Background
Osteoporosis leads to bone fractures, burdened by high mortality and disability.
Our aim was to apply an artificial intelligence-based fragility fracture predictive model to different dual energy X-ray absorptiometry (DXA) parameters.

Methods
Retrospective approved multicentric study: 174 females without vertebral fractures at baseline were enrolled. All subjects performed spine X-ray, lumbar andfemoral DXA scan to assess bone mineral density (BMD) and bone strain index (BSI). Exams were performed at baseline and after a mean follow-up period of 3 years. The retrospective analysis was performed in the first two months of 2020. The occurrence of vertebral fracture was evaluated on x-rays at follow-up. Artificial Neural Network analysis (ANNs) selected relevant input data automatically (TWIST-system) constructing a semantic connectivity map (Auto Contractive Map) of the complex biological connections among the studied variables and the two conditions (fracture vs not fracture). A final list of 13 variables was included, considering both DXA-derived indices but also age, menopause status, weight and BMI.
Results
At follow-up, 69 women developed a vertebral fracture. TWIST system selected 9/13 available variables. ANNs reached a predictive accuracy of 79.56% with training testing procedure, with a sensitivity of 80.93% and a specificity of 78.18%. The semantic connectivity map highlighted BSI's role in predicting fracture risk: a high BSI value identified those women who did not experience further fragility fracture.
Conclusion
Artificial Intelligence is a useful method to analyse the complex osteoporotic system. Femoral BSI appears.

Note: 

1. Fondazione IRCCS Ca’ Granda Ospedale Maggiore Policlinico, Milano.
2. BSE TECHNOLOGIC Srl, Torino.
3. IRCCS Istituto Ortopedico Galeazzi, Milano, Italy.
4. Dipartimento di Scienze Biomediche per la Salute, Università degli Studi di Milano.
5. Former: Fondazione IRCCS Ca’ Granda Ospedale Maggiore Policlinico, Gastroenterology and
Digestive Endoscopy Unit, Milano.
6. Scuola di Specializzazione in Medicina Fisica e Riabilitativa, Università degli Studi di Milano.
7. IRCCS Policlinico San Donato, San Donato Milanese (Milano).
8. Institute of Endocrinology, Università Vita-Salute San Raffaele, Milano.
9. Villa Santa Maria Foundation, Tavernerio (Como).