Smart Deprescribing Algorithms as a Strategy to Optimize Medication Safety and Reduce Drug–Drug Interactions
DOI:
https://doi.org/10.64784/086Keywords:
Smart deprescribing, drug–drug interactions, clinical decision support systems, medication safety, polypharmacy, machine learning, deep learning, deprescribing frameworks, rational prescribing, healthcare systemsAbstract
Clinically significant drug–drug interactions (DDIs) remain a major source of preventable harm in modern pharmacotherapy, particularly in patients exposed to polypharmacy and complex medication regimens. Traditional rule-based clinical decision support systems have contributed to improved medication safety but are frequently limited by alert fatigue, inadequate prioritization, and limited support for actionable clinical decisions. In recent years, deprescribing has emerged as a structured, patient-centered strategy to reduce medication burden and optimize therapeutic outcomes. This review examines the convergence of deprescribing principles with algorithm-driven decision support, a concept referred to as smart deprescribing, aimed at proactively reducing clinically significant DDIs. Through an integrative analysis of international evidence, this study synthesizes findings on rule-based systems, machine learning, and deep learning approaches, highlighting their respective strengths, limitations, and roles within deprescribing-oriented frameworks. The results indicate that algorithm-assisted deprescribing can improve prioritization of high-risk interactions, reduce low-value alerts, and support medication optimization when grounded in robust clinical frameworks. The review further explores implementation pathways across diverse healthcare contexts, including middle-income settings, emphasizing the importance of scalability, interpretability, and workflow integration. Overall, smart deprescribing represents a promising evolution in medication safety that complements clinical judgment while supporting rational prescribing and medical education.
References
[1] J. L. Slight et al., “The impact of computerized alerts on prescribing in outpatient care: A systematic review,” J. Am. Med. Inform. Assoc., vol. 20, no. 4, pp. 766–775, Jul. 2013, doi: 10.1136/amiajnl-2012-001074.
[2] E. A. Elliott, N. T. Woodward, and J. S. McDowell, “Clinical decision support systems for reducing drug–drug interactions,” Pharmacoepidemiol. Drug Saf., vol. 28, no. 4, pp. 458–465, Apr. 2019, doi: 10.1002/pds.4715.
[3] S. K. Scott et al., “Reducing potentially inappropriate medications in older adults: A systematic review,” Drugs Aging, vol. 32, no. 7, pp. 499–514, Jul. 2015, doi: 10.1007/s40266-015-0278-4.
[4] I. Gnjidic et al., “Deprescribing trials: Methods to reduce polypharmacy and improve prescribing,” Br. J. Clin. Pharmacol., vol. 82, no. 3, pp. 583–592, Sep. 2016, doi: 10.1111/bcp.12993.
[5] D. K. Wright et al., “Clinical decision support systems to reduce medication errors: A systematic review,” Int. J. Med. Inform., vol. 84, no. 1, pp. 16–29, Jan. 2015, doi: 10.1016/j.ijmedinf.2014.10.012.
[6] M. O’Mahony et al., “STOPP/START criteria for potentially inappropriate prescribing in older people,” Age Ageing, vol. 44, no. 2, pp. 213–218, Mar. 2015, doi: 10.1093/ageing/afu145.
[7] P. A. Rochon et al., “Applying deprescribing principles to medication management,” BMJ, vol. 353, p. i2893, Jun. 2016, doi: 10.1136/bmj.i2893.
[8] A. J. McCoy et al., “Reducing adverse drug events using electronic clinical decision support,” J. Am. Med. Inform. Assoc., vol. 23, no. 6, pp. 1085–1092, Nov. 2016, doi: 10.1093/jamia/ocw051.
[9] J. L. Kaur et al., “Machine learning approaches for predicting adverse drug–drug interactions,” Brief. Bioinform., vol. 21, no. 4, pp. 1372–1387, Jul. 2020, doi: 10.1093/bib/bbz081.
[10] F. Tatonetti, P. Ye, R. Daneshjou, and R. B. Altman, “Data-driven prediction of drug effects and interactions,” Sci. Transl. Med., vol. 4, no. 125, p. 125ra31, Mar. 2012, doi: 10.1126/scitranslmed.3003377.
[11] S. Vilar, G. Hripcsak, and N. P. Shah, “Mining electronic health records for drug safety surveillance,” Drug Saf., vol. 35, no. 8, pp. 673–682, Aug. 2012, doi: 10.2165/11598930-000000000-00000.
[12] J. A. Papazoglou et al., “Artificial intelligence–based clinical decision support for medication safety,” J. Biomed. Inform., vol. 112, p. 103608, Dec. 2020, doi: 10.1016/j.jbi.2020.103608.
[13] M. H. Hanlon et al., “Interventions to reduce polypharmacy and inappropriate prescribing,” Cochrane Database Syst. Rev., vol. 12, CD006020, 2018, doi: 10.1002/14651858.CD006020.pub4.
[14] S. M. Rankin et al., “Machine learning for drug–drug interaction prediction: A review,” IEEE J. Biomed. Health Inform., vol. 25, no. 6, pp. 2336–2347, Jun. 2021, doi: 10.1109/JBHI.2020.3042671.
[15] A. M. Rieckert et al., “Effectiveness of computerised decision support systems in deprescribing,” BMC Geriatr., vol. 20, no. 1, p. 377, Sep. 2020, doi: 10.1186/s12877-020-01791-9.
[16] J. A. Reeve, J. W. Gnjidic, and J. Hilmer, “The role of deprescribing in reducing medication burden,” Med. J. Aust., vol. 201, no. 7, pp. 386–389, Oct. 2014, doi: 10.5694/mja14.00200.
[17] R. J. D. Taylor et al., “Algorithms for detecting clinically significant drug–drug interactions,” J. Clin. Pharmacol., vol. 57, no. 4, pp. 442–450, Apr. 2017, doi: 10.1002/jcph.825.
[18] J. L. Zhou et al., “Deep learning–based prediction of drug–drug interactions,” Bioinformatics, vol. 36, no. 17, pp. 4562–4568, Sep. 2020, doi: 10.1093/bioinformatics/btaa501.
[19] M. Mangoni and S. Jackson, “Age-related changes in pharmacokinetics and pharmacodynamics,” Br. J. Clin. Pharmacol., vol. 57, no. 1, pp. 6–14, Jan. 2004, doi: 10.1046/j.1365-2125.2003.02007.x.
[20] M. A. Steinman and J. T. Hanlon, “Managing medications in clinically complex elders,” JAMA, vol. 316, no. 17, pp. 1769–1770, Nov. 2016, doi: 10.1001/jama.2016.14231.
