Developing a Local AI Cabot for Real-Time Detection of Captopril Drug-Drug Interactions in Cardiovascular Therapy
Keywords:
Captopril, Drug Interactions, Polypharmacy, Large Language Models, Artificial IntelligenceAbstract
Cardiovascular diseases (CVDs), including hypertension and heart failure, are major contributors to global morbidity and mortality, with polypharmacy being a common approach to managing these conditions. However, polypharmacy increases the risk of drug-drug interactions (DDIs), which can result in decreased efficacy, toxicity, and adverse drug reactions. This study aimed to develop and evaluate a local Cabot based on a Large Language Model (LLM) to detect Captopril drug interactions with high accuracy and specificity, ensuring data privacy and computational efficiency. The Cabot was developed using Python, LangChain, FastAPI, and a local LLM for Captopril interaction detection, with 5-fold cross-validation used to evaluate its performance using secondary data from DrugBank and Drugs.com. The system was integrated with PowerShell and ngrok to enable secure, local deployment. The Cabot achieved 100% accuracy and 100% specificity in Cycle 1, with a slight decrease in Cycle 3 to 96.7% accuracy and 92.9% specificity due to one false positive, but no false negatives were observed. The system showed an average 5-fold accuracy of 98.7% and specificity of 98.0%, confirming its potential for real-time DDI detection. The local model ensured data privacy and computational efficiency, offering a robust alternative to cloud-based systems. The Cabot developed in this study shows promise in detecting Captopril DDIs with high accuracy and specificity in cardiovascular therapy, and future improvements should focus on expanding the model’s capabilities to handle a broader range of medications and refine its detection algorithms.
References
[1] WHO, “WHO: Kardiovaskular dan Pembuluh Darah Jadi Penyebab Kematian Utama Secara Global | tempo.co,” Tempo. Accessed: Oct. 15, 2025. [Online]. Available: https://www.tempo.co/gaya-hidup/who-kardiovaskular-dan-pembuluh-darah-jadi-penyebab-kematian-utama-secara-global-395070
[2] Kemenkes, “Survei Kesehatan Indonesia (SKI) 2023,” Badan Kebijakan Pembangunan Kesehatan | BKPK Kemenkes. Accessed: Oct. 24, 2025. [Online]. Available: https://www.badankebijakan.kemkes.go.id/hasil-ski-2023/
[3] B. Papotti, C. Marchi, M. P. Adorni, and F. Potì, “Drug-drug interactions in polypharmacy patients: The impact of renal impairment,” Curr. Res. Pharmacol. Drug Discov., vol. 2, p. 100020, 2021, doi: 10.1016/j.crphar.2021.100020.
[4] R. W. Hoel, R. M. Giddings Connolly, and P. Y. Takahashi, “Polypharmacy Management in Older Patients,” Mayo Clin. Proc., vol. 96, no. 1, pp. 242–256, Jan. 2021, doi: 10.1016/j.mayocp.2020.06.012.
[5] M. Molokhia and A. Majeed, “Current and future perspectives on the management of polypharmacy,” BMC Fam. Pract., vol. 18, no. 1, p. 70, Dec. 2017, doi: 10.1186/s12875-017-0642-0.
[6] A. G. Roberts and M. E. Gibbs, “Mechanisms and the clinical relevance of complex drug–drug interactions,” Clin. Pharmacol. Adv. Appl., vol. Volume 10, pp. 123–134, Sep. 2018, doi: 10.2147/CPAA.S146115.
[7] S. Kim, E. Kim, and H. S. Suh, “Cost-Effectiveness of an Opioid Abuse–Prevention Program Using the Narcotics Information Management System in South Korea,” Value Health, vol. 24, no. 2, pp. 174–181, 2021, doi: 10.1016/j.jval.2020.12.002.
[8] I. Y. Kusuma, P. O. D. Megasari, and L. Sukiatno, “Identifikasi potensi interaksi obat pada pasien hipertensi: studi retrospektif resep polifarmasi di Apotek Karya Sehat Purwokerto,” Viva Med. J. Kesehat. Kebidanan Dan Keperawatan, vol. 12, no. 1, pp. 72–80, 2018.
[9] K. Ahmad et al., “Public health supply chain for iron and folic acid supplementation in India: Status, bottlenecks and an agenda for corrective action under Anemia Mukt Bharat strategy,” PLoS ONE, vol. 18, no. 2 February, 2023, doi: 10.1371/journal.pone.0279827.
[10] S. Bhullar, A. Shah, and N. Dhalla, “Mechanisms for the development of heart failure and improvement of cardiac function by angiotensin-converting enzyme inhibitors,” Scr. Med. (Brno), vol. 53, no. 1, pp. 51–76, 2022, doi: 10.5937/scriptamed53-36256.
[11] Y.-H. Du, C.-J. Guan, L.-Y. Li, and P. Gan, “Efficacy and safety of angiotensin converting enzyme inhibitors and angiotensin II receptor blockers in diabetic nephropathy: a systematic review and meta-analysis,” Ann. Palliat. Med., vol. 11, no. 3, pp. 1093–1101, Mar. 2022, doi: 10.21037/apm-22-212.
[12] N. Swekwi, “The effects of co-prescription of warfarin and cardiovascular medicines among atrial fibrillation patients and/or deep vein thrombosis patients in Queensland, Australia,” Oct. 01, 2020, Griffith University. doi: 10.25904/1912/3978.
