{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,10,14]],"date-time":"2026-10-14T23:14:07Z","timestamp":1792019647950,"version":"4.4.0"},"reference-count":20,"publisher":"MDPI AG","issue":"1","license":[{"start":{"date-parts":[[2026,3,22]],"date-time":"2026-03-22T00:00:00Z","timestamp":1774137600000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"funder":[{"name":"AGU RCSI-MUB Joint research","award":["2019-2"],"award-info":[{"award-number":["2019-2"]}]}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":["Medical Sciences"],"abstract":"<jats:p>Background: Warfarin pharmacogenomics is critical due to its narrow therapeutic index and significant interpatient variability. While machine learning (ML) can predict anticoagulation control status (ACS), its \u201cblack-box\u201d nature limits clinical translatability. Explainable Artificial Intelligence (XAI) addresses this by providing interpretable insights. This study applied ML and XAI to a warfarin pharmacogenomic dataset to predict poor ACS and explain model decisions. Methods: A post hoc analysis was conducted on a cross-sectional dataset of 232 patients receiving warfarin for \u22656 months. Data included age, gender, interacting drugs, SAMe-TT2R2 score, and genotypes for CYP2C9, VKORC1, and CYP4F2. Poor ACS was defined as time in therapeutic range (TTR) &lt; 70%. The dataset was split into training (70%) and testing (30%) cohorts. Three models, Random Forest, XGBoost, and Logistic Regression, were developed and evaluated using AUC-ROC, sensitivity, and specificity. XAI techniques, including permutation importance and SHapley Additive exPlanations (SHAP), were employed for global and local interpretability. Results: Of 232 patients, 141 (60.8%) had poor ACS. XGBoost and Random Forest demonstrated comparable predictive accuracy (AUC-ROC: 0.67), outperforming Logistic Regression. Sensitivity was 0.83 and 0.79 for XGBoost and Random Forest, respectively. However, specificity was modest for both ensemble methods (Random Forest: 0.48; XGBoost: 0.41) and extremely low for Logistic Regression (0.04), indicating poor discrimination, particularly for identifying patients with adequate anticoagulation control. Globally, important predictors included the age, SAMe-TT2R2 score, CYP2C9 (*2\/*2), female gender, and VKORC1 (C\/T). XAI revealed predictions were primarily driven by VKORC1, CYP4F2, SAMe-TT2R2 scores, and drug interactions. Concordance between XAI predictions and actual ACS was 78% for adequate and 88.6% for poor ACS. SHAP analysis showed VKORC1 provided a stable risk signal (mean absolute SHAP: 1.44 \u00b1 0.49 in concordant cases), while CYP2C9 was a high-variance, high-impact driver of discordance (mean SHAP: 3.44 \u00b1 3.79 in discordant cases). Conclusions: ML models, particularly ensemble methods, show modest ability to predict poor warfarin control with limited ability to correctly identify patients with adequate control from our dataset. XAI transforms these models into interpretable tools, with SHAP analysis attributing predictions to specific genetic and clinical features. While predictive accuracy remains modest, this approach enhances transparency and provides a foundation for generating hypotheses that may ultimately support clinical decision-making in pharmacogenomic-guided warfarin therapy.