Page 119 - 《中国药房》2026年1期
P. 119

心、回顾性开发转向多中心、前瞻性验证,最终通过干预                          [12]  HOSTE E A J,KELLUM J A,SELBY N M,et al. Global
          性研究证实 ADRs 预测模型的可行性,这也是 ADRs 预                          epidemiology and outcomes of acute kidney injury[J]. Nat
          测模型获得临床认可与监管批准的必经之路。(3)在整                               Rev Nephrol,2018,14(10):607-625.
          合层面,机器学习不再局限于风险评分,而是将ADRs预                         [13]  KIM J Y,KIM K Y,YEE J,et al. Risk scoring system for
          测模型与医院信息系统、电子病历等相结合,推动药物                                vancomycin-associated acute kidney injury[J]. Front Phar‐
                                                                  macol,2022,13:815188.
          安全监测向智能化、个体化方向转型。
                                                             [14]  YANG Y L,YAN H Y,WANG X Y,et al. Clinical risk as‐
              综上所述,机器学习凭借其强大的数据处理与特征
                                                                  sessment of serum creatinine abnormalities during vanco‐
          挖掘能力,在肾脏、肝脏等多器官 ADRs 预测中成效显
                                                                  mycin  therapy:a  retrospective study  using  machine  lear-
          著,为药物安全监测提供了新路径。
                                                                  ning models[J]. Int J Clin Pharm,2025,47(6):1830-1840.
          参考文献                                               [15]  ZHANG P,CHEN Q,LAO J H,et al. Machine learning
          [ 1 ]  EDWARDS I R,ARONSON J K. Adverse drug reactions:  modeling for the risk of acute kidney injury in inpatients
              definitions,diagnosis,and management[J]. Lancet,2000,  receiving  amikacin  and  etimicin[J].  Front  Pharmacol,
              356(9237):1255-1259.                                2025,16:1538074.
          [ 2 ]  国家药品不良反应监测中心. 国家药品不良反应监测年                   [16]  HUANG S H,CHU C Y,HSU Y C,et al. How platinum-
              度报告(2024 年)[J]. 中国病毒病杂志,2025,15(4):                 induced  nephrotoxicity  occurs   machine  learning  predic‐
              317-326.                                            tion  in  non-small  cell  lung  cancer  patients[J].  Comput
          [ 3 ]  PIRMOHAMED  M,JAMES  S,MEAKIN  S,et  al.  Ad‐    Methods Programs Biomed,2022,221:106839.
              verse drug reactions as cause of admission to hospital:pro‐  [17]  YU X,WU R,JI Y W,et al. Identifying patients at risk of
              spective  analysis  of  18  820  patients[J].  BMJ,2004,329  acute  kidney  injury  among  patients  receiving  immune
              (7456):15-19.                                       checkpoint  inhibitors:a  machine  learning  approach[J].
          [ 4 ]  INÁCIO P,CAVACO A,AIRAKSINEN M. The value of     Diagnostics (Basel),2022,12(12):3157.
              patient reporting to the pharmacovigilance system:a sys‐  [18]  MIAO T,LEE L H N,SUN T,et al. Safety of nonselec‐
              tematic  review[J].  Br  J  Clin  Pharmacol,2017,83(2):  tive  nonsteroidal  anti-inflammatory  drugs  in  cardiac  sur‐
              227-246.                                            gery:a historical cohort study[J]. Can J Anaesth,2025,72
          [ 5 ]  LOPEZ-GONZALEZ  E,HERDEIRO  M T,FIGUEIRAS       (7):1056-1065.
              A. Determinants of under-reporting of adverse drug reac‐  [19]  肇丽梅,缪丽燕. 药物性肝损伤研究进展及展望[J]. 药学
              tions:a  systematic  review[J].  Drug  Saf,2009,32(1):  进展,2023,47(2):81-83.
              19-31.                                         [20]  HE X,YAO P L,LI M T,et al. A risk scoring model for
          [ 6 ]  詹世鹏,马攀,刘芳. 机器学习在治疗药物监测与个体化                       high-dose  methotrexate-induced  liver  injury  in  children
              用药中的应用[J]. 中国药房,2023,34(1):117-121,128.             with  acute  lymphoblastic  leukemia  based  on  gene  poly‐
          [ 7 ]  LI Q Y,TANG B H,WU Y E,et al. Machine learning:a   morphism study[J]. Front Pharmacol,2021,12:726229.
              new approach for dose individualization[J]. Clin Pharma‐  [21]  HAN J M,YEE J,CHO S,et al. A risk scoring system uti‐
              col Ther,2024,115(4):727-744.                       lizing machine learning methods for hepatotoxicity predic‐
          [ 8 ]  HANDELMAN G S,KOK H K,CHANDRA R V,et al.         tion one year after the initiation of tyrosine kinase inhibi‐
              eDoctor:machine learning and the future of medicine[J]. J   tors[J]. Front Oncol,2022,12:790343.
              Intern Med,2018,284(6):603-619.                [22]  ZENG Y,LU H,LI S,et al. Risk prediction of liver injury
          [ 9 ]  SETO H,OYAMA A,KITORA S,et al. Gradient boosting   in pediatric tuberculosis treatment:development of an au‐
              decision  tree  becomes  more  reliable  than  logistic  regres‐  tomated  machine  learning  model[J].  Drug  Des  Devel
              sion in predicting probability for diabetes with big data[J].   Ther,2025,19:239-250.
              Sci Rep,2022,12(1):15889.                      [23]  XIAO Y,CHEN Y F,HUANG R J,et al. Interpretable ma‐
          [10]  IOSIFIDIS A,TEFAS A,PITAS  I.  Graph  embedded  ex‐  chine  learning  in  predicting  drug-induced  liver  injury
              treme learning machine[J]. IEEE Trans Cybern,2016,46  among tuberculosis patients:model development and vali‐
              (1):311-324.                                        dation  study[J].  BMC  Med  Res  Methodol, 2024,
          [11]  BRESSO E,MONNIN P,BOUSQUET C,et al. Investi‐      24(1):92.
              gating ADR mechanisms with explainable AI:a feasibility   [24]  CARDINALE  D, CICERI  F, LATINI  R, et  al.
              study with knowledge graph mining[J]. BMC Med Inform   Anthracycline-induced  cardiotoxicity:a  multicenter  ran‐
              Decis Mak,2021,21(1):171.                           domised trial comparing two strategies for guiding preven‐


          中国药房  2026年第37卷第1期                                                 China Pharmacy  2026 Vol. 37  No. 1    · 109 ·
   114   115   116   117   118   119   120   121   122   123   124