuel for cooking (6.05). Village-level inequalities are substantial enough to make district-level averages insufficient for SDG monitoring. Routine village-level monitoring can support precise targeting of public policies and strengthen local accountability. Sustaining gains among villages that have achieved SDG targets and accelerating improvement among those below target levels will be essential. This research was funded by Bill & Melinda Gates Foundation (INV-002992).