<\/jats:p>","DOI":"10.3390\/medsci14010156","type":"journal-article","created":{"date-parts":[[2026,3,23]],"date-time":"2026-03-23T08:47:42Z","timestamp":1774255662000},"page":"156","update-policy":"https:\/\/doi.org\/10.3390\/mdpi_crossmark_policy","source":"Crossref","is-referenced-by-count":1,"title":["Explainable Artificial Intelligence Unravels the Possible Distinct Roles of VKORC1 and CYP2C9 in Predicting Warfarin Anticoagulation Control"],"prefix":"10.3390","volume":"14","author":[{"ORCID":"https:\/\/orcid.org\/0000-0003-3811-6503","authenticated-orcid":false,"given":"Kannan","family":"Sridharan","sequence":"first","affiliation":[{"name":"Department of Pharmacology & Therapeutics, College of Medicine & Health Sciences, Arabian Gulf University, Manama P.O. Box 26671, Bahrain"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-5877-205X","authenticated-orcid":false,"given":"Gowri","family":"Sivaramakrishnan","sequence":"additional","affiliation":[{"name":"Bahrain Defence Force Royal Medical Services, Riffa P.O. Box 28743, Bahrain"}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"1968","published-online":{"date-parts":[[2026,3,22]]},"reference":[{"key":"ref_1","doi-asserted-by":"crossref","first-page":"1237","DOI":"10.2146\/ajhp160756","article-title":"Trends in utilization of warfarin and direct oral anticoagulants in older adult patients with atrial fibrillation","volume":"74","author":"Alalwan","year":"2017","journal-title":"Am. J. Health-Syst. Pharm."},{"key":"ref_2","first-page":"6","article-title":"Problems with Warfarin Anticoagulation Control: An InDepth Analysis and a Holistic Strategy","volume":"12","author":"Sunil","year":"2022","journal-title":"Int. J. Pharmacetical Sci. Lett."},{"key":"ref_3","doi-asserted-by":"crossref","first-page":"1417","DOI":"10.1007\/s00228-017-2296-1","article-title":"Trends in use of warfarin and direct oral anticoagulants in atrial fibrillation in Norway, 2010 to 2015","volume":"73","author":"Kjerpeseth","year":"2017","journal-title":"Eur. J. Clin. Pharmacol."},{"key":"ref_4","doi-asserted-by":"crossref","unstructured":"Neidecker, M., Patel, A.A., Nelson, W.W., and Reardon, G. (2012). Use of warfarin in long-term care: A systematic review. BMC Geriatr., 12.","DOI":"10.1186\/1471-2318-12-14"},{"key":"ref_5","unstructured":"Pratt, V.M., Scott, S.A., Pirmohamed, M., Esquivel, B., Kattman, B.L., and Malheiro, A.J. (2026, February 11). Warfarin Therapy and VKORC1 and CYP Genotype, Medical Genetics Summaries [Internet], Available online: https:\/\/www.ncbi.nlm.nih.gov\/books\/NBK84174\/."},{"key":"ref_6","doi-asserted-by":"crossref","first-page":"5248","DOI":"10.1182\/bloodadvances.2024014227","article-title":"Meta-analysis of genome-wide association studies of stable warfarin dose in patients of African ancestry","volume":"8","author":"Asiimwe","year":"2024","journal-title":"Blood Adv."},{"key":"ref_7","doi-asserted-by":"crossref","first-page":"397","DOI":"10.1002\/cpt.668","article-title":"Clinical Pharmacogenetics Implementation Consortium (CPIC) guideline for pharmacogenetics-guided warfarin dosing: 2017 update","volume":"102","author":"Johnson","year":"2017","journal-title":"Clin. Pharmacol. Ther."},{"key":"ref_8","doi-asserted-by":"crossref","first-page":"79","DOI":"10.1007\/s11096-022-01471-y","article-title":"Evaluation of supervised machine learning algorithms in predicting the poor anticoagulation control and stable weekly doses of warfarin","volume":"45","author":"Sridharan","year":"2023","journal-title":"Int. J. Clin. Pharm."},{"key":"ref_9","doi-asserted-by":"crossref","first-page":"466","DOI":"10.2174\/1389200224666230705124329","article-title":"Computational Structural Validation of CYP2C9 Mutations and Evaluation of Machine Learning Algorithms in Predicting the Therapeutic Outcomes of Warfarin","volume":"24","author":"Sridharan","year":"2023","journal-title":"Curr. Drug Metab."