Dr Vijay Kumar Foundation

The accurate diagnosis and personalized treatment of renal cancer are significant challenges due to its complexity and heterogeneity. Machine learning (ML) techniques have opened new avenues for precise cancer classification due to recent gene expression microarray analysis advancements. This study pro- poses a novel MLbased diagnostic framework for renal cancer. This framework integrates multi-source gene expression datasets with identical features, followed by advanced preprocessing, scaling, and dimensionality reduction methods. The goal is to improve classification accuracy while preserving the model's generalizability. The five classifiers-Random Forest (RF), Ex- treme Gradient Boosting (XGBoost), K-Nearest Neighbor (KNN), Decision Tree (DT), and Support Vector Machine (SVM)-were evaluated with five performance metrics and optimized using hyperparameter tuning. The uniqueness of this work is derived from the integration of datasets, the rigorous preprocessing, and feature reduction through PCA, which collectively enhance the robustness of classification. Experimental results indicate that post-PCA scaling significantly improves model performance, with KNN, XGBoost, and RF achieving an accurate improvement of 88.46 %. Furthermore, the investigation shows that the variance in effect size decreases as sample sizes increase, while classification accuracy and effect size both increases. These results emphasize the potential of the proposed framework as a dependable instru-ment for the early and precise diagnosis of renal cancer, thereby facilitating the development of more informed clinical decisionmaking.

The structural, electronic, and elastic properties of pristine and carbon‐doped boron suboxide (B6O) are calculated using density functional theory. The results indicate that it is energetically preferable for a single carbon atom to substitute into an oxygen site rather than a boron site. The lattice parameters and cell volume increase to relieve the residual stress created by the carbon substitution. The interstitial position is not favorable for a single atom substitution. However, if two carbon atoms substitute for two neighboring oxygen atoms, then it becomes energetically favorable to dope an interstitial oxygen, boron, or carbon atom along the C–C chain. If the interstitial dopant is either boron or carbon, a local B4C‐like structure with either a C–B–C or C–C–C chain is created within the boron suboxide unit cell. The resulting structure shows improvements in the bulk modulus at the expense of the shear and Young's moduli. The moduli further improve if an additional carbon is substituted within a polar or equatorial site of the neighboring B12 icosahedron. Based on these calculations, we conclude that carbon doping can either harden or soften B6O depending on the manner in which the substitutions are populated. Furthermore, as B6O samples are often oxygen deficient, C doping can occupy such sites and improve the elastic properties.

The ground state structures of neutral and anionic clusters of Na(n)Si(m) (1 ≤ n ≤ 3, 1 ≤ m ≤ 11) have been determined using genetic algorithm incorporated in first principles total energy code. The size dependence of the structural and electronic properties is discussed in detail. It is found that the lowest-energy structures of Na(n)Si(m) clusters resemble those of the pure Si clusters. Interestingly, Na atoms in neutral Na(n)Si(m) clusters are usually well separated by the Si(m) skeleton, whereas Na atoms can form Na-Na bonds in some anionic clusters. The ionization potentials, adiabatic electron affinities, and photoelectron spectra are also calculated and the results compare well with the experimental data.

We report results of ab initio pseudopotential calculations on the nanocoating of gold on an icosahedral ${\text{Al}}_{13}$ cluster and the findings of icosahedrally symmetric endohedral $\text{Al}@{\text{Al}}_{12}{\text{Au}}_{20}^{\ensuremath{-}}$ and empty cage ${\text{Al}}_{12}{\text{Au}}_{20}^{2\ensuremath{-}}$ compound fullerenes formed of metal atoms. Twelve Al atoms cap the pentagonal faces of a dodecahedral ${\text{Au}}_{20}$ cage in which each Au atom has three Al atoms and three Au atoms as nearest neighbors. Mixing of ${\text{Al}}_{13}$ and ${\text{Au}}_{20}$ magic clusters leads to a large heat of formation of 0.55 eV/atom and high stability of the $\text{Al}@{\text{Al}}_{12}{\text{Au}}_{20}$ compound fullerene. The binding energies of ${\text{Al}}_{12}{\text{Au}}_{20}$ and $\text{Al}@{\text{Al}}_{12}{\text{Au}}_{20}$ are 3.017 and 3.007 eV/atom, respectively, which are much larger than 2.457 eV/atom for ${\text{Au}}_{32}$ fullerene, leading to the possibility of their high abundance.

We study hexagonal assemblies of M6X6 (M=Mo and W and X=S, Se, and Te) nanowires from first-principles calculations to understand their structural stability, electronic properties, and the effects of Li intercalation. It is shown that due to van der Waals interactions between the nanowires, the intercalation is achieved without a significant change in their atomic structure. With an increase in Li concentration, we predict a new phase for Li3Mo6S6 compound, in which the hexagonal assembly transforms to a monoclinic structure by a change in the orientation of nanowires. The LixMo6S6 assemblies are electrically conductive and can be potentially used as cathode materials in Li-ion batteries for nanoscale applications. The voltage of such a battery, calculated to be 1.7V, can be manipulated such as by iodine doping without a significant change in the atomic structure.

