Enhancing health equity by extraction of Social Determinants of Health from unstructured text

April 12, 2023

Social determinants of health (SDoH) are the economic and social conditions that influence individual and group differences in health status. Where people live, work, and grow significantly affects quality of life and health outcomes. Many healthcare organizations aim to understand the complex array of factors, other than direct medical care, that drive health outcomes, such as underlying genetics, health behaviors, social and environmental factors. However, they face significant challenges as 80% of medical data remains unstructured and untapped after it is created.

Today, we are excited to announce addition of SDoH and ethnicity support in Text Analytics for health to unlock mentions of social, environmental, and demographics factors from unstructured biomedical data. This will enable extraction of insights, improvements in care, assessment of health inequity, tracking health outcomes, and incorporation of underrepresented groups into clinical trials and research, breaking cycles of disparity.

We introduce five new entity types: EMPLOYMENT, LIVING_STATUS, SUBSTANCE_USE, SUBSTANCE_USE_AMOUNT and ETHNICITY.

To enable deeper contextual understanding and further interpretation of extracted insights, the new capability also includes assertion detection, such as negation of substance use. Moreover, timing, frequency and amounts mentioned in the context of SDoH will be captured and associated with the entities using semantic relations that are now surfaced as well.

This new capability opens a world of possibilities for providers, payors, scientists and pharmaceutical companies, allowing them to enrich their insights with data points previously locked within unstructured text—creating a data-driven approach to social determinants of health.

At Microsoft, we hope to empower people and organizations to identify, understand and leverage the factors that impact health of patient communities, to reduce health disparities and ultimately achieve better health outcomes.

Get started today:

What is the Text Analytics for health in Azure Cognitive Service for Language? - Azure Cognitive Services | Microsoft Learn

Lean more about SDoH:

Social determinants of health (who.int)

FDA Takes Important Steps to Increase Racial and Ethnic Diversity in Clinical Trials | FDA

County Health Rankings: Relationships Between Determinant Factors and Health Outcomes - ScienceDirect