Development and external validation of a mixed-effects deep learning model to diagnose COVID-19 from CT imaging.

Background: The automatic analysis of medical images has the potential improve diagnostic accuracy while reducing the strain on clinicians. Current methods analyzing 3D-like imaging data, such as computerized tomography imaging, often treat each image slice as individual slices. This may not be able to appropriately model the relationship between slices. Methods: Our proposed method utilizes a mixed-effects model within the deep learning framework to model the relationship between slices. We externally validated this method on a data set taken from a different country and compared our results against other proposed methods. We evaluated the discrimination, calibration, and clinical usefulness of our model using a range of measures. Finally, we carried out a sensitivity analysis to demonstrate our methods robustness to noise and missing data. Results: In the external geographic validation set our model showed excellent performance with an AUROC of 0.930 (95%CI: 0.914, 0.947), with a sensitivity and specificity, PPV, and NPV of 0.778 (0.720, 0.828), 0.882 (0.853, 0.908), 0.744 (0.686, 0.797), and 0.900 (0.872, 0.924) at the 0.5 probability cut-off point. Our model also maintained good calibration in the external validation dataset, while other methods showed poor calibration. Conclusion: Deep learning can reduce stress on healthcare systems by automatically screening CT imaging for COVID-19. Our method showed improved generalizability in external validation compared to previous published methods. However, deep learning models must be robustly assessed using various performance measures and externally validated in each setting. In addition, best practice guidelines for developing and reporting predictive models are vital for the safe adoption of such models.

Bilateral adaptive graph convolutional network on CT based Covid-19 diagnosis with uncertainty-aware consensus-assisted multiple instance learning.

Coronavirus disease (COVID-19) has caused a worldwide pandemic, putting millions of people's health and lives in jeopardy. Detecting infected patients early on chest computed tomography (CT) is critical in combating COVID-19. Harnessing uncertainty-aware consensus-assisted multiple instance learning (UC-MIL), we propose to diagnose COVID-19 using a new bilateral adaptive graph-based (BA-GCN) model that can use both 2D and 3D discriminative information in 3D CT volumes with arbitrary number of slices. Given the importance of lung segmentation for this task, we have created the largest manual annotation dataset so far with 7,768 slices from COVID-19 patients, and have used it to train a 2D segmentation model to segment the lungs from individual slices and mask the lungs as the regions of interest for the subsequent analyses. We then used the UC-MIL model to estimate the uncertainty of each prediction and the consensus between multiple predictions on each CT slice to automatically select a fixed number of CT slices with reliable predictions for the subsequent model reasoning. Finally, we adaptively constructed a BA-GCN with vertices from different granularity levels (2D and 3D) to aggregate multi-level features for the final diagnosis with the benefits of the graph convolution network's superiority to tackle cross-granularity relationships. Experimental results on three largest COVID-19 CT datasets demonstrated that our model can produce reliable and accurate COVID-19 predictions using CT volumes with any number of slices, which outperforms existing approaches in terms of learning and generalisation ability. To promote reproducible research, we have made the datasets, including the manual annotations and cleaned CT dataset, as well as the implementation code, available at https://doi.org/10.5281/zenodo.6361963.

Quasiclassical trajectory study of H+SiH4 reactions in full-dimensionality reveals atomic-level mechanisms.

This work elucidates new atomic-level mechanisms that may be common in a range of chemical reactions, and our findings are important for the understanding of the nature of polyatomic abstraction and exchange reactions. A global 12-dimensional ab initio potential energy surface (PES), which describes both H+SiH(4) abstraction and exchange reactions is constructed, based on the modified Shepard interpolation method and UCCSD(T)/cc-pVQZ energy calculations at 4,015 geometries. This PES has a classical barrier height of 5.35 kcal/mol for abstraction (our best estimate is 5.35 +/- 0.15 kcal/mol from extensive ab initio calculations), and an exothermicity of -13.12 kcal/mol, in excellent agreement with experiment. Quasiclassical trajectory calculations on this new PES reveal interesting features of detailed dynamical quantities and underlying new mechanisms. Our calculated product angular distributions for exchange are in the forward hemisphere with a tail sideways, and are attributed to the combination of three mechanisms: inversion, torsion-tilt, and side-inversion. With increase of collision energy our calculated angular distributions for abstraction first peak at backward scattering and then shift toward smaller scattering angles, which is explained by a competition between rebound and stripping mechanisms; here stripping is seen at much lower energies, but is conceptually similar to what was observed in the reaction of H+CD(4) by Zare and coworkers [Camden JP, et al. (2005) J Am Chem Soc 127:11898-11899]. Each of these atomic-level mechanisms is confirmed by direct examination of trajectories, and two of them (torsion-tilt and side-inversion) are proposed and designated in this work.

Quasiclassical trajectory study of the SiH(4)+H-->SiH(3)+H(2) reaction on a global ab initio potential energy surface.

The SiH(4)+H-->SiH(3)+H(2) reaction has been investigated by the quasiclassical trajectory (QCT) method on a recent global ab initio potential energy surface [M. Wang et al., J. Chem. Phys. 124, 234311 (2006)]. The integral cross section as a function of collision energy and thermal rate coefficient for the temperature range of 300-1600 K have been obtained. At the collision energy of 9.41 kcalmol, product energy distributions and rovibrational populations are explored in detail, and H(2) rotational state distributions show a clear evidence of two reaction mechanisms. One is the conventional rebound mechanism and the other is the stripping mechanism similar to what has recently been found in the reaction of CD(4)+H [J. P. Camden et al., J. Am. Chem. Soc. 127, 11898 (2005)]. The computed rate coefficients with the zero-point energy correction are in good agreement with the available experimental data.

A global 12-dimensional ab initio potential energy surface and dynamical studies for the SiH4+H-->SiH3+H2 reaction.

A global 12-dimensional ab initio interpolated potential energy surface (PES) for the SiH(4)+H-->SiH(3)+H(2) reaction is presented. The ab initio calculations are based on the unrestricted quadratic configuration interaction treatment with all single and double excitations together with the cc-pVTZ basis set, and the modified Shepard interpolation method of Collins and co-workers [K. C. Thompson et al., J. Chem. Phys. 108, 8302 (1998); M. A. Collins, Theor. Chem. Acc. 108, 313 (2002); R. P. A. Bettens and M. A. Collins, J. Chem. Phys. 111, 816 (1999)] is applied. Using this PES, classical trajectory and variational transition state theory calculations have been carried out, and the computed rate constants are in good agreement with the available experimental data.