Using Machine Learning Classifiers to Assist Healthcare-Related Decisions: Classification of Electronic Patient Records

Surveillance Levels (SLs) are categories for medical patients (used in Brazil) that represent different types of medical recommendations. SLs are defined according to risk factors and the medical and developmental history of patients. Each SL is associated with specific educational and clinical measures. The objective of the present paper was to verify computer-aided, automatic assignment of SLs. The present paper proposes a computer-aided approach for automatic recommendation of SLs. The approach is based on the classification of information from patient electronic records. For this purpose, a software architecture composed of three layers was developed. The architecture is formed by a classification layer that includes a linguistic module and machine learning classification modules. The classification layer allows for the use of different classification methods, including the use of preprocessed, normalized language data drawn from the linguistic module. We report the verification and validation of the software architecture in a Brazilian pediatric healthcare institution. The results indicate that selection of attributes can have a great effect on the performance of the system. Nonetheless, our automatic recommendation of surveillance level can still benefit from improvements in processing procedures when the linguistic module is applied prior to classification. Results from our efforts can be applied to different types of medical systems. The results of systems supported by the framework presented in this paper may be used by healthcare and governmental institutions to improve healthcare services in terms of establishing preventive measures and alerting authorities about the possibility of an epidemic.

Contrasted crustal sources as defined by whole-rock and Sr-Nd-Pb isotope geochemistry of neoproterozoic early post-collisional granitic magmatism within the Southern Brazilian Shear Belt, Camboriu, Brazil

The early phase of post-collisional granitic magmatism in the Camboriu region, south Brazil, is represented by the porphyritic biotite +/- hornblende Rio Pequeno Granite (RPG; 630-620 Ma) and the younger (similar to 610 Ma), equigranular, biotite +/- muscovite Serra dos Macacos Granite (SMG). The two granite types share some geochemical characteristics, but the more felsic SMG constitutes a distinctive group not related to RPG by simple fractionation processes, as indicated by its lower FeOt, TiO2, K2O/Na2O and higher Zr Al2O3, Na2O, Ba and Sr when compared to RPG of similar SiO2 range. Sr-Nd-Pb isotopes require different sources. The SMG derives from old crustal sources, possibly related to the Paleoproterozoic protoliths of the Camboriu Complex, as indicated by strongly negative epsilon Nd-t (-23 to -24) and unradiogenic Pb (e.g., Pb-206/Pb-204 = 16.0-16.3; Pb-207/Pb-204 = 15.3-15.4) and confirmed by previous LA-MC-ICPMS data showing dominant zircon inheritance of Archean to Paleoproterozoic age. In contrast, the RPG shows less negative epsilon Nd-t (-12 to -15) and a distinctive zircon inheritance pattern with no traces of post-1.6 Ga sources. This is indicative of younger sources whose significance in the regional context is still unclear; some contribution of mantle-derived magmas is indicated by coeval mafic dykes and may account for some of the geochemical and isotopic characteristics of the least differentiated varieties of the RPG. The transcurrent tectonics seems to have played an essential role in the generation of mantle-derived magmas despite their emplacement within a low-strain zone. It may have facilitated their interaction with crustal melts which seem to be to a large extent the products of reworking of Paleoproterozoic orthogneisses from the Camboriu Complex. (C) 2012 Elsevier Ltd. All rights reserved.

Elemental abundances and classification of carbon-enhanced metal-poor stars

We present a detailed study of carbon-enhanced metal-poor (CEMP) stars, based on high-resolution spectroscopic observations of a sample of 18 stars. The stellar spectra for this sample were obtained at the 4.2 m William Herschel Telescope in 2001 and 2002, using the Utrecht Echelle Spectrograph, at a resolving power R similar to 52 000 and S/N similar to 40, covering the wavelength range lambda lambda 3700-5700 angstrom. The atmospheric parameters determined for this sample indicate temperatures ranging from 4750 K to 7100 K, log g from 1.5 to 4.3, and metallicities -3.0 <= [Fe/H]<=-1.7. Elemental abundances for C, Na, Mg, Sc, Ti, Cr, Cu, Zn, Sr, Y, Zr, Ba, La, Ce, Nd, Sm, Eu, Gd, Dy are determined. Abundances for an additional 109 stars were taken from the literature and combined with the data of our sample. The literature sample reveals a lack of reliable abundance estimates for species that might be associated with the r-process elements for about 67% of CEMP stars, preventing a complete understanding of this class of stars, since [Ba/Eu] ratios are used to classify them. Although eight stars in our observed sample are also found in the literature sample, Eu abundances or limits are determined for four of these stars for the first time. From the observed correlations between C, Ba, and Eu, we argue that the CEMP-r/s class has the same astronomical origin as CEMP-s stars, highlighting the need for a more complete understanding of Eu production.

A New Subgeneric Classification of Rhipsalis (Cactoideae, Cactaceae)

Most Cactaceae have succulent stems and inhabit dry or arid areas, but some are epiphytes of humid regions. Rhipsalis is the largest genus of epiphytic cacti. Species of Rhipsalis are notoriously difficult to identify, and the subgeneric classification of the genus has remained controversial. Between 1837 and 1995, eight different subgeneric classifications have been proposed for Rhipsalis. The most comprehensive taxonomic treatment of the genus recognized five subgenera, Phyllarthrorhipsalis, Rhipsalis, Epallagogonium, Calamorhipsalis, and Erythrorhipsalis, characterized mainly by stem morphology. Here, molecular phylogenetic information combined with morphological data is used to re-evaluate the former subgeneric classifications proposed for the genus. Three monophyletic subgenera are recognized, Rhipsalis, Calamorhipsalis and Erythrorhipsalis, which are mainly characterized by floral traits. The changes proposed include expanding the circumscription of Rhipsalis by the inclusion of species previously included in Phyllarthrorhipsalis and Epallagogoniwn and recognizing a broader Calamorhipsalis, also including species from subgenus Epallagogonium. The circumscription of Erythrorhipsalis remains unchanged. For each subgenus a list of synonyms, a brief description and a list of species included are presented. A key for the identification of subgenera is also provided.

A family history of serious complications due to BCG vaccination is a tool for the early diagnosis of severe primary immunodeficiency

Severe Combined Immunodeficiency (SCID) is one of the most severe forms of primary immunodeficiency (PID). Complications of BCG vaccination, especially disseminated infection and its most severe forms, are known to occur in immunodeficient patients, particularly in SCID. A carefully taken family history before BCG injection as well as delaying vaccination if PID is suspected could be a simple and effective method to avoid inappropriate vaccination of an immunodeficient child in some cases until the prospect of newborn screening for SCID has been fully developed. We describe a patient with a very early diagnosis of SCID, which was suspected on the basis of the previous death of two siblings younger than one year due to severe complications secondary to the BCG vaccine. We suggest that a family history of severe or fatal reactions to BCG should be included as a warning sign for an early diagnosis of SCID.