DEEP LEARNING IN FORENSIC ODONTOLOGY: A BIBLIOMETRIC ANALYSIS

ABSTRACT Background: Forensic odontology supports human identification, age estimation, and disaster victim identification (DVI), yet conventional approaches can be affected by examiner subjectivity, population variability, and limitations in image/data quality. Bibliometrics is useful for mapping research patterns, collaboration, and thematic structures in rapidly growing fields. Deep learning is increasingly applied to dental imaging and forensic tasks. Objective: To map the research landscape of deep learning in forensic odontology using a bibliometric approach. Methods: Scopus-indexed publications (2005–2025) were retrieved using (“forensic odontology” OR “forensic dentistry” OR “dental identification”) AND (“deep learning” OR “artificial intelligence” OR “machine learning”). Data were analyzed in RStudio with bibliometrix/Biblioshiny to assess publication trends, leading sources, country contributions, author keywords, co-occurrence networks, and thematic mapping. Results: The search identified 171 documents from 89 journal sources involving 682 authors (mean 4.98 authors/document), with 23.39% international collaboration. A total of 391 author keywords were recorded; the mean document age was 2.03 years, with 16.54 citations per document and an annual growth rate of 21.02%. Publication output rose sharply after 2019, peaking in 2024–2025. Forensic Science International was the most productive source; country contributions were led by India, followed by Brazil and China. Thematic mapping positioned AI/deep learning as the methodological core, primarily linked to dental age estimation and identification using panoramic radiography/CBCT. Conclusion: Deep learning research in forensic odontology is expanding rapidly; future work should emphasize cross-population external validation and standardized data/protocols.Keywords : bibliometric, deep learning, forensic odontology ABSTRAK  Latar belakang: Odontologi forensik berperan dalam identifikasi individu, estimasi usia, dan konteks DVI, tetapi pendekatan konvensional masih dipengaruhi subjektivitas pemeriksa, variasi populasi, serta keterbatasan kualitas. Bibliometrik membantu memetakan pola, kolaborasi, dan struktur intelektual ketika literatur berkembang cepat. Deep learning saat ini makin luas digunakan untuk analisis citra dental dan aplikasi forensik. Tujuan: Menganalisis lanskap riset deep learning dalam odontologi forensik melalui pendekatan bibliometrik. Metode: Data diambil dari Scopus (2005–2025) menggunakan kunci (“forensic odontology” OR “forensic dentistry” OR “dental identification”) AND (“deep learning” OR “artificial intelligence” OR “machine learning”). Data dianalisis dengan RStudio menggunakan bibliometrix/Biblioshiny meliputi tren publikasi, sumber, negara, kata kunci, co-occurrence, dan thematic map. Hasil: Teridentifikasi 171 dokumen dari 89 sumber jurnal, melibatkan 682 penulis (rata-rata 4,98 penulis/dokumen) dengan 23,39% kolaborasi internasional. Terdapat 391 author keywords; usia dokumen rata-rata 2,03 tahun dengan sitasi rata-rata 16,54/dokumen serta pertumbuhan tahunan 21,02%. Produksi ilmiah meningkat tajam sejak 2019 dan mencapai puncak pada 2024–2025. Jurnal Forensic Science International menjadi sumber paling dominan, dan kontribusi negara dipimpin India, diikuti Brasil dan Tiongkok. Pemetaan tema menegaskan AI/deep learning sebagai pusat metodologis dengan fokus aplikasi utama pada estimasi usia gigi dan identifikasi berbasis radiografi panoramik/CBCT. Kesimpulan: Riset deep learning dalam odontologi forensik tumbuh pesat, namun penelitian selanjutnya perlu menekankan validasi eksternal lintas populasi, dan standardisasi data/protokol. Kata kunci : bibliometrik, deep learning, odontologi forensik