Dataset for "Machine learning outperforms clinical experts in classification of hip fractures"

Hip fractures are a major cause of morbidity and mortality in the elderly, and incur high health and social care costs. Given projected population ageing, the number of incident hip fractures is predicted to increase globally. As fracture classification strongly determines the chosen surgical treatment, differences in fracture classification influence patient outcomes and treatment costs. We aimed to create a machine learning method for identifying and classifying hip fractures, and to compare its performance to experienced human observers. We used 3659 hip radiographs, classified by at least two expert clinicians. The machine learning method was able to classify hip fractures with 19% greater accuracy than humans, achieving overall accuracy of 92%.

This data set contains the source data for figures 2 and 4, which are the main Results figures. Data are given in both csv and MAT file formats. The MATLAB scripts for generating the figures are also provided.

Subjects:
Mathematical sciences
Medical and health interface
Tools, technologies and methods

Cite this dataset as:
Gill, R., Ehrhardt, B., Murphy, E., 2022. Dataset for "Machine learning outperforms clinical experts in classification of hip fractures". Bath: University of Bath Research Data Archive. Available from: https://doi.org/10.15125/BATH-01011.

Export

[QR code for this page]

Data

data_files.zip
application/zip (500kB)
Creative Commons: Attribution 4.0

Data for figures 2 and 4, as well as MATLAB scripts to generate figures

Creators

Richie Gill
University of Bath

Beate Ehrhardt
University of Bath

Ellen Murphy
University of Bath

Contributors

University of Bath
Rights Holder

Coverage

Collection date(s):

From 1 December 2017 to 29 March 2019

Geographical coverage:

UK, Southwest, Bath & Bristol

Documentation

Data collection method:

Machine learning recognition of the hip joint in antero-posterior (AP) pelvis x-rays. Identification and classification of hip fractures using machine learning. Anonymised x-rays were obtained, with ethical approval, from North Bristol NHS Trust and the Royal United Hospital NHS Foundation Trust.

Data processing and preparation activities:

NA

Technical details and requirements:

The Caffe 1.0 machine learning system was used to generate the machine learning algorithms. MATLAB (version 2017) was used to process training and test data, and to analyse output.

Additional information:

The data is organised clearly into the source data for each figure

Documentation Files

Readme.txt
text/plain (155B)
Creative Commons: Attribution 4.0

Readme file

Funders

Arthroplasty for Arthritis

ARCHi: Automated Recognition and Classification of Hip Fracture

Publication details

Publication date: 8 February 2022
by: University of Bath

Version: 1

DOI: https://doi.org/10.15125/BATH-01011

URL for this record: https://researchdata.bath.ac.uk/id/eprint/1011

Related papers and books

Murphy, E. A., Ehrhardt, B., Gregson, C. L., von Arx, O. A., Hartley, A., Whitehouse, M. R., Thomas, M. S., Stenhouse, G., Chesser, T. J. S., Budd, C. J., and Gill, H. S., 2022. Machine learning outperforms clinical experts in classification of hip fractures. Scientific Reports, 12(1). Available from: https://doi.org/10.1038/s41598-022-06018-9.

Contact information

Please contact the Research Data Service in the first instance for all matters concerning this item.

Contact person: Richie Gill

Departments:

Faculty of Science
Mathematical Sciences

Faculty of Engineering & Design
Mechanical Engineering

Research Centres & Institutes
Centre for Therapeutic Innovation