Scholarlink

<p>Unmet need for contraception remains a major public health concern across Sub-Saharan Africa, with Benin reporting persistently high rates despite national efforts to expand access. This study investigates the ability of artificial intelligence (AI) algorithms to identify women of reproductive age most vulnerable to unmet need for family planning, using nationally representative data from the 2017–2018 Demographic and Health Survey (DHS). Among 6,669 eligible women aged 15–49, the prevalence of unmet need for contraception was 37.2%, with regional hotspots such as Plateau and Collines exceeding 43%. To address class imbalance, the SMOTE algorithm was applied. Eight supervised machine learning models—including Random Forest, XGBoost, and Support Vector Machines—were benchmarked using sensitivity, precision, F-measure, and AUROC. Random Forest consistently outperformed the other models, achieving an AUROC of 1.00 and an F1 score of 0.96 under tenfold cross-validation. Variable importance assessments revealed marital status, age, and region of residence as the top predictors. High-risk profiling identified specific combinations of characteristics associated with elevated probabilities of unmet need—for example, married or cohabiting women aged 30–34 in Ouémé with seven or more children who had heard family planning messages on the radio showed a predicted probability of 87%. These findings demonstrate the potential of AI-driven approaches to enhance the precision of family planning interventions. By moving beyond conventional regression methods, this study offers scalable, equity-centered insights to guide strategic planning, community outreach, and policy formulation in reproductive health programming across Benin.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Assessment of AI algorithms in identifying vulnerable women with unmet need for family planning in Benin

  • 期刊论文

  • 发表日期：

    2026年5月18日

查看全文
  • Justin Dansou,
  • Alphonse M. Affo,
  • Fidelia A. A. Dake,
  • Pacôme E. Acotchéou,
  • Jacques Z. Saizonou

摘要

Unmet need for contraception remains a major public health concern across Sub-Saharan Africa, with Benin reporting persistently high rates despite national efforts to expand access. This study investigates the ability of artificial intelligence (AI) algorithms to identify women of reproductive age most vulnerable to unmet need for family planning, using nationally representative data from the 2017–2018 Demographic and Health Survey (DHS). Among 6,669 eligible women aged 15–49, the prevalence of unmet need for contraception was 37.2%, with regional hotspots such as Plateau and Collines exceeding 43%. To address class imbalance, the SMOTE algorithm was applied. Eight supervised machine learning models—including Random Forest, XGBoost, and Support Vector Machines—were benchmarked using sensitivity, precision, F-measure, and AUROC. Random Forest consistently outperformed the other models, achieving an AUROC of 1.00 and an F1 score of 0.96 under tenfold cross-validation. Variable importance assessments revealed marital status, age, and region of residence as the top predictors. High-risk profiling identified specific combinations of characteristics associated with elevated probabilities of unmet need—for example, married or cohabiting women aged 30–34 in Ouémé with seven or more children who had heard family planning messages on the radio showed a predicted probability of 87%. These findings demonstrate the potential of AI-driven approaches to enhance the precision of family planning interventions. By moving beyond conventional regression methods, this study offers scalable, equity-centered insights to guide strategic planning, community outreach, and policy formulation in reproductive health programming across Benin.