BackgroundAccurate prediction of prognosis is critical for therapeutic decisions in chondrosarcoma patients. Several prognostic models have been created utilizing multivariate Cox regression or binary classification-based machine learning approaches to predict the 3- and 5-year survival of patients with chondrosarcoma, but few studies have investigated the results of combining deep learning with time-to-event prediction. Compared with simplifying the prediction as a binary classification problem, modeling the probability of an event as a function of time by combining it with deep learning can provide better accuracy and flexibility.Materials and methodsPatients with the diagnosis of chondrosarcoma between 2000 and 2018 were extracted from the Surveillance, Epidemiology, and End Results (SEER) registry. Three algorithms—two based on neural networks (DeepSurv, neural multi-task logistic regression [NMTLR]) and one on ensemble learning (random survival forest [RSF])—were selected for training. Meanwhile, a multivariate Cox proportional hazards (CoxPH) model was also constructed for comparison. The dataset was randomly divided into training and testing datasets at a ratio of 7:3. Hyperparameter tuning was conducted through a 1000-repeated random search with 5-fold cross-validation on the training dataset. The model performance was assessed using the concordance index (C-index), Brier score, and Integrated Brier Score (IBS). The accuracy of predicting 1-, 3-, 5- and 10-year survival was evaluated using receiver operating characteristic curves (ROC), calibration curves, and the area under the ROC curves (AUC).ResultsA total of 3145 patients were finally enrolled in our study. The mean age at diagnosis was 52 ± 18 years, 1662 of the 3145 patients were male (53%), and mean survival time was 83 ± 67 months. Two deep learning models outperformed the RSF and classical CoxPH models, with the C-index on test datasets achieving values of 0.832 (DeepSurv) and 0.821 (NMTLR). The DeepSurv model produced better accuracy and calibrated survival estimates in predicting 1-, 3- 5- and 10-year survival (AUC:0.895-0.937). We deployed the DeepSurv model as a web application for use in clinical practice; it can be accessed through https://share.streamlit.io/whuh-ml/chondrosarcoma/Predict/app.py.ConclusionsTime-to-event prediction models based on deep learning algorithms are successful in predicting chondrosarcoma prognosis, with DeepSurv producing the best discriminative performance and calibration.

No abstract

No abstract

This paper presents a comparison of studies on the local distributed electrical conductivity in stress corrosion crack (SCC) from signals of eddy current testing (ECT) and direct current potential drop (DCPD) aiming to improve SCC sizing accuracy when using electromagnetic non-destructive testing (NDT) methods. Experimental setups of ECT and DCPD were established, respectively, to collect measurement signals due to artificial SCCs in a plate of austenitic stainless steel. The local conductivity in the SCC region was reconstructed from the feature parameters extracted from the measured ECT and DCPD signals through inverse analyses. The inversion strategies for ECT and DCPD, each including an efficient forward simulation and an optimization scheme, were introduced from the viewpoint of conductivity reconstruction. Inversion results obtained from the measured ECT and DCPD signals showed the consistent trend which proved the validity of the predicted electrical conductivity indirectly. It is clarified that the electrical conductivity in a SCC is relatively high at the crack tip area and may become as high as 17% of that of the base material. These results provide a good reference to enhance the sizing accuracy of SCC with an electromagnetic NDT method such as ECT by updating the conductive crack model based on the results of this work. This article is part of the theme issue ‘Advanced electromagnetic non-destructive evaluation and smart monitoring’.

This paper proposes a new crack model to parameterize profile of a natural crack in order to improve the inversion precision of crack profile with eddy current testing (ECT) signals. In this novel crack model, the edge profile of a planar crack is parametrized with 6 feature parameters of three curves, i.e., two quarter-elliptic curves and a straight line. Compared with the conventional rectangular and semi-elliptical crack model, the new model gives better profile fitting of actual cracks such as a fatigue crack or a stress corrosion crack. In addition, the reconstruction procedure with the new crack model is more stable than using the arbitrary-shaped crack model outlined by a piecewise line as the unknown parameters to be reconstructed is significantly reduced. To apply the new crack model for crack inversion with the conventional conjugate gradient method, the formulae for calculating the gradient vector regarding to the feature parameters of the new crack model are deduced. To investigate the performance of the new model for crack sizing, cracks of different sizes and complex shapes were reconstructed using the new crack model and the fast forward ECT signal simulator using unflawed field database and the multimedia finite element scheme. Meanwhile, inverse analyses for sizing cracks of complicated shapes were also carried out based on ECT signals containing random noise to clarify the robustness of the new inversion scheme. The numerical results show that reconstructed crack parameters are identical with the true ones in acceptable accuracy, which proved the validity and efficiency of the new method.