Volume 28, Issue 6 (Iran South Med J 2026)
Iran South Med J 2026, 28(6): 911-925 |
Back to browse issues page
Download citation:
BibTeX | RIS | EndNote | Medlars | ProCite | Reference Manager | RefWorks
Send citation to:
BibTeX | RIS | EndNote | Medlars | ProCite | Reference Manager | RefWorks
Send citation to:
Zarooj Hosseini K, Emami H, Nazari E, Golabpour A. Neural Network Performance in Drug Dose Prediction: A Systematic Review: A Systematic Review. Iran South Med J 2026; 28 (6) :911-925
URL: http://ismj.bpums.ac.ir/article-1-2469-en.html
URL: http://ismj.bpums.ac.ir/article-1-2469-en.html
1- Department of Health Information Technology and Management, School of Allied Medical Sciences, Shahid Beheshti Universi-ty of Medical Sciences, Tehran, Iran
2- Department of Health Information Technology and Management, School of Allied Medical Sciences, Shahid Beheshti Universi-ty of Medical Sciences, Tehran, Iran ,haemami@sbmu.ac.ir
3- Department of Health Informatics Technology, School of Allied Medical Sciences, Shahroud University of Medical Sciences, Shahroud, Iran
2- Department of Health Information Technology and Management, School of Allied Medical Sciences, Shahid Beheshti Universi-ty of Medical Sciences, Tehran, Iran ,
3- Department of Health Informatics Technology, School of Allied Medical Sciences, Shahroud University of Medical Sciences, Shahroud, Iran
Abstract: (215 Views)
Background: The therapeutic efficacy of an agent should be balanced against its potential adverse effects. Accurate dosage is critical for successful treatment, yet even a correctly administered dose within a therapeutic regimen fan result in death. Recent years have witnessed the emergence of machine learning approaches in medicine, offering predictive and prescriptive capabilities regarding dosage optimization. These technologies have the potential to expand the potential of individualized prescription and decrease the incidence of prescription errors.
Materials and Methods: The author conducted this systematic review, covering the years 2000 to 2025. Three major bibliographic databases were searched, and after duplicates were removed and irrelevant articles were screened out, the author included manuscripts published in English with valid methodologies that presented quantitative measures of algorithm performance. Data were extracted and analyzed, focusing on key characteristics of the algorithms under review, evaluation metrics (Accuracy, MAE, RMSE, R, and AUC), and the frequency of the algorithm use.
Results: Out of 460 initially identified studies, 18 were included in the final analysis after screening. The results indicated that artificial neural networks and their advanced variants demonstrated satisfactory performance in drug dose prediction, particularly for drugs with a narrow therapeutic window such as warfarin. Evaluation metrics including MAE, RMSE, R², Accuracy, and AUC showed moderate to high predictive performance in most studies. However, the limited use of external validation and the absence of prospective clinical evaluations were identified as major limitations.
Conclusion: The findings of this systematic review suggest that neural networks hold a substantial potential for improving the accuracy of drug dose prediction and enhancing treatment safety, especially for high-risk drugs with narrow therapeutic windows. However, the lack of external validation and prospective clinical assessments restricts their widespread clinical application. Conducting standardized clinical studies and developing more interpretable models may facilitate the practical implementation of these approaches in clinical settings.
Materials and Methods: The author conducted this systematic review, covering the years 2000 to 2025. Three major bibliographic databases were searched, and after duplicates were removed and irrelevant articles were screened out, the author included manuscripts published in English with valid methodologies that presented quantitative measures of algorithm performance. Data were extracted and analyzed, focusing on key characteristics of the algorithms under review, evaluation metrics (Accuracy, MAE, RMSE, R, and AUC), and the frequency of the algorithm use.
Results: Out of 460 initially identified studies, 18 were included in the final analysis after screening. The results indicated that artificial neural networks and their advanced variants demonstrated satisfactory performance in drug dose prediction, particularly for drugs with a narrow therapeutic window such as warfarin. Evaluation metrics including MAE, RMSE, R², Accuracy, and AUC showed moderate to high predictive performance in most studies. However, the limited use of external validation and the absence of prospective clinical evaluations were identified as major limitations.
Conclusion: The findings of this systematic review suggest that neural networks hold a substantial potential for improving the accuracy of drug dose prediction and enhancing treatment safety, especially for high-risk drugs with narrow therapeutic windows. However, the lack of external validation and prospective clinical assessments restricts their widespread clinical application. Conducting standardized clinical studies and developing more interpretable models may facilitate the practical implementation of these approaches in clinical settings.