},{"key":"ref_10","doi-asserted-by":"crossref","first-page":"82","DOI":"10.1016\/j.inffus.2019.12.012","article-title":"Explainable Artificial Intelligence (XAI): Concepts, taxonomies, opportunities and challenges toward responsible AI","volume":"58","author":"Arrieta","year":"2020","journal-title":"Inf. Fusion"},{"key":"ref_11","doi-asserted-by":"crossref","first-page":"1405","DOI":"10.1007\/s43440-021-00256-w","article-title":"Influence of CYP2C9, VKORC1, and CYP4F2 polymorphisms on the pharmacodynamic parameters of warfarin: A cross-sectional study","volume":"73","author":"Sridharan","year":"2021","journal-title":"Pharmacol. Rep."},{"key":"ref_12","doi-asserted-by":"crossref","unstructured":"van Miert, J.H., Bos, S., Veeger, N.J., and Meijer, K. (2018). Clinical usefulness of the SAMe-TT2R2 score: A systematic review and simulation meta-analysis. PLoS ONE, 13.","DOI":"10.1371\/journal.pone.0194208"},{"key":"ref_13","doi-asserted-by":"crossref","unstructured":"Kamthornthanakarn, I., and Krittayaphong, R. (2019). Optimal INR level for warfarin therapy after mechanical mitral valve replacement. BMC Cardiovasc. Disord., 19.","DOI":"10.1186\/s12872-019-1078-3"},{"key":"ref_14","doi-asserted-by":"crossref","first-page":"972","DOI":"10.1097\/MBC.0000000000000312","article-title":"Is the time in therapeutic range using the ratio of tests equivalent to the Rosendaal method?","volume":"26","author":"Caldeira","year":"2015","journal-title":"Blood Coagul. Fibrinolysis"},{"key":"ref_15","doi-asserted-by":"crossref","first-page":"1357","DOI":"10.1016\/j.jacc.2018.06.065","article-title":"Outcomes among patients with atrial fibrillation and appropriate anticoagulation control","volume":"72","author":"Bonde","year":"2018","journal-title":"J. Am. Coll. Cardiol."},{"key":"ref_16","first-page":"11","article-title":"A comparative evaluation of random forest and XGBoost models for disease detection using medical indicators","volume":"19","author":"Tuama","year":"2025","journal-title":"Int. J. Prof. Stud."},{"key":"ref_17","doi-asserted-by":"crossref","first-page":"208","DOI":"10.1111\/j.1365-2125.2012.04335.x","article-title":"The pharmacokinetics and pharmacodynamics of single dose (R)-and (S)-warfarin administered separately and together: Relationship to VKORC1 genotype","volume":"75","author":"Maddison","year":"2013","journal-title":"Br. J. Clin. Pharmacol."},{"key":"ref_18","doi-asserted-by":"crossref","unstructured":"Zhu, X., Xiao, X., Wang, S., Chen, X., Lu, G., and Li, X. (2022). Rosendaal linear interpolation method appraising of time in therapeutic range in patients with 12-week follow-up interval after mechanical heart valve replacement. Front. Cardiovasc. Med., 9.","DOI":"10.3389\/fcvm.2022.925571"},{"key":"ref_19","doi-asserted-by":"crossref","first-page":"e37833","DOI":"10.2196\/37833","article-title":"Implementation of machine learning pipelines for clinical practice: Development and validation study","volume":"10","author":"Kanbar","year":"2022","journal-title":"JMIR Med. Inform."},{"key":"ref_20","doi-asserted-by":"crossref","first-page":"e53207","DOI":"10.2196\/53207","article-title":"How explainable artificial intelligence can increase or decrease clinicians\u2019 trust in AI applications in health care: Systematic review","volume":"3","author":"Rosenbacke","year":"2024","journal-title":"JMIR AI"}],"container-title":["Medical Sciences"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/www.mdpi.com\/2076-3271\/14\/1\/156\/pdf","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,3,23]],"date-time":"2026-03-23T09:07:43Z","timestamp":1774256863000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.mdpi.com\/2076-3271\/14\/1\/156"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,3,22]]},"references-count":20,"journal-issue":{"issue":"1","published-online":{"date-parts":[[2026,3]]}},"alternative-id":["medsci14010156"],"URL":"https:\/\/doi.org\/10.3390\/medsci14010156","relation":{},"ISSN":["2076-3271"],"issn-type":[{"value":"2076-3271","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,3,22]]}}}