Predictive model for a second hip fracture occurrence using natural language processing and machine learning on...

Area of Science:

• Gerontology and Public Health
• Artificial Intelligence in Medicine
• Orthopedic Surgery

Background:

• Hip fractures (HFx) are a major cause of morbidity and mortality in older adults, severely limiting mobility and quality of life.
• Understanding the clinical characteristics of patients experiencing initial and subsequent hip fractures is crucial for effective patient management and prevention strategies.

Purpose of the Study:

• To analyze real-world data on the clinical profiles of patients with initial and second hip fractures (2HFx).
• To develop and validate an artificial intelligence-based predictive model for identifying patients at high risk of sustaining a second hip fracture.

Main Methods:

• Analysis of electronic health records from a Spanish hospital (2011-2019) using natural language processing and machine learning.
• Inclusion of 1,960 patients with HFx, with 124 in the 2HFx subgroup.
• Development of a competitive risk model using 16 demographic and clinical characteristics identified as significant predictors.

Main Results:

• The study identified 16 key predictors for second hip fractures, including visual deficit, malnutrition, walking assistance, hypothyroidism, female sex, and history of osteoporosis.
• The predictive model demonstrated good performance with an AUC of 0.69 (dependent) and 0.75 (apparent).
• Comorbidities like hypertension, cognitive impairment, diabetes, heart failure, and chronic kidney disease were more prevalent in patients with hip fractures.

Conclusions:

• The developed AI model can effectively identify older patients at higher risk for a second hip fracture.
• These findings provide valuable insights into HFx patient characteristics in Spain and support the implementation of targeted preventive measures to reduce refracture rates.