Skip to content

Latest commit

 

History

History
54 lines (43 loc) · 2.23 KB

README.md

File metadata and controls

54 lines (43 loc) · 2.23 KB
Raw

Predicting bloodstream infection outcome using machine learning

Scientific Reports | medRxiv | PDF

Yazeed Zoabi1,†, Orli Kehat2,†, Dan Lahav1, Ahuva Weiss-Meilik2,‡, Amos Adler2,‡, Noam Shomron1,‡
1 Tel Aviv University
2 Tel-Aviv Sourasky Medical Center
†These authors contributed equally
‡These authors jointly supervised this work

Among the implications of this work is implementation of the models as a basis for selective rapid microbiological identifcation, toward earlier administration of appropriate antibiotic therapy. Additionally, our models may help reduce the development of BSI and its associated adverse health outcomes and complications.

Model Predictors and their respective units

  • RDW - (%)
  • Albumin - (g/L)
  • Age - Age in years
  • Creatinine - (mg/dL)
  • RBC - (10e6/ϻL)
  • Surgery - True = 1, False = 0
  • NRBC/100_WBC - (%)
  • Mean_platelet_volume - (fL)
  • AST - (U/L)
  • HCT - (%)
  • MCHC - (g/dL)
  • Indirect_Bilirubin - (mg/dL)
  • Cerebrovascular_disease - True = 1, False = 0
  • Alkaline - (U/L)
  • Platelet_Automated_Count - (10e3/ϻL)
  • MCV - (fL)
  • Catheterization - True = 1, False = 0
  • Respiratory_diseases - True = 1, False = 0
  • Direct_bilirubin - (mg/dL)
  • ALT - (U/L)
  • Sex - Male=1, Female=0
  • Lymphocytes - (10e3/ϻL)
  • Neutrophils - (10e3/ϻL)
  • Infectious_background - True = 1, False = 0
  • Insulin - True = 1, False = 0

Model Outcome

The probability of having the composite outcome as defined by the manuscript.

Use

  1. Import lgbm_compact_model.txt using LightGBM 2.3.1 on Python 3.6.

  2. Predict using your data.

Files in this repository

  • lgbm_compact_model.txt - The compact model that uses 25 features
  • hyperparameters.txt - The hyperparameters used by LightGBM to train the inclusive and the compact model