A Novel Prediction Tool for Overall Survival of Patients Living with Spinal Metastatic Disease.
2020
Abstract Objective To identify the significant prognostic factors for overall survival in patients with spinal metastases, and to establish an online widget for predicting survival with a interactive visual approach. Methods Patients operated for spinal metastases between 2010 and 2018 were retrospectively enrolled, and were randomly divided into training and validation samples with a ratio of 7:3. Patients’ characteristics were analyzed with univariate and multivariate COX analyses to identify independent prognostic factors basing on the training sample. A shiny web tool was developed by transform the fitted multivariable COX model into a visual interface. Time-dependent area under curve (AUC) plot and calibration curve were generated to assess the discrimination ability and consistency of the novel model, both for the training and validation samples. Results A total of 265 consecutive patients were finally included, with 185 in training sample and 80 in validation sample. The primary tumor types, lesion site of metastasis, visceral metastasis, Frankel grade, operation category, number of surgical segments and the pre-operative percentage of lymphocyte were demonstrated to be significantly associated with overall survival. A novel shiny model ( https://yang1209xg.shinyapps.io/predictspinalmetastasis/ ), which could provide predicted survival curve and median survival time, was established, with favorable discrimination ability and consistency between predicted and actual survival both in internal and external data, according to time-dependent AUC plots and calibration curves. Conclusions A user-friendly shiny app with favorable discrimination ability and consistency was released online for predicting the survival of patients with spinal metastases. A continuous survival curve and the predicted median survival time are available to guide the treatment planning.
Keywords:
- Correction
- Source
- Cite
- Save
- Machine Reading By IdeaReader
41
References
0
Citations
NaN
KQI