Search

The COVID-19 corona virus has claimed 4.1 million lives, as of July 24, 2021. A variety of machine learning models have been applied to related data to predict important factors such as the severity of the disease, infection rate and discover important prognostic factors. Often the usefulness of the findings from the use of these techniques is reduced due to lack of method interpretability. Some recent progress made on the interpretability of machine learning models has the potential to unravel more insights while using conventional machine learning models. In this work, we analyze COVID-19 blood work data with some of the popular machine learning models; then we employ state-of-the-art post-hoc local interpretability techniques(e.g.- SHAP, LIME), and global interpretability techniques(e.g. - symbolic metamodeling) to the trained black-box models to draw interpretable conclusions. In the gamut of machine learning algorithms, regressions remain one of the simplest and most explainable models with clear mathematical formulation. We explore one of the most recent techniques called symbolic metamodeling to find the mathematical expression of the machine learning models for COVID-19. We identify Acute Kidney Injury (AKI), initial Albumin level (ALBI), Aspartate aminotransferase (ASTI), Total Bilirubin initial(TBILI) and D-Dimer initial (DIMER) as major prognostic factors of the disease severity. Our contributions are- (i) uncover the underlying mathematical expression for the black-box models on COVID-19 severity prediction task (ii) we are the first to apply symbolic metamodeling to this task, and (iii) discover important features and feature interactions., Comment: accepted by SIPAIM 2021, code repository: https://github.com/ananyajana/interpretablecovid19

Non-alcoholic fatty liver disease (NAFLD) is one of the most common causes of chronic liver diseases (CLD) which can progress to liver cancer. The severity and treatment of NAFLD is determined by NAFLD Activity Scores (NAS)and liver fibrosis stage, which are usually obtained from liver biopsy. However, biopsy is invasive in nature and involves risk of procedural complications. Current methods to predict the fibrosis and NAS scores from noninvasive CT images rely heavily on either a large annotated dataset or transfer learning using pretrained networks. However, the availability of a large annotated dataset cannot be always ensured andthere can be domain shifts when using transfer learning. In this work, we propose a self-supervised learning method to address both problems. As the NAFLD causes changes in the liver texture, we also propose to use texture encoded inputs to improve the performance of the model. Given a relatively small dataset with 30 patients, we employ a self-supervised network which achieves better performance than a network trained via transfer learning. The code is publicly available at https://github.com/ananyajana/fibrosis_code., Comment: 5 pages, 2 figures, accepted at ISBI 2021, code at this URL: https://github.com/ananyajana/fibrosis_code

Non-Alcoholic Fatty Liver Disease (NAFLD) is becoming increasingly prevalent in the world population. Without diagnosis at the right time, NAFLD can lead to non-alcoholic steatohepatitis (NASH) and subsequent liver damage. The diagnosis and treatment of NAFLD depend on the NAFLD activity score (NAS) and the liver fibrosis stage, which are usually evaluated from liver biopsies by pathologists. In this work, we propose a novel method to automatically predict NAS score and fibrosis stage from CT data that is non-invasive and inexpensive to obtain compared with liver biopsy. We also present a method to combine the information from CT and H\&E stained pathology data to improve the performance of NAS score and fibrosis stage prediction, when both types of data are available. This is of great value to assist the pathologists in computer-aided diagnosis process. Experiments on a 30-patient dataset illustrate the effectiveness of our method., Comment: 6 pages, 3 figures. Accepted in IEEE BIBE 2020

Esophageal cancer is the 6th leading cause of cancer death worldwide. Its development is associated with a variety of risk factors including tobacco use, heavy alcohol consumption, human papilloma virus infection, and certain dietary factors such as trace mineral and vitamin deficiencies. An association with ionizing radiation exposure is revealed by the high excess relative risk for squamous cell carcinoma of the esophagus observed in the survivors of the atomic bomb detonations in Japan. It is also seen as a secondary malignancy in patients who received radiotherapy for breast and thoracic cancers; additionally, patients with head/neck and oral squamous cell cancers are at increased risk for metachronous esophageal squamous cell cancers. This malignancy is rapidly fatal, mainly because it remains asymptomatic until late, advanced stages when the disease is rarely curable. The stromal microenvironment plays an essential role in the maintenance and modulation of normal epithelial cell growth and differentiation and cross talk between the epithelial and stromal compartments can influence many aspects of malignant progression, including tumor cell proliferation, migration, invasion and recruitment of new blood vessels. To test the hypothesis that radiation exposure plays a role in esophageal carcinogenesis via non-targeted mechanisms involving stromal-epithelial cell communication, we are studying radiation effects on hTERT-immortalized human esophageal epithelial cells and genetic variants grown in co-culture with human esophageal stromal fibroblasts (Okawa et al., Genes & Dev. 2007. 21: 2788-2803). We examined how radiation treatment of stromal fibroblasts affected epithelial migration and invasion, behaviors associated with cancer promotion and progression. Chemotactic and haptotactic migration of epithelial cells stimulated by conditioned media from irradiated fibroblasts was measured using assays conducted in Transwell cell culture chambers. Our results using low LET radiation showed a dose-dependent increase in migration of epithelial cells when exposed to conditioned media from irradiated vs. non-irradiated fibroblasts. We also observed enhanced invasion through a basement membrane simulant. To identify chemotactic proteins secreted by irradiated stromal fibroblasts, we used antibody capture cytokine arrays and have identified several proteins as candidates. Increased secretion of these factors by irradiated fibroblasts was confirmed using ELISA. We are currently analyzing the contribution of these individual factors on epithelial migration and invasion, as well as their influence on cell survival and DNA repair. Studies using high-LET radiation will help determine radiation quality effects on these processes. These results should further our understanding of the mechanisms by which radiation impacts the tissue microenvironment and how it influences cancer development processes.

