Search

Peroxisome proliferator-activated receptor alpha (PPARα), a ligand-dependent transcription factor, regulates fatty acid metabolism in heart and skeletal muscle. The intron 7 G/C polymorphism (rs4253778) has been associated with athletic performance. The rare C-allele was predominant in power athletes, whereas the G-allele was more frequent in endurance athletes. In the present study, we investigated the association between this polymorphism and strength characteristics in nonathletic, healthy young adults (n = 500; age 24.2 ± 4.4 years). Knee torque was measured during concentric knee flexion and extension movements at 60°/s, 120°/s, and 240°/s during 3, 25, and 5 repetitions, respectively. Also, resistance to muscle fatigue (i.e. work last 20% repetitions/work first 20% repetitions *100) was calculated. Differences in knee strength phenotypes between GG homozygous individuals and C-allele carriers were analyzed. The polymorphism did not influence the ability to produce isometric or dynamic knee flexor or extensor peak torque during static or dynamic conditions in this population (0.23

ispartof: International Journal of Body Composition Research vol:9 issue:2 pages:49-56 status: published

status: published

In this chapter we have demonstrated that Learning Automata are interesting building blocks for multi-agent Reinforcement learning algorithms. LA can be viewed as policy iterators, that update their action probabilities based on private information only. Even in multi-automaton settings, each LA is updated using only the environment response, and not on the basis of any knowledge regarding the other automata, i.e. nor their strategies, nor their feedback. As such LA based agent algorithms are relatively simple and the resulting multi-automaton systems can still be treated analytically. Convergence proofs already exist for a variety of settings ranging from a single automaton model acting in a simple stationary random environment to a distributed automata model interacting in a complex environment. The above properties make LA attractive design tools for multi-agent learning applications, where communication is often expensive and payoffs are inherently stochastic. They allow to design multi-agent learning algorithms with different learning objectives. Furthermore, LA have also proved to be able to work in asynchronous settings, where the actions of the LA are not taken simultaneously and where reward comes with delay. We have demonstrated this design approach in 2 distinct multi-agent learning settings. In ergodic markov games each agent defers its action selection to a local automaton, associated with the current system state. Convergence to an equilibrium between agent policies can be established by approximating the problem by a limiting normal form game. In episodic multi-stage learning problems agents were designed as tree-structured hierarchies of automata, mimicking the structure of the environment. Convergence of this algorithm can again be established based on existing automata properties. By using Intermediate Rewards instead of Monte Carlo rewards, the hierarchical learning automata are shown (both empirically and theoretically) to have a faster and more accurate convergence by even using less information. However, the Intermediate Rewards update mechanism is still an off-line algorithm in which the updating happens at explicit end-states. The general n-step algorithm solves this problem by handing immediate rewards to the automata which use bootstrapping to compensate for the absence of reward of the remainder of the path. Empirical experiments show that the n-step rewards (with an appropriate value for n) outperform both the Monte Carlo technique as well as the Intermediate Rewards.

In this paper we tested the practical use of an existing learning algorithm, called Exploring Selfish Reinforcement Learning (ESRL), on conflicting interest games. This algorithm is based on the principles of learning automata and, for games where agents have conflicting interests, on a Homo egualis society. Furthermore we propose some variations on the exploration heuristic of the algorithm and observe how these influence the convergence, fairness and speed results.

This document describes the result of a thorough study towards the possible development and application of e-services in the home healthcare field (denoted as eHomeCare services). This vast study was essential for the successful start of the Coplintho project, as the design and development of e-services in a fairly new and complex application field such as the home healthcare sector comprises many unknown factors. The Coplintho consortium set out a discussion and interaction at large with care-givers as well as patients and possible users of future eHomeCare services. This study took more time and effort than originally planned, but the consortium invested the resources needed to come up with concise and valuable results. Literature surveys were done in depth and the available information analysed. Inclusion criteria were set up in view of a profound and funded choice with respect to the selection of patients for the proof-of-concept demo. The choice was made to base these inclusion criteria on pathologies, although many other user-specific aspects are also taken into the design phase of the proof-of-concept demo. In parallel to the above a thorough study of the existing juridical framework was performed to reveal how the introduction of e-services in the home healthcare field such as depicted in the Coplintho project should be designed and used to be in line with existing and emerging legal constraints. Within the report a thorough overview and analyses is given of the different legislations involved. It must be noted however that the juridical framework is evolving fast in the domain of e-health and privacy-protection so that evolving or changing legal aspects should be tracked carefully. The overall study resulted in several so-called use cases for the two selected patient groups (being diabetes and multiple sclerosis) being merely case stories to illustrate possible daily use in real-life scenarios. These use cases have proven to be a very effective tool to bridge between the user-groups and the technical groups within the consortium. It has proven to be a challenge to align and translate in an optimum way the user-driven needs with the technological-driven concepts within the complex application domain of home healthcare. It is clear that the elaboration of the use cases has leveraged to a great extent this precious process. It should be noted by the reader that the information contained in this study is only valid for the time frame is was conducted within. Some items (such as the juridical framework, the existing preconditions) may change during time as the e-health domain is rapidly evolving. The information contained in these sections will be updated in the last deliverable of WP1 (D1.2, M24). It is believed by the authors however that many parts of the document can be very useful for other researchers developing e-services of related nature, even beyond the domain of e-health. The consortium is willingly to share experience and research results with other projects / interested parties on request.