김청수 (아주대학교)

한빛사 논문

조회212

아주대학교

CV File 139 kb

Int. J. Antimicrob. Agents, Sep 14 2023, S0924-8579(23)00245-5 | https://doi.org/10.1016/j.ijantimicag.2023.106966 Translation of Machine Learning-Based Prediction Algorithms to Personalized Empiric Antibiotic Selections: A Population-Based Cohort Study.

Chungsoo Kim ¹, Young Hwa Choi ², Jung Yoon Choi ³, Hee Jung Choi ^4,5, Rae Woong Park^{1,6 †}, Sandy Jeong Rhie^{3,7 †}

¹Department of Biomedical Sciences, Ajou University Graduate School of Medicine, Suwon, Gyeonggi-do, Republic of Korea

²Department of Infectious Diseases, Ajou University School of Medicine, Suwon, Republic of Korea

³Graduate School of Pharmaceutical Sciences, Ewha Womans University, Seoul, Republic of Korea

⁴College of Medicine, Ewha Womans University, Seoul, Republic of Korea

⁵Department of Internal Medicine, Ewha Womans University Mokdong Hospital, Seoul, Republic of Korea

⁶Department of Biomedical Informatics, Ajou University School of Medicine, 164, World cup-ro, Yeongtong-gu, Suwon-si, Gyeonggi-do 16499, Republic of Korea

⁷College of Pharmacy, Ewha Womans University, 52, Ewhayeodae-gil, Seodaemun-gu, Seoul 03760, Republic of Korea

†These authors contributed equally to this manuscript as co-senior authors.

Corresponding author: Sandy Jeong Rhie

Abstract

Background: Prediction of antibiotic non-susceptibility based on patient characteristics and clinical status may support to select empiric antibiotics for suspected hospital-acquired urinary tract infections (HA-UTIs).

Methods: Prediction models were developed to predict non-susceptible results of eight antibiotics susceptibility tests ordered for suspected HA-UTI. Eligible patients were those with urine culture and susceptibility test results after 48 hours of admissions between 2010 and 2021. We utilized patient demographics, diagnosis, prescriptions, exposure to multidrug-resistant organisms, transfer history, and daily calculated antibiogram as predictors. We used Lasso logistic regression (LLR), extreme gradient boosting (XGB), random forest (RF), and stacked ensemble methods for development. Parsimonious models were also developed for clinical utility. Discrimination was assessed using the area under the receiver operating characteristic curve (AUROC).

Results: The mean age was 62.1± 16.2 years and 48.1% were male in 10,474 suspected HA-UTI cases. Non-susceptibility prediction for ampicillin/sulbactam, cefepime, ciprofloxacin, imipenem, piperacillin/tazobactam, and trimethoprim/sulfamethoxazole performed best using the stacked ensemble (AUROC=76.9, 76.1, 77.0, 80.6, 76.1, and 76.5, respectively). The model for ampicillin performed best with LLR (AUROC=73.4). Only for gentamicin did the XGB perform best (AUROC=66.9). In the parsimonious models, the LLR yielded the highest AUROC for ampicillin, ampicillin/sulbactam, cefepime, gentamicin, and trimethoprim/sulfamethoxazole (AUROC=70.6, 71.8, 73.0, 65.9, and 73.0, respectively). The model for ciprofloxacin performed best with XGB (AUROC=70.3), while the model for imipenem performed best in the stacked ensemble (AUROC=71.3). A personalized application using the parsimonious models was released publicly.

Conclusions: We developed prediction models for antibiotic non-susceptibility to support the empiric antibiotic selection for HA-UTI.

논문정보

형식Research article
게재일2023년 09월 (BRIC 등록일 2023-09-20)
연구진국내 연구진
분야 의약학 > 의약학 기타

댓글 작성 로그인

CV 열람하기

연락처 보기