Machine learning-based prediction of medication refill adherence among first-time insulin users with type 2 diabetes

被引:2
作者
Chen, Ya-Lin [1 ,5 ]
Nguyen, Phung-Anh [2 ,3 ,4 ]
Chien, Chia-Hui [5 ,6 ,7 ]
Hsu, Min-Huei [8 ,9 ]
Liou, Der-Ming [5 ]
Yang, Hsuan-Chia [3 ,5 ,6 ,10 ,11 ]
机构
[1] Univ Washington, Dept Biomed Informat & Med Educ, Seattle, WA USA
[2] Taipei Med Univ, Clin Data Ctr, Off Data Sci, Taipei, Taiwan
[3] Taipei Med Univ, Taipei Med Univ Hosp, Clin Big Data Res Ctr, Taipei 110301, Taiwan
[4] Taipei Med Univ, Coll Management, Res Ctr Hlth Care Ind Data Sci, Taipei, Taiwan
[5] Taipei Med Univ, Coll Med Sci & Technol, Grad Inst Biomed Informat, Taipei, Taiwan
[6] Taipei Med Univ, Coll Med Sci & Technol, Int Ctr Hlth Informat Technol ICHIT, Taipei, Taiwan
[7] Taipei Med Univ, Off Med Affairs, Taipei, Taiwan
[8] Taipei Med Univ, Off Data Sci, Taipei, Taiwan
[9] Taipei Med Univ, Grad Inst Data Sci, Coll Management, Taipei, Taiwan
[10] Taipei Med Univ, Wan Fang Hosp, Res Ctr Big Data & Meta Anal, Taipei, Taiwan
[11] 9F,Educ & Res Bldg,Shuang Ho Campus,301,Yuantong R, New Taipei City 235, Taiwan
关键词
Observational study; Machine learning; Medication adherence; Type; 2; diabetes; Insulin; PERSISTENCE; MELLITUS; THERAPY; BIAS;
D O I
10.1016/j.diabres.2023.111033
中图分类号
R5 [内科学];
学科分类号
1002 ; 100201 ;
摘要
Aims: The prevalence of Type 2 Diabetes Mellitus (T2DM) is projected to be 7 % in 2030. Despite its need for long-term diabetes care, the adherence rate of injectable medications such as insulin is around 60 %, lower than the acceptable threshold of 80 %. This study aims to create classification models to predict insulin adherence among adult T2DM naive insulin users.Methods: Clinical data were extracted from Taipei Medical University Clinical Research Database (TMUCRD) from January 1st, 2004 to December 30th, 2020. A patient was regarded as adherent if his/her medication possession ratio (MPR) was at least 80 %. Seven domains of predictors were created, including demographics, baseline medications, baseline comorbidities, baseline laboratory data, healthcare resource utilization, index insulins, and the concomitant non-insulin T2DM medications. We built two Xgboost models for internal and external testing respectively.Results: Using a cohort of 4134 patients from Taiwan, our model achieved the Area Under the curve of the Receiver Operating Characteristic (AUROC) of the internal test was 0.782 and the AUROC of the external test was 0.771. the SHAP (SHapley Additive exPlanations) value showed that the number of prescribed medications, the number of outpatient visits, and laboratory data were predictive of future insulin adherence.Conclusions: This is the first study to predict adherence among adult naive insulin users. The developed model is a potential clinical decision support tool to identify possible non-adherent patients for healthcare providers to design individualized education plans.
引用
收藏
页数:7
相关论文
共 38 条
[2]   8. Obesity and Weight Management for the Prevention and Treatment of Type 2 Diabetes: Standards of Medical Care in Diabetes-2022 [J].
American Diabetes Association Professional Practice Committee* .
DIABETES CARE, 2022, 45 :S113-S124
[3]   4. Comprehensive Medical Evaluation and Assessment of Comorbidities: Standards of Medical Care in Diabetes-2022 [J].
American Diabetes Association Professional Practice Committee .
DIABETES CARE, 2022, 45 :S46-S59
[4]   6. Glycemic Targets: Standards of Medical Care in Diabetes-2022 [J].
American Diabetes Association Professional Practice Committee .
DIABETES CARE, 2022, 45 :S83-S96
[5]   How payment scheme affects patients' adherence to medications? A systematic review [J].
Aziz, Hamiza ;
Hatah, Ernieda ;
Bakry, Mohd Makmor ;
Islahudin, Farida .
PATIENT PREFERENCE AND ADHERENCE, 2016, 10 :837-+
[6]   A Systematic Review of Medication Adherence Thresholds Dependent of Clinical Outcomes [J].
Baumgartner, Pascal C. ;
Haynes, R. Brian ;
Hersberger, Kurt E. ;
Arnet, Isabelle .
FRONTIERS IN PHARMACOLOGY, 2018, 9
[7]   XGBoost: A Scalable Tree Boosting System [J].
Chen, Tianqi ;
Guestrin, Carlos .
KDD'16: PROCEEDINGS OF THE 22ND ACM SIGKDD INTERNATIONAL CONFERENCE ON KNOWLEDGE DISCOVERY AND DATA MINING, 2016, :785-794
[8]   The prescribing trend of oral antidiabetic agents for type 2 diabetes in Taiwan: An 8-year population-based study [J].
Chu, Wei-Min ;
Ho, Hsin-En ;
Huang, Kuang-Hua ;
Tsan, Yu-Tse ;
Liou, Yi-Sheng ;
Wang, Yueh-Hsin ;
Lee, Meng-Chih ;
Li, Yu-Ching .
MEDICINE, 2017, 96 (43)
[9]   A systematic review of adherence with medications for diabetes [J].
Cramer, JA .
DIABETES CARE, 2004, 27 (05) :1218-1224
[10]   Medication compliance and persistence: Terminology and definitions [J].
Cramer, Joyce A. ;
Roy, Anuja ;
Burrell, Anita ;
Fairchild, Carol J. ;
Fuldeore, Mahesh J. ;
Ollendorf, Daniel A. ;
Wong, Peter K. .
VALUE IN HEALTH, 2008, 11 (01) :44-47