Aspect-Based Sentiment Analysis of Indonesian Healthcare Application Reviews Using Latent Dirichlet Allocation and IndoBERT

The increasing use of digital healthcare applications has generated large volumes of user reviews that reflect user experiences and satisfaction. This study aimed to perform Aspect-Based Sentiment Analysis (ABSA) of Indonesian healthcare application reviews using Latent Dirichlet Allocation (LDA) and IndoBERT. A total of 3,000 reviews were collected from Halodoc, Alodokter, and SATUSEHAT Mobile through the Google Play Store. After preprocessing, 1,460 valid reviews were analyzed. LDA was applied to identify discussion aspects, while IndoBERT was used to classify review sentiments into positive, neutral, and negative categories. The analysis identified seven major aspects related to consultation services, doctor responsiveness, medication delivery, payment processes, and system accessibility. Negative sentiment dominated the dataset (51.78%), followed by positive (40.82%) and neutral (7.40%) sentiments. Doctor Responsiveness and Service Quality achieved the highest positive sentiment (86.57%), whereas Login, OTP, and System Access Problems showed the highest negative sentiment (89.71%). The findings indicated that users appreciated healthcare professionals but frequently reported technical and operational issues. The study concluded that improvements in system reliability, authentication mechanisms, payment services, and medication delivery processes are essential to enhance user satisfaction with Indonesian healthcare applications