Ensemble Learning in Public Health and Community‐Engaged Research: Advances, Applications, and Ethical Considerations

Ensemble Learning to Predict Short Birth Intervals

Short birth intervals, defined as periods between consecutive live births of less than 33 months, are associated with adverse maternal and infant health outcomes. Recent research conducted using the Ethiopian Demographic and Health Survey (EDHS) 2016–2019 dataset applied ensemble learning algorithms to predict short birth intervals among reproductive‐age women. With a weighted sample of over 12,500 women, the study demonstrated that ensemble learning approaches—particularly the Random Forest classifier—can achieve high performance in predicting the outcome.

Data Preprocessing and Model Development

The research team undertook extensive data preprocessing steps, including:

• Data Cleaning: Removing inconsistent or missing values and performing single imputation using the mean.
• Dimensionality Reduction: Techniques such as Principal Component Analysis (PCA) were applied to manage multicollinearity while preserving data integrity.
• Balancing the Dataset: The Synthetic Minority Over-sampling Technique (SMOTE) was employed to correct class imbalances between short and normal birth intervals.
• Data Splitting and Validation: An 80:20 train-test split was used along with k-fold cross-validation to ensure robust model assessment.

Key Findings and Model Performance

Table 1 below presents sociodemographic characteristics of the study sample, while Table 2 compares the performance of various ensemble models.

Table 1. Sociodemographic Characteristics of Women (EDHS 2016–2019 Dataset)

*Regional percentages vary across different groups.

Table 2. Model Performance Comparison (Test Set Results)

*The Random Forest model outperformed other methods; its excellent performance metrics provide evidence that ensemble learning can serve as a decision-support tool for health policymakers and practitioners.

Community-Engaged Research in Community Health Centers

CHCs are at the forefront of providing care to socioeconomically disadvantaged populations and have the potential to reduce health disparities. Recent studies underscore the impact of community advisory boards (CABs) in driving the implementation of evidence-based interventions (EBIs) in CHCs. A mixed methods study conducted in Massachusetts combined quantitative surveys and qualitative interviews with CHC staff and CAB members to assess the quality of engagement and to develop a practical toolkit for implementing CABs.

Study Components and Toolkit Development

The research project involved several steps:

1. Quantitative Surveys: Members of CHC-led CABs completed the Research Engagement Survey Tool (REST). The survey evaluated nine engagement principles such as community focus, partner input, co-learning, and trust. High internal consistency was noted, with mean ratings generally above 4 out of 5 across quality and quantity scales.

2. Qualitative Interviews: Semi-structured interviews delved deeper into the experiences, challenges, and successes of CAB membership. Themes such as role definition, meeting facilitation, and transparency emerged as critical for effective CAB operation.

3. Cost Analysis and Structured Feedback: CHC leaders were engaged through meetings to conduct cost analyses and gather structured feedback. This informed the budgeting, staffing, and operational guidance integrated into the final toolkit.

4. Toolkit Creation: Integrating findings from surveys, interviews, and cost analyses, a “how-to” implementation strategy was developed. The toolkit addresses CAB leadership structure, member recruitment, meeting logistics, facilitation techniques, prioritization strategies, cost and sustainability planning, and evaluation.

This comprehensive approach has potential to improve the uptake and sustainability of EBIs in CHCs, ultimately enhancing community health outcomes.

Portable MRI and Ethical Considerations in Community Research

Breakthroughs in portable magnetic resonance imaging (pMRI) are transforming the landscape of neuroimaging by making it feasible to conduct imaging outside traditional hospital settings. pMRI enables scanning in community centers, homes, and remote areas—expanding access for populations that historically have been underrepresented in neuroimaging research. However, these advances raise a host of ELSI issues that must be carefully addressed.

Ethical, Legal, and Social Implications (ELSI)

Recent guidance has been produced to help researchers navigate the complexities associated with pMRI in community settings. Key considerations include:

• Research Protocol Development: Researchers must clearly articulate the purpose of scanning in a community setting, ensuring that the intended benefits extend to the local population. This involves forming community partnership agreements and engaging local stakeholders from the outset.
• Staff and Oversight Adequacy: All research personnel must be appropriately trained. Establishing an MRI Safety Committee and a Community Advisory Board (CAB) is essential for oversight of both safety and ethical standards.
• Data Management and Confidentiality: Plans for secure data handling and storage are critical, along with clear processes for managing incidental findings. Transparent informed consent procedures must be implemented to maintain trust.
• Community Engagement: Continuous input from community partners is vital. Researchers need to remain responsive to community feedback both before and during scanning, and ensure results are communicated back to the community in a meaningful way.

