Machine Learning in Maternal Care – Johns Hopkins Bloomberg School of Public Health

Report on the Application of Artificial Intelligence in Maternal Health to Advance Sustainable Development Goals

Introduction: Aligning Technological Innovation with Global Health Targets

This report details the integration of artificial intelligence (AI) into maternal health, highlighting its capacity to advance key Sustainable Development Goals (SDGs). The application of AI-driven tools is facilitating a critical shift from reactive responses to proactive, preventive healthcare strategies. This transformation directly supports the achievement of SDG 3: Good Health and Well-being, particularly Target 3.1, which aims to reduce the global maternal mortality ratio. By enabling early risk identification and standardizing data collection, AI presents an innovative pathway to improve health outcomes for mothers and newborns worldwide.

AI-Driven Predictive Analytics for Maternal Complications

Research demonstrates the significant potential of AI models in predicting life-threatening maternal health complications. These models contribute directly to SDG 3 by providing actionable lead time for clinical interventions.

• Predictive Modeling: AI models trained on routinely collected electronic health records (EHR) and ultrasound data have shown strong performance in predicting severe conditions such as postpartum hemorrhage and severe maternal morbidity.
• Early Intervention: The primary benefit of these models is their ability to identify at-risk individuals early, allowing healthcare systems to implement risk-appropriate and preventive measures, thereby reducing mortality and morbidity.
• Data-Driven Decision-Making: By leveraging large datasets, AI augments clinical decision-making, ensuring that interventions are informed by robust, evidence-based predictions.

Enhancing Equity and Reducing Disparities through AI

A core function of this technological advancement is its potential to mitigate systemic inequalities in healthcare, aligning with SDG 10: Reduced Inequalities and SDG 5: Gender Equality. By ensuring equitable access to high-quality diagnostics and care, AI can address long-standing disparities.

1. Standardization of Care: AI-enabled workflows, such as automated fetal biometrics from ultrasound scans, reduce operator variability and standardize measurements. This ensures a consistent quality of care, irrespective of geographic location or provider.
2. Addressing Disparities: Through careful calibration, subgroup analysis, and fairness assessments, these models can be designed to expose and help correct for existing geographic and racial disparities in maternal health outcomes.
3. Improved Continuity of Care: Standardized, AI-assisted data collection enhances the continuity of care, which is crucial for underserved populations and contributes to the empowerment of all women through better health.

Framework for Responsible Implementation and Innovation

The successful deployment of AI in public health requires a structured and ethical framework. This approach supports SDG 9: Industry, Innovation, and Infrastructure by promoting the development of resilient and technologically advanced health systems.

• Robust Data Pipelines: Establishing secure and comprehensive data infrastructure is fundamental for training and validating effective AI models.
• Transparent Reporting and Validation: To ensure transportability and trust, models must undergo external validation with transparent reporting on performance metrics, including AUC, calibration, and net benefit analysis.
• Human-in-the-Loop Governance: A critical guardrail is the integration of human oversight into AI-driven systems to ensure clinical relevance, ethical application, and ultimate accountability in decision-making.

Analysis of Sustainable Development Goals in the Article

1. Which SDGs are addressed or connected to the issues highlighted in the article?

   The article primarily addresses issues related to the following Sustainable Development Goals (SDGs):

   • SDG 3: Good Health and Well-being: The core focus of the article is on maternal health, specifically using AI to predict and prevent life-threatening complications. This directly aligns with the goal of ensuring healthy lives and promoting well-being for all at all ages.
   • SDG 10: Reduced Inequalities: The article explicitly mentions that AI tools can “expose inequities” and “address geographic and racial disparities in maternal health,” which connects directly to the goal of reducing inequality within and among countries.
   • SDG 5: Gender Equality: While not mentioned directly, improving maternal health is a critical component of achieving gender equality. Ensuring the health and safety of women during childbirth empowers them and is fundamental to this goal.

2. What specific targets under those SDGs can be identified based on the article’s content?

   Based on the article’s discussion, the following specific SDG targets can be identified:

   • Target 3.1: By 2030, reduce the global maternal mortality ratio to less than 70 per 100,000 live births. The article directly supports this target by discussing how AI models can “predict life-threatening complications (e.g., postpartum hemorrhage and severe maternal morbidity).” By enabling “proactive prevention through early risk identification,” these tools aim to reduce the incidents that lead to maternal mortality.
   • Target 3.8: Achieve universal health coverage, including financial risk protection, access to quality essential health-care services and access to safe, effective, quality and affordable essential medicines and vaccines for all. The article connects to the quality of care aspect of this target. It states that AI-enabled workflows can “standardize measurements (e.g., automated fetal biometrics), reduce operator variability, and enhance continuity of care,” thereby improving the quality and reliability of maternal healthcare services.
   • Target 10.2: By 2030, empower and promote the social, economic and political inclusion of all, irrespective of age, sex, disability, race, ethnicity, origin, religion or economic or other status. The article’s emphasis on using AI for “subgroup analysis, and fairness assessment” to “address geographic and racial disparities in maternal health” directly relates to this target of promoting inclusion and reducing health outcome inequalities based on race and location.

3. Are there any indicators mentioned or implied in the article that can be used to measure progress towards the identified targets?

   The article implies several indicators that can be used to measure progress:

   • Incidence of Postpartum Hemorrhage and Severe Maternal Morbidity: The article names these as “life-threatening complications” that AI models can predict. A reduction in the rates of these conditions would be a direct indicator of progress towards Target 3.1 (reducing maternal mortality), as they are major contributors to it.
   • Disparities in Maternal Health Outcomes by Race and Geography: The article’s goal to “address geographic and racial disparities” implies that progress would be measured by tracking and reducing the gap in maternal health outcomes between different racial and geographic groups. This serves as a practical indicator for Target 10.2.
   • AI Model Performance Metrics: The article mentions specific technical metrics for evaluating the tools themselves, such as “AUC, calibration, decision-relevant thresholds, and net benefit.” These are indicators of the effectiveness and reliability of the AI interventions being used to achieve the broader health goals. While not official SDG indicators, they are crucial for measuring the success of the technological approach discussed.

4. Create a table with three columns titled ‘SDGs, Targets and Indicators” to present the findings from analyzing the article. In this table, list the Sustainable Development Goals (SDGs), their corresponding targets, and the specific indicators identified in the article.

   SDGs	Targets	Indicators (Mentioned or Implied in the Article)
   SDG 3: Good Health and Well-being	Target 3.1: Reduce the global maternal mortality ratio.	Rates of postpartum hemorrhage and severe maternal morbidity.
   SDG 3: Good Health and Well-being	Target 3.8: Achieve universal health coverage, including access to quality essential health-care services.	Standardization of measurements (e.g., fetal biometrics) and reduction in operator variability.
   SDG 10: Reduced Inequalities	Target 10.2: Promote the inclusion of all, irrespective of race, ethnicity, etc.	Measurement of geographic and racial disparities in maternal health outcomes.

Source: publichealth.jhu.edu