Article-level classification of scientific publications: A comparison of deep learning, direct citation and bibliographic coupling.

Classification schemes for scientific activity and publications underpin a large swath of research evaluation practices at the organizational, governmental, and national levels. Several research classifications are currently in use, and they require continuous work as new classification techniques becomes available and as new research topics emerge. Convolutional neural networks, a subset of "deep learning" approaches, have recently offered novel and highly performant methods for classifying voluminous corpora of text. This article benchmarks a deep learning classification technique on more than 40 million scientific articles and on tens of thousands of scholarly journals. The comparison is performed against bibliographic coupling-, direct citation-, and manual-based classifications-the established and most widely used approaches in the field of bibliometrics, and by extension, in many science and innovation policy activities such as grant competition management. The results reveal that the performance of this first iteration of a deep learning approach is equivalent to the graph-based bibliometric approaches. All methods presented are also on par with manual classification. Somewhat surprisingly, no machine learning approaches were found to clearly outperform the simple label propagation approach that is direct citation. In conclusion, deep learning is promising because it performed just as well as the other approaches but has more flexibility to be further improved. For example, a deep neural network incorporating information from the citation network is likely to hold the key to an even better classification algorithm.

Correction to: Informing materials: drugs as tools for exploring cancer mechanisms and pathways.

The original version of this article unfortunately contained a mistake. Three entries are incorrect in the reference list. The corrected references are given below.

Informing materials: drugs as tools for exploring cancer mechanisms and pathways.

This paper builds on previous work that investigated anticancer drugs as 'informed materials', i.e., substances that undergo an informational enrichment that situates them in a dense relational web of qualifications and measurements generated by clinical experiments and clinical trials. The paper analyzes the recent transformation of anticancer drugs from 'informed' to 'informing material'. Briefly put: in the post-genomic era, anti-cancer drugs have become instruments for the production of new biological, pathological, and therapeutic insights into the underlying etiology and evolution of cancer. Genomic platforms characterize individual patients' tumors based on their mutational landscapes. As part of this new approach, drugs targeting specific mutations transcend informational enrichment to become tools for informing (and destabilizing) their targets, while also problematizing the very notion of a 'target'. In other words, they have become tools for the exploration of cancer pathways and mechanisms. While several studies in the philosophy and history of biomedicine have called attention to the heuristic relevance and experimental use of drugs, few have investigated concrete instances of this role of drugs in clinical research.

Translational research policies: disruptions and continuities in biomedical innovation systems in Austria, Finland and Germany.

Increasing the rate of biomedical research that is relevant to clinical innovation has been an intensifying concern of the research community and of policy-makers. In response, some of these actors have recently promoted varied approaches they label as translational research (TR) and translational medicine. This movement started in the USA in the early 1990s, and has since evolved to encompass large and ambitious initiatives. Its advocates contend that the productivity of biomedical innovation systems can be bolstered by: (1) the extension of large-scale development collaborations; (2) the strengthening of clinical experimental platforms; (3) training and supporting dedicated human capital; (4) achieving higher collective coordination of research teams than was previously common practice. In this paper, we examine to which extent these objectives have been put into practice by communities of biomedical actors and policymakers, by characterizing current translational initiatives in three European countries-Austria, Finland and Germany. This research draws on an analysis of policy documents and 26 semi-structured interviews conducted with policy-makers and TR advocates from these countries. Traditions of science and technology policy-making in each country have made them differentially receptive to the TR movement. German biomedical actors have most fully put into practice TR propositions, while Finland has seen policy-level debate of the notions but little in the way of concrete implementation and Austria appears to be a middle case.

EU policies in personalized medicine-related technologies.

Against the background of a number of first drug-diagnostic co-products developed and introduced into the European market, European decision-makers feel impelled to react and position themselves in the field of personalized medicine. Their reactions cover a broad range, from the analysis of knowledge requirements for market approval to the need for translational activities and the possible contribution of pharmacogenetics to public health. This article summarizes the current positions of European institutions, based on literature review and expert consultation for three items associated with personalized medicine: biobanks, genetic diagnostics and drug-diagnostic co-products, and provides an outlook on requirements for an effective future European policy on personalized medicine.