This study looks at unequal antenatal care use in Ghana and proposes a fairness aware machine learning approach to address it. The goal is to predict which women are at risk of inadequate antenatal visits while ensuring that there is ethical and transparent use of artificial intelligence. Data came from the 2022 Ghana Demographic and Health Survey and included over 3,000 women with a recent live birth. Several supervised learning models were trained and tested using careful validation and balanced evaluation methods. The models were designed not only for accuracy but also for fairness across wealth, region ethnicity and religion. Explainable methods were used to show how factors such as education wealth and contact with health workers influenced predictions. Fairness testing revealed that initial predictions reflected already existing social inequalities. Bias reduction techniques improved equity without strongly reducing model performance. The final framework combines prediction fairness analysis and causal reasoning to support responsible decision making in maternal health.