Abstract
Diabetic ketoacidosis (DKA) is a serious complication that affects millions of individuals
globally and presents significant health complications. Hyperchloremia, an electrolyte imbalance
characterized by high levels of chloride in the blood, may result in gastrointestinal problems, kidney damage,
and even death, especially in DKA patients. Early detection and treatment of hyperchloremia are of utmost
importance in the management of DKA. This study explores the potential of the bootstrap aggregating
ensemble with random subspaces machine learning approach to predict the occurrence of hyperchloremia,
providing a basis for early intervention and improved patient outcomes. We tested our approach with the
retrospective MIMIC-III database containing 1177 DKA patients and compared it with previous studies
with an area under the curve (AUC) of 100%. Our approach showed significant performance outperforming
other methods. The combination of this approach may enhance the early detection and timely intervention
of hyperchloremia cases, ultimately leading to improved patient outcomes and a more effective management
of DKA-associated complications. Our work aims to contribute to the development of decision support tools
for healthcare professionals, assisting them in making informed decisions for DKA patients, with a focus on
preventing and managing hyperchloremia.