Adenosine deaminase and tuberculous meningitis-A systematic review with meta-analysis

Tuberculous meningitis (TBM) is a severe infection of the central nervous system, particularly in developing countries. Prompt diagnosis and treatment are necessary to decrease the high rates of disability and death associated with TBM. The diagnosis is often time and labour intensive; thus, a simple, accurate and rapid diagnostic test is needed. The adenosine deaminase (ADA) activity test is a rapid test that has been used for the diagnosis of the pleural, peritoneal and pericardial forms of tuberculosis. However, the usefulness of ADA in TBM is uncertain. The aim of this study was to evaluate ADA as a diagnostic test for TBM in a systematic review. A systematic search was performed of the medical literature (MEDLINE, LILACS, Web of Science and EMBASE). The ADA values from TBM cases and controls (diagnosed with other types of meningitis) were necessary to calculate the sensitivity and specificity. Out of a total of 522 studies, 13 were included in the meta-analysis (380 patients with TBM). The sensitivity, specificity and diagnostic odds ratios (DOR) were calculated based on arbitrary ADA cut-off values from 1 to 10 U/l. ADA values from 1 to 4 U/l (sensitivity > 93% and specificity < 80%) helped to exclude TBM; values between 4 and 8 U/l were insufficient to confirm or exclude the diagnosis of TBM (p = 0.07), and values > 8 U/l (sensitivity < 59% and specificity > 96%) improved the diagnosis of TBM (p < 0.001). None of the cut-off values could be used to discriminate between TBM and bacterial meningitis. In conclusion, ADA cannot distinguish between bacterial meningitis and TBM, but using ranges of ADA values could be important to improve TBM diagnosis, particularly after bacterial meningitis has been ruled out. The different methods used to measure ADA and the heterogeneity of data do not allow standardization of this test as a routine.

The application of adaptive partitioned random search in feature selection problem

Feature selection is one of important and frequently used techniques in data preprocessing. It can improve the efficiency and the effectiveness of data mining by reducing the dimensions of feature space and removing the irrelevant and redundant information. Feature selection can be viewed as a global optimization problem of finding a minimum set of M relevant features that describes the dataset as well as the original N attributes. In this paper, we apply the adaptive partitioned random search strategy into our feature selection algorithm. Under this search strategy, the partition structure and evaluation function is proposed for feature selection problem. This algorithm ensures the global optimal solution in theory and avoids complete randomness in search direction. The good property of our algorithm is shown through the theoretical analysis.

Plant-antivenom: Database of anti-venom medicinal plants

Plant-antivenom is a computational Websystem about medicinal plants with anti-venom properties. The system consists of a database of these plants, including scientific publications on this subject and amino acid sequences of active principles from venomous animals. The system relates these data allowing their integration through different search applications. For the development of the system, the first surveys were conducted in scientific literature, allowing the creation of a publication database in a library for reading and user interaction. Then, classes of categories were created, allowing the use of tags and the organization of content. This database on medicinal plants has information such as family, species, isolated compounds, activity, inhibited animal venoms, among others. Provision is made for submission of new information by registered users, by the use of wiki tools. Content submitted is released in accordance to permission rules defined by the system. The database on biological venom protein amino acid sequences was structured from the essential information from National Center for Biotechnology Information (NCBI). Plant-antivenom`s interface is simple, contributing to a fast and functional access to the system and the integration of different data registered on it. Plant-antivenom system is available on the Internet at http://gbi.fmrp.usp.br/plantantivenom.

Asymmetric page split generalized index search trees for formal concept analysis

Formal Concept Analysis is an unsupervised machine learning technique that has successfully been applied to document organisation by considering documents as objects and keywords as attributes. The basic algorithms of Formal Concept Analysis then allow an intelligent information retrieval system to cluster documents according to keyword views. This paper investigates the scalability of this idea. In particular we present the results of applying spatial data structures to large datasets in formal concept analysis. Our experiments are motivated by the application of the Formal Concept Analysis idea of a virtual filesystem [11,17,15]. In particular the libferris [1] Semantic File System. This paper presents customizations to an RD-Tree Generalized Index Search Tree based index structure to better support the application of Formal Concept Analysis to large data sources.

Feasibility of an evidence-based literature search service for general practitioners

Objective: To test the feasibility of an evidence-based clinical literature search service to help answer general practitioners' (GPs') clinical questions. Design: Two search services supplied GPs who submitted questions with the best available empirical evidence to answer these questions. The GPs provided feedback on the value of the service, and concordance of answers from the two search services was assessed. Setting: Two literature search services (Queensland and Victoria), operating for nine months from February 1999. Main outcome measures: Use of the service; time taken to locate answers; availability of evidence; value of the service to GPs; and consistency of answers from the two services. Results: 58 GPs asked 160 questions (29 asked one, 11 asked five or more). The questions concerned treatment (65%), aetiology (17%), prognosis (13%), and diagnosis (5%). Answering a question took a mean of 3 hours 32 minutes of personnel time (95% Cl, 2.67-3.97); nine questions took longer than 10 hours each to answer, the longest taking 23 hours 30 minutes. Evidence of suitable quality to provide a sound answer was available for 126 (79%) questions. Feedback data for 84 (53%) questions, provided by 42 GPs, showed that they appreciated the service, and asking the questions changed clinical care. There were many minor differences between the answers from the two centres, and substantial differences in the evidence found for 4/14 questions. However, conclusions reached were largely similar, with no or only minor differences for all questions. Conclusions: It is feasible to provide a literature search service, but further assessment is needed to establish its cost effectiveness.