Application of Supervised Machine Learning Algorithms to Identify the Prevalence and Determinants of Spontaneous Abortion Among Ever-Married Women in Somaliland: Insights from SLDHS Data 2020
摘要
This study examined the prevalence and determinants of spontaneous abortion among ever-married women in Somaliland using the 2020 Somaliland Demographic Health Survey. The results showed that 95.85% of the respondents were aborted without intention, while 4.15% were unaborted. The results were presented using the Chi-square test with 95% confidence interval (CI) and a p value ≤ 0.05. The analysis showed that abortion rates vary across regions in Somaliland. Waqooyi Galbeed region had a statistically significant association with abortion compared to other regions (AOR = 2.4, p = 0.002). Additionally, individuals in the second wealth index category were more likely to have abortions (AOR = 2.01, p = 0.012). The study also found out that difficulties in accessing medical help are significantly related to abortion with AOR = 1.6, p value = 0.005 and personalities with more than five children are more likely to have spontaneous abortions with a statistically significant association with AOR = 1.9, p value = 0.003. Five machine learning models were used to analyze predictors of spontaneous abortion, with Random Forest and KNN being the most accurate models, with accuracy rates of 95.9% and 95.5%, respectively.