Predicting treatment recommendations in postmenopausal osteoporosis

G Bonaccorsi; M Giganti; M Nitsenko; G Pagliarini; G Piva; G Sciavicco

doi:10.1016/j.jbi.2021.103780

Predicting treatment recommendations in postmenopausal osteoporosis

J Biomed Inform. 2021 Jun:118:103780. doi: 10.1016/j.jbi.2021.103780. Epub 2021 Apr 20.

Authors

G Bonaccorsi¹, M Giganti², M Nitsenko³, G Pagliarini³, G Piva³, G Sciavicco⁴

Affiliations

¹ Dept. of Translat. Med. and for Romagna, Menopause and Osteoporosis Center, University of Ferrara, Italy.
² Dept. of Translat. Med. and for Romagna, and Exper. Med., Section of Radiology, Univ. of Ferrara, Italy.
³ Dept. of Mathematics and Computer Science, Univ. of Ferrara, Italy.
⁴ Dept. of Mathematics and Computer Science, Univ. of Ferrara, Italy. Electronic address: guido.sciavicco@unife.it.

PMID: 33857641
DOI: 10.1016/j.jbi.2021.103780

Abstract

We designed, implemented, and tested a clinical decision support system at the Research Center for the Study of Menopause and Osteoporosis within the University of Ferrara (Italy). As an independent module of our system, we implemented an original machine learning system for rule extraction, enriched with a hierarchical extraction methodology and a novel rule evaluation technique. Such a module is used in everyday operation protocol, and it allows physicians to receive suggestions for prevention and treatment of osteoporosis. In this paper, we design and execute an experiment based on two years of data, in order to evaluate and report the reliability of our suggestion system. Our results are encouraging, and in some cases reach expected accuracies of around 90%.

Keywords: Clinical decision support system; Keywords: Osteoporosis treatment; Machine learning; Rule extraction.

MeSH terms

Decision Support Systems, Clinical*
Female
Humans
Italy
Machine Learning
Osteoporosis, Postmenopausal* / drug therapy
Reproducibility of Results