Expert System for Pregnancy Risk Diagnosis Using Decision Tree and Dempster-Shafer Method

Abstract

The high Maternal Mortality Rate (MMR) remains a severe concern in maternal healthcare. One of the reasons is the delay in recognizing early danger signs during pregnancy. To address this issue, there is a proposed solution in the form of developing an expert system aimed at swiftly and efficiently diagnosing pregnancy risks in pregnant women using the Decision Tree and Dempster Shafer methods. The Decision Tree method is employed for symptom classification, while Dempster Shafer provides confidence values for existing facts. This research collects data from the dataset, the Poedji Rochjati Score Card (KSPR), and qualitative data through expert interviews. From the collected data, knowledge acquisition processes are then carried out to extract knowledge using the ID3 Decision Tree and combine all symptoms from the gathered data. The processed data is then represented as a decision tree and assigned confidence values. The development of this expert system utilizes the Laravel framework with PHP language and MySQL database. System validation involves patients as participants and midwives as experts and testers. Testing was conducted on March 13 and 16, 2024, involving 16 patients at the Gatak Community Health Center. The system evaluation results show an accuracy rate of 93.75%. This value indicates that the system can operate effectively. Thus, it can be recommended for use in diagnosing pregnancy risks.