A practical “Portable MRI Research ELSI Checklist” has been developed to guide researchers through these stages—from creating the research protocol to post-scan community engagement.

Summary and Integration

Recent advances in ensemble learning have led to impressive accuracies in predicting adverse maternal outcomes such as short birth intervals, which can inform timely interventions. Simultaneously, community health centers are integrating structured community advisory boards into their operations to better implement evidence-based practices, while new pMRI technologies and ethical frameworks are expanding the reach of neuroimaging research. Together, these developments highlight a broader trend toward data-driven decision-making and increased community engagement in public health. The synergy of advanced analytics and effective community strategies promises improved health outcomes and a reduction in disparities.

Frequently Asked Questions (FAQ)

What is ensemble learning and why is it useful in public health research?
Ensemble learning is a machine learning technique that combines predictions from multiple models to achieve higher accuracy and robustness. In public health research, it is particularly useful for predicting complex outcomes such as short birth intervals, as models like Random Forest demonstrate excellent performance by reducing overfitting and improving generalization.

How does the Random Forest algorithm compare to other ensemble methods in predicting short birth intervals?
In the recent study using the EDHS 2016–2019 data, the Random Forest algorithm achieved 97.84% accuracy, 99.70% recall, and an AUC of 98%, outperforming other methods such as Bagging, XGBoost, and AdaBoost. Its superior performance validates its use as a reliable decision-support tool.

What are the benefits of community advisory boards (CABs) in CHCs?
CABs enhance community engagement by ensuring that implementation strategies for evidence-based interventions are informed by local perspectives. They improve the relevance and sustainability of health programs, foster trust between CHCs and their communities, and support shared decision-making.

What unique ethical challenges does portable MRI (pMRI) bring to community research?
pMRI enables neuroimaging in non-traditional settings but raises ethical, legal, and social issues. These include ensuring proper oversight and training, managing incidental findings in remote settings, maintaining data confidentiality, and engaging local communities in a respectful and transparent manner.

How can policymakers use the outcomes of ensemble learning models in public health?
Policymakers can use predictive models to identify populations at higher risk for adverse outcomes—such as short birth intervals—and tailor interventions accordingly. This data-driven approach aids in resource allocation and designing targeted early intervention programs.

References

1. Kelkay, J. M., Ramos, B. F., Bittar, R. S. M., Cal, R. V. R., Luís, L. A., Albernaz, P. L., … & [others]. (2025). Ensemble learning to predict short birth interval among reproductive-age women in Ethiopia: Evidence from EDHS 2016–2019. BMC Pregnancy and Childbirth. https://doi.org/10.1186/s12884-025-07248-1

2. [SOA Abstracts]. (2025). Selected abstracts and e-posters from the SOA24, ICC Birmingham UK, 1–3 July 2024. Journal of Intensive Care Society. https://pubmed.ncbi.nlm.nih.gov/11795607/

3. Shen, F. X. W., Wolf, S. M., Lawrenz, F., Klein, E. (2025). Portable accessible MRI in dementia research: Ethical considerations about research representation and dementia-friendly technology. Journal of Law, Medicine & Ethics. https://doi.org/10.1017/jme.2024.157

4. Shen, F. X. W., Wolf, S. M., Lawrenz, F., Comeau, D. S., Evans, B. J., Fair, D., … & Thomas, J. (2025). Conducting research with highly portable MRI in community settings: A practical guide to navigating ethical issues and ELSI checklist. Journal of Law, Medicine & Ethics. https://doi.org/10.1017/jme.2024.162

5. Maia, F. C. Z., Ramos, B. F., Bittar, R. S. M., Cal, R. V. R., et al. (2025). The near point of convergence in patients with vestibular migraine. International Archives of Otorhinolaryngology