Comparing manual vs. automated machine learning and deep learning models for predicting one-year mortality in elderly hip fracture patients - Takeaways - MDSpire

Comparing manual vs. automated machine learning and deep learning models for predicting one-year mortality in elderly hip fracture patients

  • By

  • Adi Shuchami

  • Maxim Glebov

  • Maksim Katsin

  • Yotam Portnoy

  • Haim Berkenstadt

  • Dina Orkin

  • Teddy Lazebnik

  • June 1, 2026

  • 0 min

Share

  • 1

    The study evaluated machine learning and deep learning algorithms for predicting one-year mortality in elderly hip fracture patients.

  • 2

    Manually optimised Extreme Gradient Boosting (XGB) algorithm outperformed other models with an AUC of 0.846 and accuracy of 0.791.

  • 3

    Key predictors of mortality included serum albumin and urea levels, patient age, intraoperative hypothermia, and chronic disease count.

  • 4

    An automated ML model using a large language model showed comparable performance to XGB, with a higher recall but lower precision.

  • 5

    The study highlights the potential of automated ML frameworks to democratize advanced predictive analytics in clinical settings.

Original Source(s)

Comparing manual vs. automated machine learning and deep learning models for predicting one-year mortality in elderly hip fracture patients

  • by Adi Shuchami, Maxim Glebov, Maksim Katsin, Yotam Portnoy, Haim Berkenstadt, Dina Orkin, Teddy Lazebnik

  • June 1, 2026

Related Content

Defining orthoplastic limb salvage centers: a systematic review

BAR Could Predict Mortality After Hip Fracture

Preoperative lab ratio shows stronger association with short-term mortality risk vs other biomarkers.

Case Report: Revision surgery for a missed posterior dislocation of the humeral head in a shoulder fracture-dislocation and literature review