Type of Study: Review |
Subject:
medicine and food
Received: 2025/11/18 | Accepted: 2026/02/24 | Published: 2026/06/10
Received: 2025/11/18 | Accepted: 2026/02/24 | Published: 2026/06/10
References
1. Khammarnia M, Setoodehzadeh FJHS. Medical error as a challenge in Iran’s health system. Health Scope 2017; 6(6): e61597. [DOI]
2. Isfahani P, Bazi A, Alirezaei S, et al. Medication error rates in Iranian hospitals: a meta-analysis. BMC Health Services Research 2024; 24(1): 743. [DOI]
3. Noorain, Srivastava V, Parveen B, Parveen RJCdm. Artificial intelligence in drug formulation and development: applications and future prospects. Current Drug Metabolism 2023; 24(9): 622-634. [DOI]
4. Bhatt P, Singh S, Kumar V, et al. Artificial intelligence in pharmaceutical industry: Revolutionizing drug development and delivery. Current Artificial Intelligence 2024; 2(1): E051223224198. [DOI]
5. Derraz B, Breda G, Kaempf C, et al. New regulatory thinking is needed for AI-based personalised drug and cell therapies in precision oncology. NPJ Preci-sion Oncology 2024; 8(1): 23. [DOI]
6. Huang S, Xu Q, Yang G, et al. Machine learning for prediction of drug concentrations: application and challenges. Clinical Pharmacology and Therapeutics 2025; 117(5): 1236-1247. [DOI]
7. Blasiak A, Khong J, Kee T. CURATE.AI: Optimizing Personalized Medicine with Artificial Intelligence. SLAS Technology 2020; 25(2): 95-105. [DOI]
8. Vora LK, Gholap AD, Jetha K, et al. Artificial intelligence in pharmaceutical technology and drug delivery design. Pharmaceutics 2023; 15(7): 1916. [DOI]
9. Serrano DR, Luciano FC, Anaya BJ, et al. Artificial Intelligence (AI) Applications in Drug Discovery and Drug Delivery: Revolutionizing Personalized Medicine. Pharmaceutics 2024; 16(10): 1328. [DOI]
10. Teplytska O, Ernst M, Koltermann LM, et al. Machine learning methods for precision dosing in anticancer drug therapy: A scoping review. Clin Pharmacokinet 2024; 63(9): 1221-1237. [DOI]
11. Li B, Tan K, Lao AR, et al. A comprehensive review of artificial intelligence for pharmacology research. Frontiers in genetics 2024; 15: 1450529. [DOI]
12. Aghajanpour S, Amiriara H, Esfandyari-Manesh M, et al. Utilizing machine learning for predicting drug release from polymeric drug delivery systems. Computers in Biology and Medicine 2025; 188: 109756. [DOI]
13. Emami H, Radfar R. Presentation of a novel model for evaluation of commercialization of Research and Development: Case study of the pharmaceutical biotechnology industry. IJPR 2017; 16(3): 1214-1222. [Article]
14. Varela-Rey I, Bandín-Vilar E, TojaCamba FJ, et al. Artificial Intelligence and Machine Learning Applications to Pharmacokinetic Modeling and Dose Prediction of Antibiotics: A Scoping Review. Antibiotics (Basel, Switzerland) 2024; 13(12): 1203. [DOI]
15. Mulani AO, Deshmukh M, Jadhav V, et al. Trans-forming Drug Therapy with Deep Learning: The Future of Personalized Medicine. Drug Res (Stuttg) 2025; 75(8): 326-333. [DOI]
16. Li QY, Tang BH, Wu YE, et al. Machine Learning: A New Approach for Dose Individualization. Clinical pharmacology and therapeutics 2024; 115(4): 727-744. [DOI]
17. Haidar SH, Johnson SB, Fossler MJ, et al. Modeling the Pharmacokinetics and Pharmacodynamics of a Unique Oral Hypoglycemic Agent Using Neural Net-works. Pharmaceutical Research 2002; 19(1): 87-91. [DOI]
18. Podda G. PREDICTION OF OPTIMAL WARFARIN MAINTENANCE DOSE USING ADVANCED ARTIFI-CIAL NEURAL NETWORKS. 2013. [Article]
19. Grossi E, Podda GM, Pugliano M, et al. Prediction of optimal warfarin maintenance dose using advanced artificial neural networks. Pharmacogenomics 2014; 15(1): 29-37. [DOI]
20. Zhou Q, Kwong J, Chen J, et al. Use of artificial neural network to predict warfarin individualized dosage regime in Chinese patients receiving lowintensity anticoagulation after heart valve replacement. International journal of cardiology 2014; 176(3): 1462-1464. [DOI]
21. Isma'eel HA, Sakr GE, Habib RH, et al. Improved accuracy of anticoagulant dose prediction using a pharmacogenetic and artificial neural network-based method. European journal of clinical phar-macology 2014; 70(3): 265-273. [DOI]
