EFEKTIVITAS PEMANFAATAN ARTIFICIAL INTELLIGENCE DALAM MANAJEMEN INFORMASI KESEHATAN TERHADAP PENINGKATAN KUALITAS DATA KESEHATAN: SYSTEMATIC LITERATURE REVIEW

Abstract

Digital transformation has expanded the volume and complexity of data managed through electronic health records and health information systems, while incompleteness, inconsistency, duplication, coding errors, and poor standardization remain common. This scoping review mapped the use of artificial intelligence (AI) in health information management and examined whether reported outcomes directly measured data quality or only downstream technical performance. Reporting followed PRISMA-ScR. Publications from January 2021 to 18 July 2026 were searched in PubMed and supplementary publisher platforms. Two reviewers independently screened records and resolved disagreements by consensus. Ten sources were included and separated into three primary studies and seven secondary reviews. Direct primary evidence showed that K-nearest-neighbor imputation reduced missing values in a processed dataset and that large-language-model preprocessing improved clinical concept extraction (F1-score 0.40 to 0.61). These findings reflect improvements in dataset completeness after imputation and downstream extraction performance, not proof that missing patient information was actually collected or that clinical outcomes improved. Evidence for workflow, organizational, and real-world clinical impact remained limited. AI therefore shows potential as a data-quality support and quality-control tool, but implementation requires validation, human oversight, bias monitoring, privacy protection, and alignment with interoperability standards.