Download This Paper
Preprints with The Lancet is a collaboration between The Lancet Group of journals and SSRN to facilitate the open sharing of preprints for early engagement, community comment, and collaboration. Preprints available here are not Lancet publications or necessarily under review with a Lancet journal. These preprints are early-stage research papers that have not been peer-reviewed. The usual SSRN checks and a Lancet-specific check for appropriateness and transparency have been applied. The findings should not be used for clinical or public health decision-making or presented without highlighting these facts. For more information, please see the FAQs.
Prediction of Recurrence in Pyogenic Vertebral Osteomyelitis: Application of Artificial Neural Network Using Time-Series Data of C-Reactive Protein
22 Pages Posted: 1 Jul 2020
More...Abstract
Background: Prediction of recurrence in pyogenic vertebral osteomyelitis (PVO) is crucial to avoid additional 6-week antibiotics course with hospitalisation, aggressive spinal surgery, and to reduce the mortality rate. However, prediction of PVO recurrence by pre-treatment, initial risk factors is limited in PVO patients who require long-term intravenous antibiotics with hospitalisation and experience various events affecting treatment outcome during hospitalisation. We hypothesised that time-series analysis of sequential C-reactive protein (CRP) during the 6-week antibiotics course could reflect such long treatment process.
Methods: A retrospective study was performed to develop a PVO recurrence-prediction model, including initial risk factors and time-series data of CRP. Of 739 PVO patients, 516 and 223 were divided into training and test cohorts, respectively. Conventional stepwise logistic regression and artificial neural network (ANN) models were created from the training cohort, and their prediction of recurrence in test cohort were compared.
Findings: Prediction models using initial risk factors showed poor sensitivity (2·3%) and accuracy (80·7% and 81·2%), whereas baseline ANN models using time-series CRP data showed remarkably increased sensitivity (55·8 and 60·5%) and accuracy (89·7% and 88·8%). Ensemble ANN model using both initial risk factors and time-series CRP data showed additional benefit in prediction power.
Interpretation: ANN models using time-series CRP data and their ensemble model showed favourable outcome in prediction of PVO recurrence, and was superior to the conventional model using initial risk factors.
Funding Information: No funds were received in support of this work.
Declaration of Interests: The authors declare no conflicts of interest.
Ethical Approval Statement: The study protocol was approved by the Institutional Review Board(No. 2019-12-004-001) of our institution.
Keywords: pyogenic; vertebral osteomyelitis; recurrence; risk factor; infection burden; CRP; time-series data; artificial neural network; deep learning
Suggested Citation: Suggested Citation