A machine learning strategy to mitigate the inappropriateness of procalcitonin request in clinical practice

Luisa Agnello; Matteo Vidali; Anna Maria Ciaccio; Bruna Lo Sasso; Alessandro Iacona; Giuseppe Biundo; Concetta Scazzone; Caterina Maria Gambino; Marcello Ciaccio

doi:10.1016/j.heliyon.2024.e26556

A machine learning strategy to mitigate the inappropriateness of procalcitonin request in clinical practice

Heliyon. 2024 Feb 17;10(5):e26556. doi: 10.1016/j.heliyon.2024.e26556. eCollection 2024 Mar 15.

Authors

Affiliations

¹ Institute of Clinical Biochemistry, Clinical Molecular Medicine and Clinical Laboratory Medicine, Department of Biomedicine, Neurosciences and Advanced Diagnostics, University of Palermo, 90127, Palermo, Italy.
² Foundation IRCCS Ca' Granda Ospedale Maggiore Policlinico, 20122, Milan, Italy.
³ Internal Medicine and Medical Specialties "G. D'Alessandro", Department of Health Promotion, Maternal and Infant Care, University of Palermo, 90127, Palermo, Italy.
⁴ Department of Laboratory Medicine, AOUP "P. Giaccone", Palermo, Italy.

Abstract

Aim: The aim of this study was to develop machine learning (ML) models to mitigate the inappropriate request of Procalcitonin (PCT) in clinical wards.

Material and methods: We built six different ML models based on both demographical data, i.e., sex and age, and laboratory parameters, i.e., cell blood count (CBC) parameters, inclusive of monocyte distribution width (MDW), and C-reactive protein (CRP). The dataset included 1667 PCT measurements of different patients. Based on a PCT cut-off of 0.50 ng/mL, we found 1090 negative (65.4%) and 577 positive (34.6%) results. We performed a 70:15:15 train:validation:test splitting based on the outcome.

Results: Random Forest, Support Vector Machine and eXtreme Gradient Boosting showed optimal performances for predicting PCT positivity, with an area under the curve ranging from 0.88 to 0.89.

Conclusions: The ML models developed could represent a useful tool to predict PCT positivity, avoiding unusefulness PCT requests. ML models are based on laboratory tests commonly ordered together with PCT but have the great advantage to be easy to measure and low-cost.

Keywords: Artificial intelligence; CBC; CRP; Laboratory medicine; MDW; Procalcitonin; Sepsis.