The variations in the atomic and electronic structures as well as the elastic properties of perovskite-type RRh3BX (R=Sc, Y, La, and Ce) compounds are understood as a function of X in terms of the changes in the cohesive energy and the covalent and ionic bonding using ab initio calculations. A strong covalent bonding is obtained between B 2p and Rh 4d orbitals in all cases. For R=Ce, f-d hybridization is significant and the Fermi level lies in a pseudogap similar to the case where B is replaced by C, leading to further possibilities of designing such compounds.

Using ab initio calculations, we find high stability of octahedral Mo6S8 clusters, which can further be condensed to form Mo3nS3n+2 (n, an integer) nanowires. These linear structures are energetically more favorable compared with other closed-packed polyhedral isomers of Mo-S clusters. The octahedral units in nanowires are stabilized by strong Mo-Mo interactions and p-d hybridization between Mo 4d and S 2p orbitals. There is a free electron-like band that crosses the Fermi energy in infinite nanowires and leads to their metallic character. Iodine doping acts as electron donor and can be used to tailor the electronic conductivity. For Mo12S8I4 nanowires, both electrons and holes are found to contribute to conduction. These nanowires are energetically more favorable than the experimentally obtained Mo12S6I12 nanowires.

Using first principles calculations within the generalized gradient approximation, the structural stability of compound Mo-S clusters at subnanometer level is discussed based on results obtained for various Mo5Sn isomers with n ¼ 5{15. The symmetrical Mo-S bonds generally lead to higher stability, while the formation of S-S bond and the observation of magnetism in these clusters lower the structural stability. Mo5Sn isomers with n ¼ 6, 8, and 11 are found to be most stable. The addition of sulfur atoms is quite favorable in the edge capped square pyramid and face capped trigonal bipyramid isomers. [doi:10.2320/matertrans.48.658]

Using ab initio calculations, we show that the encapsulation of Y, La, and Ac metal $(M)$ atom stabilizes the dodecahedral fullerene anion $M@\mathrm{Si}_{20}{}^{\ensuremath{-}}$ in the icosahedral symmetry. Similar to ${\mathrm{C}}_{60}$, it is the ideal cage of silicon and the largest that can be stabilized by an $M$ atom. Doping of other rare earths is further shown to stabilize magnetic dodecahedral fullerenes $\mathrm{Pa}@{\mathrm{Si}}_{20}$, $\mathrm{Sm}@{\mathrm{Si}}_{20}$, $\mathrm{Pu}@{\mathrm{Si}}_{20}$, and $\mathrm{Tm}@{\mathrm{Si}}_{20}$ with $1{\ensuremath{\mu}}_{\mathrm{B}}$, $4{\ensuremath{\mu}}_{\mathrm{B}}$, $4{\ensuremath{\mu}}_{\mathrm{B}}$, and $3{\ensuremath{\mu}}_{\mathrm{B}}$ spin magnetic moments, respectively, in contrast to most previous studies on $M$-encapsulated Si clusters in which the magnetic moment is completely quenched. The highest spin magnetic moment of $7{\ensuremath{\mu}}_{\mathrm{B}}$ is achieved for $\mathrm{Gd}@\mathrm{Si}_{20}{}^{\ensuremath{-}}$ with half-filled $4f$ states. The orbital magnetic moment is also calculated and it is $\ensuremath{\sim}1{\ensuremath{\mu}}_{\mathrm{B}}$ in most cases. Neutral $M@{\mathrm{Si}}_{20}$ ($M=\mathrm{Y}$, La, Ac, and Gd) behaves like superhalogen and interaction with a noble or alkali metal atom leads to salt like behavior. These findings could pave way for the realization of silicon fullerenes by doping of several elements.

The ability of silica to form different phases can be used to stabilize its nanostructures. Here we explore the stability of thin nanotubes of ${\mathrm{SiO}}_{x}$ ($x=1$ and 2) using ab initio calculations with the generalized gradient approximation for the exchange-correlation energy. We find that the pentagonal nanotubes are energetically most stable. The pentagonal SiO nanotube is a semiconductor with the largest calculated band gap of 0.90 eV, which is close to the value for bulk Si. The ${\mathrm{SiO}}_{2}$ nanotubes are, however, insulating similar to bulk silica and could be promising as the thinnest insulating layers for nanodevices. Our results demonstrate that we can get the most important circuit elements for nanoelectronics, namely semiconducting, as well as insulating nanotubes based on silicon in the subnanometer regime.