<![CDATA[Unstructured medical notes in SOAP format are crucial assets for clinical analysis; however, their automated processing in the Indonesian language remains a significant challenge due to limited support from global NLP technologies. This study evaluates the integration of Apache cTAKES and the Unified Medical Language System (UMLS) to extract medical information from Indonesian electronic health records. The primary obstacle lies in the cTAKES architecture, which is optimized for English, causing direct application to Indonesian texts to yield a very low detection rate (Recall) of only 17.9%. As a pragmatic solution to bridge this linguistic barrier, this research proposes a preprocessing pipeline based on automatic translation using the Google Translate API prior to the cTAKES extraction process. The evaluation was conducted on a dataset of 50 SOAP-format medical records identifying 840 medical entities. Experimental results demonstrate that the automatic translation approach significantly improves entity detection, achieving a Recall of 90.2% and an F1-Score of 93.4%. Despite challenges such as information loss from local medical abbreviations and translation ambiguities, this study proves that automatic translation serves as an effective transitional strategy in resource-limited environments. This approach not only supports clinical information extraction but also enables the automatic mapping of medical terminology to international standards such as ICD-10, SNOMED-CT, and RxNorm to foster national health data interoperability.]]>
