The aim is to make an AI model recognizing possibly avoidable blood draws for serum potassium among pediatric patients following heart medical procedure. We gathered factors identified with potassium homeostasis, including serum science, hourly potassium admission, diuretics, and pee yield. Utilizing set up AI methods, including irregular woods classifiers, and hyperparameter tuning, we made models anticipating whether a patient’s potassium would be typical or unusual dependent on the latest potassium level, prescriptions controlled, pee yield, and markers of renal capacity. We built up numerous models dependent on various age-classifications and transient nearness of the latest potassium estimation. We surveyed the prescient execution of the models utilizing a free test set. Of the 7,269 affirmations (6,196 patients) included, serum potassium was estimated on normal of 1 (interquartile range, 0–1) time every day. Roughly 96% of patients got in any event one portion of IV diuretic and 83% got a type of potassium supplementation. AI techniques can be utilized to anticipate avoidable blood tests precisely for serum potassium in fundamentally sick pediatric patients. A middle of 27.2% of tests might have been saved, with diminished expenses and danger of disease or paleness.
Reference link- https://pdfs.journals.lww.com/pccmjournal/9000/00000/avoidable_serum_potassium_testing_in_the_cardiac.97884.pdf