22. Saleh MI, Alzubiedi S. Dosage Individualization of Warfarin Using Artificial Neural Networks. Molecular Diagnosis & Therapy 2014; 18(3): 371-379. [DOI]
23. Pavani A, Naushad SM, Kumar RM, et al. Artificial neural network-based pharmacogenomic algorithm for warfarin dose optimization. Pharmacogenomics 2016; 17(2): 121-131. [DOI]
24. Tao Y, Xiang D, Zhang Y, et al. Swarm ANN/SVR-based modeling method for warfarin dose prediction in Chinese. InInternational Conference on Swarm Intelligence 2017 (pp. 351-358). Cham: Springer International Publishing. [Article]
25. Tao H, Li Q, Zhou Q, et al. A prediction study of war-farin individual stable dose after mechanical heart valve replacement: adaptive neural-fuzzy inference system prediction. BMC Surgery 2018; 18(1): 10. [DOI]
26. Li Q, Wang J, Tao H, et al. The prediction model of warfarin individual maintenance dose for patients undergoing heart valve replacement, based on the back propagation neural network. Clinical Drug Investigation 2020; 40(1): 41-53. 10. [DOI]
27. Ma W, Li H, Dong L, et al. Warfarin maintenance dose prediction for Chinese after heart valve replacement by a feedforward neural network with equal stratified sampling. Scientific Reports 2021; 11(1): 13778. [DOI]
28. Lee H, Kim HJ, Chang HW, et al. Development of a system to support warfarin dose decisions using deep neural networks. Scientific Reports 2021; 11(1): 14745. [DOI]
29. Jamali N, Razavi H, Gharib MR. Optimization of Propofol Dose Estimated During Anesthesia Through Artificial Intelligence by Genetic Algorithm: Design and Clinical Assessment. Neural Processing Letters 2022; 54(4): 3019-3043. [DOI]
30. Gu ZC, Huang SR, Dong L, et al. An Adapted Neural-Fuzzy Inference System Model Using Preprocessed Balance Data to Improve the Predictive Accuracy of Warfarin Maintenance Dosing in Patients After Heart Valve Replacement. Cardiovasc Drugs Ther 2021; 36(5): 879-889. [DOI]
31. Jahmunah V, Chen S, Lih Oh SL, et al. Automated warfarin dose prediction for Asian, American, and Caucasian populations using a deep neural network. Computers in Biology and Medicine 2023; 153. [DOI]
32. Liu Y, Yu Z, Ye X, et al. Personalized venlafaxine dose prediction using artificial intelligence technolo-gy: a retrospective analysis based on real-world da-ta. International journal of clinical pharmacy 2024; 46(4): 926-936. 10.1007/s11096-024-01729-7 [DOI]
33. Hu Z, Pan G, Wang X, et al. Intelligent algorithm based on deep learning to predict the dosage for anesthesia: A study on prediction of drug efficacy based on deep learning. Health science reports 2024; 7(5): e2113. [DOI]
34. Mohan M, Thiyagu T. Predicting Customer Purchases with TabNet: A Deep Learning Approach to Consumer Behaviour. 2025 8th International Conference on Computing Methodologies and Communication (ICCMC); Erode, India, 2025: 1915-1919. [DOI]
35. Raghuvanshi KP. A systematic literature review on the role of LSTM networks in capturing temporal dependencies in data mining algorithms. International Journal for Research in Applied Science and Engineering Technology 2024; 12(10): 1219-1224. [DOI]
36. Robeson SM, Willmott CJ. Decomposition of the mean absolute error (MAE) into systematic and unsystematic components. Plos one 2023; 18(2): e0279774. [DOI]
37. Park SH, Cha WC. Application strategies for artificial intelligence–based clinical decision support system: From the simulation to the real-world. Healthcare Informatics Research 2022; 28(3): 185-187. [DOI]
38. Ramspek CL, Jager KJ, Dekker FW, et al. External validation of prognostic models: what, why, how, when and where? Clinical Kidney Journal 2021; 14(1): 49-58. [DOI]
39. Shah RU, Bress AP, Vickers AJ. Do prediction models do more harm than good? Circulation. Cardio-vascular Quality and Outcomes 2022; 15(4): e008667. [DOI]
40. Li Q, Tao H, Wang J, et al. Warfarin maintenance dose Prediction for Patients undergoing heart valve replacement - A hybrid model with genetic algorithm and Back-Propagation neural network. Scientific Reports 2018; 8(1): 9712. [DOI]
Send email to the article author
| Rights and Permissions | |
 |
This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License. |
