A reinforcement learning approach for individualizing erythropoietin dosages in hemodialysis patients

This paper presents a reinforcement learning (RL) approach for anemia management in patients undergoing chronic renal failure. Erythropoietin (EPO) is the treatment of choice for this kind of anemia but it is an expensive drug and with some dangerous side-effects that should be considered especially for patients who do not respond to the treatment. Therefore, an individualized treatment appears to be necessary. RL is a suitable approach to tackle this problem. Moreover, resulting policies are similar to medical protocols, and hence, they can easily be transferred to daily practice. A cohort of 64 patients are included in the study. An implementation of the Q-learning algorithm based on a state-aggregation table and another implementation using the multi-layer perceptron as a function approximator (Q-MLP) are compared with the protocols followed in the Nephrology Unit. The policy obtained by the Q-MLP approach outperforms the hospital policy in terms of the ratio of patients that are within the targeted range of hemoglobin (11.5–12.5 g/dl) at the end of the analyzed period, since an increase of 25% is observed. It ensures an improvement in patients’ quality-of-life and considerable economic savings for the health care system due to both the expensiveness of EPO treatment and the costs incurred by the health care system in order to alleviate problems related to EPO over-dosing. It should be pointed out that the approach presented here is completely general, and therefore, it can be applied to any problem of drug dosage optimization.     

A note on 3D orthogonal drawings with direction constrained edges

The recent interest in three-dimensional graph drawing has been motivating studies on how to extend two-dimensional techniques to higher dimensions. A common 2D approach for computing an orthogonal drawing separates the task of defining the shape of the drawing from the task of computing its coordinates. First results towards finding a three-dimensional counterpart of this approach are presented by G. Di Battista, et al. [Graph Drawing (Proc. GD'00), Lecture Notes in Comput. Sci., vol. 1984, Springer, Berlin, 2001; Theoret. Comput. Sci. 289 (2002) 897], where characterizations of orthogonal representations of paths and cycles are studied. In this paper we show that the characterization for cycles given by G. Di Battista, et al. [Theoret. Comput. Sci. 289 (2002) 897] does not immediately extend to even seemingly simple graphs.     

Silicene on non-metallic substrates: Recent theoretical and experimental advances

Silicene, the silicon counterpart of graphene, has been successfully grown on metallic substrates such as Ag(111), ZrB₂(0001), and Ir(111) surfaces. However, characterization of its electronic structure is hampered by the metallic substrate. In addition, potential applications of silicene in nanoelectronic devices will require its growth on or integration with semiconducting and insulating substrates. We herein present a review of recent theoretical works regarding the interaction of silicene with non-metallic templates, distinguishing between the weak van-der-Waals-like interactions of silicene with, for example, layered metal (di)chalcogenides, and the stronger covalent bonding between silicene and, for example, ZnS surfaces. We then present a methodology to effectively compare the stability of diverse silicene structures using thermodynamics and molecular dynamics density functional theory calculations. Recent experimental results on the growth of silicene on MoS₂ are also reported and compared to the theoretical predictions.

Open image in new window

     

Managing schema evolution in a container‐based persistent system

Managing schema evolution is a problem every persistent system has to cope with to be useful in practice. Schema evolution consists basically of supporting class modification and dealing with data objects created and stored under the old class definitions. Several proposals have been made to handle this problem in systems that follow a full orthogonally persistent approach, but, until now, there has not been any proposal to support it in container‐based persistent systems. In this paper we describe a schema evolution management system designed for Barbados. Barbados is a complete programming environment which is based on an architecture of containers to provide persistent storage. Barbados does not provide full orthogonal persistence, but, as will be described in this paper, its architecture has several other advantages. Among them is the fact that this model is especially suitable for solving the schema evolution problem. Copyright © 2002 John Wiley & Sons, Ltd.     

Automatic assignment of reviewers in an online peer assessment task based on social interactions

Online peer assessment tasks are very popular and have unique characteristics that improve learning and encourage social interactions in a distance education environment. Unfortunately, social factors have usually been ignored in the process of selecting reviewers for online peer assessment tasks. We hypothesise that this fact could have some influence on the lack of engagement and participation by some learners. For this reason, we propose an approach in which social network analysis techniques, expert criteria, and Bayesian reasoning are applied to select reviewers with the objective of increasing participation in peer review tasks. The approach is divided into two elements. On the one hand, we have developed an influence diagram template that structures a set of proposed social network analysis variables in accordance with expert criteria. This influence diagram template can be easily updated for any course simply by eliciting a minimal set of parameters. On the other hand, we have instantiated the proposed influence diagram template to produce an influence diagram network to quantify the quality of reviewer assignment for an online peer assessment task. In an online experiment, we verified that the consideration of social factors can increase participation in a peer assessment task.     

A SiGe BiCMOS phase shifter based on quarter‐wave coupled resonators

This work presents the first example of monolithically integrated phase shifter based on a pass‐band filter architecture. The proposed configuration was realized mapping a classical quarter‐wave coupled filter circuit into its lumped element equivalent. Phase control is achieved by controlling the pass‐band through tunable tanks employing varactor diodes. A demonstrator was prototyped in the 24 GHz ISM band using a 0.25μm SiGe BiCMOS technology. Experimental results show 180° of phase range and maximum transmission losses of 8 dB. The main feature of this configuration is that it allows controlling the transmission losses by design and that its size is extremely compact.     

The contributions of “minimum primary emissions” and “new particle formation enhancements” to the particle number concentration in urban air

In an urban site affected by fresh vehicle exhaust emissions, the ambient air number concentrations of particles coarser than 3 nm (N) was split into two components, N=N1+N2. This was done using a method based on the high correlation between black-carbon (BC) and number (N) concentrations which is typically observed in ambient air and is the result of vehicle exhaust emissions. The component N1 accounts for “those aerosol components directly emitted in the particle phase” and “those components nucleating immediately after emission”. The component N2 accounts for the new particle formation enhancements during the “dilution and cooling of the vehicle exhaust” and is also influenced by “in situ new particle formation in ambient air”. The contribution of N1 to N exhibits a maximum of 55% during the morning rush hours (07:00–08:00). The contribution of N2 to N exhibits a daily evolution with a broad maximum during daylight (as solar radiation intensity), while for about 7 h (11:00–17:00) the N2 contribution to N is about 70%. During some “afternoon N2 events”, N2 contributions exceeded 90%. Enhancements in the new particle formation processes may increase the N/BC concentrations ratio in one order of magnitude, from 4.82×10⁶ particles/ng BC to 47×10⁶ particles/ng BC and during some events up to 97×10⁶ particles/ng BC. The results show evidence of the high potential of the vehicle exhausts and of the urban atmosphere to trigger new particle formation if the ambient air conditions are favourable. The method used in this study is useful in assessing future changes in the number to BC relationship due to forthcoming regulations in the vehicle exhaust emissions.     

Kinetics of mechanically induced anatase-to-rutile phase transformations under inelastic impact conditions

This work investigates the kinetics of the anatase-to-rutile phase transformation of TiO₂ powders during ball milling. Two elementary stages are observed, namely the transformation of anatase to the TiO₂ II phase and of this latter to rutile. Such consecutive reactions were studied under inelastic impact regimes. Based on a suitable modeling of transformation kinetics, the fraction of powder processed on average at each collision was estimated. The mass of powder involved in phase transformation and microstructural refinement processes at each collision was then worked out by a systematic variation of the powder charge inside the reactor.