Academic Journal
Detection of cancer recurrence from Thai-English electronic medical records using sentence embeddings.
| Title: | Detection of cancer recurrence from Thai-English electronic medical records using sentence embeddings. |
|---|---|
| Authors: | Sangariyavanich E; National Cancer Institute of Thailand, Bangkok, Thailand.; Department of Clinical Epidemiology and Biostatistics, Faculty of Medicine Ramathibodi Hospital, Mahidol University, Bangkok, Thailand., Ponthongmak W; Department of Clinical Epidemiology and Biostatistics, Faculty of Medicine Ramathibodi Hospital, Mahidol University, Bangkok, Thailand wanchana.pon@mahidol.ac.th., Theera-Ampornpunt N; Department of Clinical Epidemiology and Biostatistics, Faculty of Medicine Ramathibodi Hospital, Mahidol University, Bangkok, Thailand., Tangchitnob N; Department of Clinical Epidemiology and Biostatistics, Faculty of Medicine Ramathibodi Hospital, Mahidol University, Bangkok, Thailand., McKay GJ; Centre for Public Health, Queen's University Belfast, Belfast, Northern Ireland, UK., Thakkinstian A; Department of Clinical Epidemiology and Biostatistics, Faculty of Medicine Ramathibodi Hospital, Mahidol University, Bangkok, Thailand. |
| Source: | BMJ health & care informatics [BMJ Health Care Inform] 2026 Jul 02; Vol. 33 (1). Date of Electronic Publication: 2026 Jul 02. |
| Publication Type: | Journal Article |
| Language: | English |
| Journal Info: | Publisher: BMJ Publishing Country of Publication: England NLM ID: 101745500 Publication Model: Electronic Cited Medium: Internet ISSN: 2632-1009 (Electronic) Linking ISSN: 26321009 NLM ISO Abbreviation: BMJ Health Care Inform Subsets: MEDLINE |
| Imprint Name(s): | Original Publication: London : BMJ Publishing, [2019]- |
| MeSH Terms: | Neoplasm Recurrence, Local*/diagnosis , Electronic Health Records* , Natural Language Processing*, Humans ; Thailand |
| Abstract: | Objective: This study developed and validated monolingual and bilingual sentence-bidirectional encoder representations from transformers (SBERT) models for detecting cancer recurrence within Thai-English electronic medical records (EMRs) from Thai cancer hospitals. Method: A multicentre dataset of 32 436 documents from 1250 patients was used for model development. External validation involved an independent dataset of 9244 documents from 384 patients across two Thai cancer hospitals. Performance was benchmarked against a fine-tuned PubMedBERT (MetBERT). Results: The development dataset included breast (43.9%), colorectal (12.1%), cervical (28.0%) and head and neck (16.0%) cancers. MetBERT achieved the highest area under the precision-recall curve (AUPRC) for locoregional versus no recurrence (11.1%) and locoregional versus distant recurrence (91.7%), while monolingual-SBERT excelled at distant versus no recurrence (32.0%). External validation demonstrated MetBERT superiority for locoregional versus no recurrence (9.30%-21.50%). For distant versus no recurrence, bilingual-SBERT performed best with AUPRC 17.55%-24.39%. While MetBERT led in distinguishing locoregional versus distant recurrence (88.30%-94.70%), bilingual-SBERT demonstrated robust external validation performance (AUPRC 85.25%-91.80%). Discussion: Low AUPRC values (9%-32%) reflect the extreme class imbalance in real-world data (~1% recurrence prevalence). Despite this, fine-tuned MetBERT achieved highest performance, while bilingual-SBERT demonstrated superior robustness during external validation. This validates sentence embedding models for handling mixed Thai-English medical records in multilingual clinical environments. Conclusion: Sentence embedding frameworks provide a practical, generalisable solution for detecting cancer recurrence within multilingual EMRs. Despite text-length constraints, these models are suitable for clinical integration as a screening tool for cancer registry workflows. (© Author(s) (or their employer(s)) 2026. Re-use permitted under CC BY-NC. No commercial re-use. See rights and permissions. Published by BMJ Group.) |
| Competing Interests: | Competing interests: None declared. |
| References: | AMIA Jt Summits Transl Sci Proc. 2022 May 23;2022:331-338. (PMID: 35854741) JMIR Med Inform. 2022 Sep 2;10(9):e37770. (PMID: 35981230) Science. 2015 Jul 17;349(6245):261-6. (PMID: 26185244) JCO Clin Cancer Inform. 2019 Oct;3:1-12. (PMID: 31584836) JAMA Oncol. 2022 Mar 01;8(3):420-444. (PMID: 34967848) BMC Bioinformatics. 2018 Dec 28;19(Suppl 17):498. (PMID: 30591037) Stud Health Technol Inform. 2022 May 25;294:93-97. (PMID: 35612023) J Affect Disord. 2024 Oct 15;363:340-347. (PMID: 39029695) Sci Data. 2016 May 24;3:160035. (PMID: 27219127) JCO Clin Cancer Inform. 2021 Apr;5:469-478. (PMID: 33929889) J Healthc Inform Res. 2019;3:283-299. (PMID: 33225204) Nat Med. 2024 Apr;30(4):1134-1142. (PMID: 38413730) JCO Clin Cancer Inform. 2024 Dec;8:e2400107. (PMID: 39705642) Nat Rev Clin Oncol. 2019 May;16(5):312-325. (PMID: 30700859) Radiology. 2021 Oct;301(1):115-122. (PMID: 34342503) J Surg Oncol. 2015 May;111(6):657-62. (PMID: 25690414) |
| Contributed Indexing: | Keywords: Electronic Health Records; Natural Language Processing |
| Entry Date(s): | Date Created: 20260702 Date Completed: 20260702 Latest Revision: 20260813 |
| Update Code: | 20260814 |
| PubMed Central ID: | PMC13330879 |
| DOI: | 10.1136/bmjhci-2025-101997 |
| PMID: | 42392671 |
| Database: | MEDLINE |
| ISSN: | 2632-1009 |
|---|---|
| DOI: | 10.1136/bmjhci-2025-101997 |