A comparative study of three theoretical formalisms by Anholt et al. (1976), Scofield (1978), and Becker et al. (1985) for K-shell ionization by highly relativistic protons has been carried out. The importance of the spin-flip mechanism is emphasized. It is concluded that in the analysis of such inner-shell ionization processes, an accurate description of the projectile-target interaction is of much more significance than the accuracy of the wave functions of the atomic electrons.

The semirelativistic approach, as applied to sum-rule calculations, can yield very inaccurate results for high-Z systems; the apparently good agreement previously reported is fortuitous, due to the application of an approximate eigenstate of the semirelativistic Hamiltonian. Attention is given to one-electron atomic systems, and the results obtained are discussed with a view to their implications for many-electron systems. The many-electron problem is complicated by a lack of clarity as to the relative importance of contributions due to correlation effects among the electrons, their effect on the E2 oscillator strength, and the retardation correction to the E1 transition.

The theory of two-photon ionization of a hydrogenic state in the nonrelativistic dipole approximation is generalized for elliptically polarized light. An application to the metastable 2S state of atomic hydrogen is made. Significant differences in the angular distribution of the outgoing electrons are found depending upon the polarization of the photons. It is claimed that two-photon ionization employing elliptically polarized photons from lasers may provide an additional test for the theories of multiphoton ionization.

The recoil spectrum of nuclei in polymers for incident protons below 10 MeV in energy is calculated by using the Rutherford cross-section with screening corrections. Employing the work of McKinley and Feshbach on the Coulomb scattering of relativistic electrons by nuclei, the recoil spectrum of the nuclei is also calculated for electron energies varying from threshold to 10 MeV. The partitioning of energy between electronic excitation and ionization and nuclear recoil is then studied to search for possible difference in polymer radiation response as a function of radiation energy and type.

In view of the inapplicability of the asymptotic expressions for the stopping number available in the literature at high energies, an alternative approach is taken to compute the shell correction to the stopping number of K electrons. Anholt's formula for the K-shell ionization has been used to calculate the excitation function for longitudinal interaction and numerical integration over energy has been carried out to evaluate the shell correction. Comparison with other theoretical calculations is made. It is proposed that, with the inclusion of relativistic effects, an asymptotic expansion of the stopping number with a leading-term logarithmic in the energy of the incident particle would be more meaningful and might enable one to extract the relativistic contribution to the shell correction from it.

Relativistic corrections to the Bethe sum rule have been obtained for a one-electron system. The relativistic sum is first derived assuming completeness by employing the Dirac Hamiltonian. Then, a semirelativistic approach to order v-squared/c-squared is presented following the Foldy-Wouthuysen transformation scheme to exclude the contributions from the negative-energy states. The results have wide applications in high-Z systems at large momentum transfers.

The kinetic energies of ions produced by dissociative photoionization of NO have been measured at the discrete resonance lines of He (584A) and Ne (736A), and with undispersed synchrotron radiation. O sup + ions were identified with energies from 0 to approximately 0.5 eV and two groups of N sup + ions one with energy of 0.36 eV and another with energies between 0.9 and 1.5 eV, apparently produced by predissociation of the C sup 3 P 1 and B'1 sigma states respectively.

The Weinberg-Salam model is applied to quantify the energy loss of antineutrinos and neutrinos encountering polymers. The scattering cross-sectional energy due to encounters with electrons is calculated, along with the probability that an antineutrino will remain the same particle. The energy loss reaches a maximum, i.e., stopping occurs, when the probability is unity. The technique is applied to study the energy losses in kapton, a solid organic insulator used for antennas on spacecraft exposed to solar neutrinos with energies ranging from 0.5-10 MeV. The energy loss is found to be negligible.

Calculations are presented on the effect of the two-body charge and current density effects on the observables in the photodisintegration of a deuteron. The computations are performed in a cartesian coordinate system, initially for gamma rays plane-polarized in the x-direction. Unpolarized photons are treated with an incoherent superposition of transition amplitudes for gamma rays polarized in the x- and y- directions. Account is taken of recoil and interaction energies after an initial state is defined. Consideration is focused on the cross section, polarization, and asymmetry of the reaction at low and medium energy incident gamma rays.

Transmittance of thin films and photoionization of gases in selective detectors set for photons in ultraviolet spectrum

Optical and photoelectric properties of thin tin films in vacuum UV