Kenyan Neonatal Mortality Risk Predictor: Protocol for a User-Centered Design Evaluation

Comprehensive Summary

Nyatuka et al. investigate how machine learning (ML) based neonatal risk predictors can be used in Kenyan healthcare facilities to identify high-risk newborns and decrease the number of neonatal deaths. Neonatal mortality is a major public health challenge in both low and middle-income countries (LMIC). Specifically, sub-Saharan Africa, is a country where limited resources, understaffed facilities, and delays in newborn risk assessments often prevent timely maternal care. Nyatuka et al. conducted a mixed methods, 3 phase study. Phase 1 involved interviewing staff, including neonatal department heads, facility managers, and frontline healthcare workers, to understand potential barriers for implementing ML models. Phase 2 involved using a paper-based neonatal risk tool for four months to evaluate how effectively it could be integrated into clinical settings. Phase 3 involved administering surveys to assess the tool’s usability, perceived usefulness, and overall feasibility in clinical practice. Initial observations from the ongoing study suggest that the tool can be integrated into existing triage systems, allowing high-risk neonates to be identified within the first 48 hours of birth. Healthcare workers reported positive experiences, noting the ML model’s relevance and usability. Although data analysis is ongoing, these preliminary observations suggest that contextually designed ML tools may enhance the speed and accuracy of clinical decision-making in LMIC settings.

Outcomes and Implications

Neonatal mortality in LMICs is significantly high, with many deaths resulting from preventable conditions including preterm birth, infections, and complications during delivery. Nyatuka et al.’s findings suggest that ML based neonatal risk predictors could improve early identification of high-risk infants, help prioritize clinical interventions, and make more efficient use of limited healthcare resources. Implementing these tools into clinical practices could reduce neonatal mortality and support efforts to meet national health priorities, including the Sustainable Development Goal 3.2 (SDG 3.2). With testing completed in 2025 and data analysis expected by the middle of 2026, this work provides evidence-based guidance for integrating data-driven solutions into routine neonatal care in Kenya and similar LMIC settings, offering a practical step toward stronger and more responsive health systems.