Our smartphones can recognize the pictures of our family, loved ones and friends. Face recognition software leverages artificial intelligence (AI), image recognition and other advanced technology to map, analyze and confirm the identity of a face.

We humans do a poor job when classifying the injury related to a patient sustaining a proximal humeral fracture. In consequence, there is great heterogeneity in the treatment of proximal humerus fractures. Moreover, offering relevant information to patients regarding the risk of complications or fracture sequelae is challenging, given that the current series are based on obsolete classifications, and the published series bring together just over hundreds of patients analyzed. With these limitations, patients have few opportunities to participate in decision-making about their injury.

The present project aim is to integrate new technologies for the prediction of relevant clinical results for the patients presenting a proximal humeral fracture. In brief, AI can help identify similar fracture patterns without human inference, while humans can feed the algorithm with variables of interest such as the functional outcomes and complications related to this particular type of fracture.

Inclusion Criteria:

Patients sustaining a proximal humerus fracture treated nonoperatively under the criteria of the treating surgeon and patients' preference.

Subjects evaluated within the first 3 weeks after the injury. Patients between 18 and 90 years of age. Patients who have been studied with simple shoulder radiographs in anteroposterior and scapular outlet projections.

Participants who accept 1-year time follow-up.

Exclusion Criteria:

Patients with dementia or difficulty completing the evaluation after one year of follow-up.

Patients who have previously received surgical treatment on the affected limb. Patients who have suffered a previous fracture in the affected limb. Surgically treated patients.

Patients sustaining proximal humeral fractures, nonoperatively treated