A research group led by Assistant Professor Keisuke Nakamura of Shinshu University used Multi-Sigma® to build a deep neural network (DNN) model and analyze walking independence after hip fracture surgery. Their study was published under the title “Deep neural network–based prediction of walking independence after hip fracture surgery and analysis of contributing factors.”
The study included 368 patients aged 65 or older who underwent surgery for femoral neck or trochanteric fractures. The researchers developed models to predict walking independence at one and four weeks after surgery. Predictors included age, pre-injury living arrangements and walking status, cognitive function, comorbidities, fracture type, waiting times for surgery and rehabilitation to begin, and whether walking started within two days after surgery.
The researchers compared the DNN developed using Multi-Sigma® with logistic regression and support vector machines. At one week after surgery, the DNN showed the highest sensitivity among the three methods. At four weeks, it showed the most balanced classification performance in terms of precision, sensitivity, and specificity.
The researchers also used Multi-Sigma®’s Contribution Analysis (sensitivity analysis) feature to quantify how each variable contributed to the predictions.
For further details, please refer to the paper below.
Publication Details
Title: Deep neural network–based prediction of walking independence after hip fracture surgery and analysis of contributing factors
Authors: Keisuke Nakamura et al.
Journal: Journal of Orthopaedic Science
Publication Year: 2026
