PERFORMA ARTIFICIAL INTELLIGENCE-LARGE LANGUAGE MODEL PADA PENAPISAN LITERATUR PENELITIAN KOMPRESI TABLET FARMASI
Keywords:
Kecerdasan Buatan, Model Bahasa Besar (LLM, Penapisan Literatur, Manufaktur Farmasi, Kompresi TabletAbstract
Tinjauan pustaka yang efisien sangat penting untuk memajukan proses manufaktur farmasi, khususnya dalam kompresi tablet. Penelitian ini membandingkan dua Artificial Intelligence-Large Language Models (AI-LLM) untuk mengotomatisasi penapisan (screening) literatur. Model A, sebuah model baru dengan 4 miliar parameter, dibandingkan dengan Model B, sebuah model lama dengan 8 miliar parameter. Kami mengevaluasi kemampuan keduanya dalam memberikan skor relevansi (0-10) dan tingkat kepercayaan (confidence levels)terhadap suatu judul dan abstrak penelitian. Hasil penelitian menunjukkan bahwa Model A mencapai nilai rata-rata yang secara signifikan lebih tinggi (7,52 ± 1,83) dibandingkan dengan Model B (5,52 ± 3,61), dengan perbedaan yang signifikan (p < 0,000001). Meskipun Model B menunjukkan koefisien korelasi yang lebih tinggi, Model A menunjukkan keunggulan, dengan mempertahankan hubungan yang lebih linear dan terprediksi antara skor dan tingkat kepercayaan. Model B menunjukkan kesenjangan kepercayaan-akurasi (confidence-accuracy gap) yang nyata, di mana model terkadang mampu mengidentifikasi sampel dengan benar namun dengan tingkat kepercayaan yang tidak konsisten. Temuan kami menunjukkan bahwa model yang lebih baru meskipun parameternya lebih kecil dapat memberikan performa yang lebih andal untuk tugas-tugas penapisan literatur ilmiah secara otomatis.
References
Chigbu, U. E., Atiku, S. O., & Du Plessis, C. C. (2023). The Science of Literature Reviews: Searching, Identifying, Selecting, and Synthesising. Publications, 11(1), 2. https://doi.org/10.3390/publications11010002
Corradini, F., Leonesi, M., & Piangerelli, M. (2025). State of the Art and Future Directions of Small Language Models: A Systematic Review. Big Data and Cognitive Computing, 9(7), 189. https://doi.org/10.3390/bdcc9070189
Divecha, C., Tullu, M., & Karande, S. (2023). The art of referencing: Well begun is half done! Journal of Postgraduate Medicine, 69(1), 1–6. https://doi.org/10.4103/jpgm.jpgm_908_22
Doubleday, A. F., Cheverko, C. M., Bolgova, O., Mavrych, V., Mohamed, F. R. R., Westrick, J., Juarez, L., Rush, E., Solka, K. A., Byram, J. N., Beacker, R., Gomez, V., Ganeng, B. K. A., Hoffman, L. A., Roach, V. A., Brown, K. M., DeVaul, N., Garnett, C. N., Herriott, H. L., … Wilson, A. B. (2026). Temporal trends in large language model (LLM) accuracy: A meta-analysis of multiple-choice question performance in dentistry and dental education. Journal of Dentistry, 171, 106724. https://doi.org/10.1016/j.jdent.2026.106724
Forstmeier, W., Wagenmakers, E., & Parker, T. H. (2017). Detecting and avoiding likely false‐positive findings – a practical guide. Biological Reviews, 92(4), 1941–1968. https://doi.org/10.1111/brv.12315
Li, D., Wu, L., Zhang, M., Shpyleva, S., Lin, Y.-C., Huang, H.-Y., Li, T., & Xu, J. (2024). Assessing the performance of large language models in literature screening for pharmacovigilance: A comparative study. Frontiers in Drug Safety and Regulation, 4, 1379260. https://doi.org/10.3389/fdsfr.2024.1379260
McIntosh, F., Murina, S., Chen, L., Vargas, H. A., & Becker, A. S. (2025). Keeping private patient data off the cloud: A comparison of local LLMs for anonymizing radiology reports. European Journal of Radiology Artificial Intelligence, 2, 100020. https://doi.org/10.1016/j.ejrai.2025.100020
Nguyen, D.-T., & Kim, J.-M. (2025). Tool Wear Detection Using Novel Acoustic Emission Features and a Two-Stage Mann-Whitney U Test. Applied Acoustics, 240, 110952. https://doi.org/10.2139/ssrn.4996667
Oami, T., Okada, Y., & Nakada, T. (2024). Performance of a Large Language Model in Screening Citations. JAMA Network Open, 7(7), e2420496. https://doi.org/10.1001/jamanetworkopen.2024.20496
Olmsted, J. (2024). Research Reliability and Validity: Why do they matter? Journal of Dental Hygiene, 98(6), 53–57.
Podder, S., Date, H., & Murthy, S. (2026). Green prompt engineering for sustainable generative AI. Environmental Science and Ecotechnology, 30, 100684. https://doi.org/10.1016/j.ese.2026.100684
Qian, S. S., Refsnider, J. M., Moore, J. A., Kramer, G. R., & Streby, H. M. (2020). All tests are imperfect: Accounting for false positives and false negatives using Bayesian statistics. Heliyon, 6(3), e03571. https://doi.org/10.1016/j.heliyon.2020.e03571
Rico, I. C., & Espada, J. P. (2025). Expert system for extracting keywords in educational texts and textbooks based on transformers models. Expert Systems with Applications, 282, 127735. https://doi.org/10.1016/j.eswa.2025.127735
Rohrschneider, D., Pehlke, M., Handmann, U., & Jansen, M. (2026). LLM-based JSON Mapping and Blockchain Integration for Digital Product Passports. Digital Business, 6(1), 100167. https://doi.org/10.1016/j.digbus.2026.100167
Tran, A. T., & Klinken-Uth, S. (2025). Influences of variations of the amount of compressed material on compressibility, tabletability and compactability. Journal of Pharmaceutical Sciences, 114(7), 103831. https://doi.org/10.1016/j.xphs.2025.103831
Wiest, I. C., Wolf, F., Leßmann, M.-E., Van Treeck, M., Ferber, D., Zhu, J., Boehme, H., Bressem, K. K., Ulrich, H., Ebert, M. P., & Kather, J. N. (2025). A software pipeline for medical information extraction with large language models, open source and suitable for oncology. Npj Precision Oncology, 9(1), 313. https://doi.org/10.1038/s41698-025-01103-4
Yang, B., Xiao, H., Zeng, Z., Lai, S., Han, J., Tan, Y., & Ni, Y. (2025). Boosting expertise and efficiency in LLM: A knowledge-enhanced framework for construction support. Alexandria Engineering Journal, 130, 525–542. https://doi.org/10.1016/j.aej.2025.09.029
Zakowiecki, D., Kukuls, K., Cal, K., Pelloux, A., & Mohylyuk, V. (2025). Investigating the Mechanical Behaviour of Viscoelastic and Brittle Pharmaceutical Excipients During Tabletting: Revealing the Unobvious Potential of Advanced Compaction Simulation. Pharmaceutics, 17(12), 1606. https://doi.org/10.3390/pharmaceutics17121606
Downloads
Published
How to Cite
Issue
Section
License
This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.
