Not Just Par for the Course: 73 Quaternary Germanides RE4 M2 XGe4 ( RE = La-Nd, Sm, Gd-Tm, Lu; M = Mn-Ni; X = Ag, Cd) and the Search for Intermetallics with Low Thermal Conductivity.

A total of 73 new quaternary rare-earth germanides RE4 M2 XGe4 ( RE = rare-earth metal; M = Mn-Ni; X = Ag, Cd) were prepared through reactions of the elements. The solid solution Nd4Mn2Cd(Ge1- ySi y)4 was also prepared under the same conditions and found to be complete over the entire range. All of these compounds adopt the monoclinic Ho4Ni2InGe4-type structure (space group C2/ m, a = 14.2-16.7 Å, b = 4.0-4.6 Å, c = 6.8-7.5 Å, ß = 106-109°), as revealed by powder X-ray diffraction analysis and single-crystal X-ray diffraction analysis on selected members. The structure determination of Nd4(Mn0.78(1)Ag0.22(1))2Ag0.83(1)Ge4 disclosed disorder of Mn and Ag atoms within the tetrahedral site and Ag deficiencies within the square planar site. Within the solid solution Nd4Mn2Cd(Ge1- ySi y)4, the end-members and two intermediate members were structurally characterized; as the Si content increases, the Cd sites become less deficient and the individual [Mn2 Tt2] layers contract but become further apart from each other. Electronic band structure calculations confirm that the Ag-Ge or Cd-Ge bonds are the weakest in the structure and thus prone to distortion. Thermal property measurements confirm expectations from machine-learning predictions that these quaternary germanides should exhibit low thermal conductivity, which was found to be <10 W m-1 K-1 for Nd4Mn2AgGe4.

Benchmark datasets incorporating diverse tasks, sample sizes, material systems, and data heterogeneity for materials informatics.

Materials discovery via machine learning has become an increasingly popular method due to its ability to rapidly predict materials properties in a time-efficient and low-cost manner. However, one limitation in this field is the lack of benchmark datasets, particularly those that encompass the size, tasks, material systems, and data modalities present in the materials informatics literature. This makes it difficult to identify optimal machine learning model choices including algorithm, model architecture, data splitting, and data featurization for a given task. Here, we attempt to address this lack of benchmark datasets by assembling a unique repository of 50 different datasets for materials properties. The data contains both experimental and computational data, data suited for regression as well as classification, sizes ranging from 12 to 6354 samples, and materials systems spanning the diversity of materials research. Data were extracted from 16 publications. In addition to cleaning the data where necessary, each dataset was split into train, validation, and test splits. For datasets with more than 100 values, train-val-test splits were created, either with a 5-fold or 10-fold cross-validation method, depending on what each respective paper did in their studies. Datasets with less than 100 values had train-test splits created using the Leave-One-Out cross-validation method. These benchmark data can serve as a basis for a more diverse benchmark dataset in the future to further improve their effectiveness in the comparison of machine learning models.

Skin Electrical Resistance as a Diagnostic and Therapeutic Biomarker of Breast Cancer Measuring Lymphatic Regions.

Skin changes associated with alterations in the interstitial matrix and lymph system might provide significant and measurable effects due to the presence of breast cancer. This study aimed to determine if skin electrical resistance changes could serve as a diagnostic and therapeutic biomarker associated with physiological changes in patients with malignant versus benign breast cancer lesions. Forty-eight women (24 with malignant cancer, 23 with benign lesions) were enrolled in this study. Repeated skin resistance measurements were performed within the same session and 1 week after the first measurement in the breast lymphatic region and non-breast lymphathic regions. Intraclass correlation coefficients were calculated to determine the technique's intrasession and intersession reproducibility. Data were then normalized as a mean of comparing cross-sectional differences between malignant and benign lesions of the breast. Six months longitudinal data from six patients that received therapy were analyzed to detect the effect of therapy. Standard descriptive statistics were used to compare ratiometric differences between groups. Skin resistance data were used to train a machine learning random forest classification algorithm to diagnose breast cancer lesions. Significant differences between malignant and benign breast lesions were obtained (p<0.01), also pre- and post-treatment (p<0.05). The diagnostic algorithm demonstrated the capability to classify breast cancer with an area under the curve of 0.68, sensitivity of 66.3%, specificity of 78.5%, positive predictive value 70.7% and negative predictive value 75.1%. Measurement of skin resistance in patients with breast cancer may serve as a convenient screening tool for breast cancer and evaluation of therapy. Further work is warranted to improve our approach and further investigate the biophysical mechanisms leading to the observed changes.