[13] A. Cesaro, S. C. Hoffman, P. Das, and C. de la Fuente-Nunez, “Challenges and applications of artificial intelligence in infectious diseases and antimicrobial resistance,” Npj Antimicrob. Resist., vol. 3, no. 1, p. 2, 2025.
[14] R. Qardafil, “Peningkatan Efektivitas Kualitas Chatbot Jala Tech Melalui Implementasi Chat GPT, Auto-GPT, Dan Langchain,” PhD Thesis, Universitas Islam Indonesia, 2023.
[15] Y. N. Pratiwi, Z. Shafa, T. Nike, and P. Perawati, “Mengolah Kata, Menyembuhkan Luka:: Pentingnya Bahasa Indonesia dalam hal Komunikasi Praktik Kedokteran,” Indones. Res. J. Educ., vol. 5, no. 1, pp. 757–764, 2025.
[16] S. Yessy Asri et al., A Large Language Model (Llm) Penerapan Large Language Model (Llm) Dalam Chatbot. Uwais Inspirasi Indonesia, 2025.
[17] A. Osheba, A. Abou-El-Ela, O. Adel, Y. Maaod, T. Abuhmed, and S. El-Sappagh, “Leveraging Large Language Models for Smart Pharmacy Systems: Enhancing Drug Safety and Operational Efficiency,” in 2025 19th International Conference on Ubiquitous Information Management and Communication (IMCOM), Bangkok, Thailand: IEEE, Jan. 2025, pp. 1–8. doi: 10.1109/IMCOM64595.2025.10857582.
[18] P. K. Narayanan, “Engineering Machine Learning and Data REST APIs using FastAPI,” in Data Engineering for Machine Learning Pipelines, Berkeley, CA: Apress, 2024, pp. 323–359. doi: 10.1007/979-8-8688-0602-5_10.
[19] E. M. Tosca, L. Aiello, A. De Carlo, and P. Magni, “Pharmacometrics in the Age of Large Language Models: A Vision of the Future,” Pharmaceutics, vol. 17, no. 10, p. 1274, Sep. 2025, doi: 10.3390/pharmaceutics17101274.
[20] B. K. Kalita, K. A. Singh, and M. Kalita, “Fivefold Cross-Validation Approach in Evaluating the Robustness of Machine Learning Models for Prediction of Esophageal Cancer,” Indian J. Community Med., vol. 51, no. Suppl 1, pp. S180–S188, Feb. 2026, doi: 10.4103/ijcm.ijcm_454_24.
[21] H. Khude and P. Shende, “AI-driven clinical decision support systems: Revolutionizing medication selection and personalized drug therapy,” Adv. Integr. Med., vol. 12, no. 4, p. 100529, Dec. 2025, doi: 10.1016/j.aimed.2025.100529.
[22] F. Y. Al-Ashwal, M. Zawiah, L. Gharaibeh, R. Abu-Farha, and A. N. Bitar, “Evaluating the Sensitivity, Specificity, and Accuracy of ChatGPT-3.5, ChatGPT-4, Bing AI, and Bard Against Conventional Drug-Drug Interactions Clinical Tools,” Drug Healthc. Patient Saf., vol. Volume 15, pp. 137–147, Sep. 2023, doi: 10.2147/DHPS.S425858.
[23] Z. Ş. Aksoyalp and B. R. Erdoğan, “COMPARATIVE EVALUATION OF ARTIFICIAL INTELLIGENCE AND DRUG INTERACTION TOOLS: A PERSPECTIVE WITH THE EXAMPLE OF CLOPIDOGREL,” Ank. Univ. Eczacilik Fak. Derg., vol. 48, no. 3, pp. 22–22, Aug. 2024, doi: 10.33483/jfpau.1460173.
[24] M. W. Shiferaw, T. Zheng, A. Winter, L. A. Mike, and L.-N. Chan, “Assessing the accuracy and quality of artificial intelligence (AI) chatbot-generated responses in making patient-specific drug-therapy and healthcare-related decisions,” BMC Med. Inform. Decis. Mak., vol. 24, no. 1, p. 404, Dec. 2024, doi: 10.1186/s12911-024-02824-5.
[25] S. M. AlJassas, N. A. Alzaidi, S. M. Algharqan, S. H. Aljawad, B. M. Alsofyani, and F. Ali Hadadi, “Evaluating the Impact of Drug-Drug Interactions in Polypharmacy Strategies for Safe Medication Management,” J. Healthc. Sci., vol. 04, no. 12, pp. 1051–1056, 2024, doi: 10.52533/JOHS.2024.41255.
[26] Rishi Reddy Kothinti, “Deep learning in healthcare: Transforming disease diagnosis, personalized treatment, and clinical decision-making through AI-driven innovations,” World J. Adv. Res. Rev., vol. 24, no. 2, pp. 2841–2856, Nov. 2024, doi: 10.30574/wjarr.2024.24.2.3435.
[27] H.-X. Huang, K.-K. Liu, L. Zhang, Y.-M. Huang, W.-H. Liu, and J. Xiao, “Influencing factors of end-of-dose failure in patients with cancer pain after oral oxycodone sustained-release tablets: A retrospective, case-control study,” Jpn. J. Clin. Oncol., vol. 51, no. 6, pp. 932–941, 2021, doi: 10.1093/jjco/hyab002.
