Matteo Manca

Development and validation of a general approach to predict and quantify the synergism of anti-cancer drugs using experimental design and artificial neural networks.

The combination of two or more drugs using multidrug mixtures is a trend in the treatment of cancer. The goal is to search for a synergistic effect and thereby reduce the required dose and inhibit the development of resistance. An advanced model-free approach for data exploration and analysis, based on artificial neural networks (ANN) and experimental design is proposed to predict and quantify the synergism of drugs. The proposed method non-linearly correlates the concentrations of drugs with the cytotoxicity of the mixture, providing the possibility of choosing the optimal drug combination that gives the maximum synergism. The use of ANN allows for the prediction of the cytotoxicity of each combination of drugs in the chosen concentration interval. The method was validated by preparing and experimentally testing the combinations with the predicted highest synergistic effect. In all cases, the data predicted by the network were experimentally confirmed. The method was applied to several binary mixtures of cisplatin and [Cu(1,10-orthophenanthroline)2(H2O)](ClO4)2, Cu(1,10-orthophenanthroline)(H2O)2(ClO4)2 or [Cu(1,10-orthophenanthroline)2(imidazolidine-2-thione)](ClO4)2. The cytotoxicity of the two drugs, alone and in combination, was determined against human acute T-lymphoblastic leukemia cells (CCRF-CEM). For all systems, a synergistic effect was found for selected combinations.

Everywhere run: a virtual personal trainer for supporting people in their running activity

In the last years many medical researches have reported an increase of health problems in developed countries, mostly related to a sedentary lifestyle (as obesity and linked pathologies like diabetes and cardiovascular diseases). As a consequence. many research efforts have been carried out for finding strategies for motivating people to exercise regularly. In this paper we present an Android-based mobile application, called Everywhere Run [1], that aims at motivating and supporting people during their running activities, behaving as a virtual personal trainer. Everywhere Run fosters the interaction between users and real personal trainers, in order to make it easy to non expert people to start working out in a healthy and safe way.

Design and architecture of a friend recommender system in the social bookmarking domain

Social media systems are becoming more and more popular nowadays. In order to face the overload in the amount of users and content available in these systems, social recommender systems have been developed and are largely studied in the literature. A form of social media, known as social bookmarking system, allows to share bookmarks in a social network. A user adds as a friend or follows another user and receives updates on the bookmarks added by that user. However, no approach in the literature proposes friend recommender systems in the social bookmarking domain. In this paper, we present an analysis of the state-of-the-art on user recommendation in social environments and of the structure of a social bookmarking system, in order to derive a design and an architecture of a friend recommender system in the social bookmarking domain. This study can be useful for any future research in this area, by highlighting the aspects that characterize this domain and the features that this type of recommender system has to offer.

Behavioral data mining to produce novel and serendipitous friend recommendations in a social bookmarking system

In the last few years, social media systems have experienced a fast growth. The amount of content shared in these systems increases fast, leading users to face the well known “interaction overload” problem, i.e., they are overwhelmed by content, so it becomes difficult to come across interesting items. To overcome this problem, social recommender systems have been recently designed and developed in order to filter content and recommend to users only interesting items. This type of filtering is usually affected by the “over-specialization” problem, which is related to recommendations that are too similar to the items already considered by the users. This paper proposes a friend recommender system that operates in the social bookmarking application domain and is based on behavioral data mining, i.e., on the exploitation of the users activity in a social bookmarking system. Experimental results show how this type of mining is able to produce accurate friend recommendations, allowing users to get to know bookmarked resources that are both novel and serendipitous. Using this approach, the impact of the “interaction overload” and the “over-specialization” problems is strongly reduced.

Using social media to characterize urban mobility patterns: State-of-the-art survey and case-study

Abstract The knowledge of the urban mobility is a crucial aspect for city planners and administrators. The huge amount of geo-spatial data, generated by the combination of social media systems and the wide use of smart devices, is creating new challenges and opportunities to satisfy this thirst of knowledge. In this work, we explore how social media data can be used to infer knowledge about urban dynamics and mobility patterns in a urban area. Specifically, in order to highlight the main advantages, limitations, and open issues, we focus on mobility patterns by presenting a survey of the state of the art and a case-study based on the city of Barcelona.

Using Behavioral Data Mining to Produce Friend Recommendations in a Social Bookmarking System

Social recommender systems have been developed to filter the large amounts of data generated by social media systems. A type of social media, known as social bookmarking system, allows the users to tag bookmarks of interest and to share them. Although the popularity of these systems is increasing and even if users are allowed to connect both by following other users or by adding them as friends, no friend recommender system has been proposed in the literature. Behavioral data mining is a useful tool to extract information by analyzing the behavior of the users in a system. In this paper we first perform a preliminary analysis that shows that behavioral data mining is effective to discover how similar the preferences of two users are. Then, we exploit the analysis of the user behavior to produce friend recommendations, by analyzing the resources tagged by a user and the frequency of each used tag. Experimental results highlight that, by analyzing both the tagging and bookmarking behaviors of a user, our approach is able to mine preferences in a more accurate way with respect to a state-of-the-art approach that considers only the tags.

Lipid profiles in brains from sheep with natural scrapie.

Prion diseases are fatal neurodegenerative disorders affecting many mammals, ovine scrapie being the archetypal prion disease. Several independent studies in murine and cell-based models of scrapie have highlighted the presence of a link between prion generation and lipid alterations; yet, no data on natural disease are available. In this study we investigated levels of total lipids and cholesterol as well as profiles of fatty acids in brain homogenates from symptomatic and asymptomatic scrapie-infected sheep vs. healthy sheep, all belonging to the same flock. Lipid extracts were analyzed by means of gas chromatography and high performance liquid chromatography. Data of fatty acids were submitted to multivariate statistical analysis to give a picture of the brain lipid profiles of sheep. Interestingly, results revealed abnormalities in the brain fatty acid unsaturation of infected/symptomatic animals. Significant reduction of monoene 18:1 n-9 was detected in brain lipids from infected/symptomatic sheep, as compared to healthy and infected/asymptomatic animals, and this alteration occurred in combination with a significant increase in 18:0 level. The unsupervised Principal Component Analysis showed that infected/symptomatic and healthy sheep samples lie in two different regions of the plot, infected/asymptomatic lie mostly next to healthy. The increase of cerebral saturated fatty acids provides a rough indication of presumed alterations in lipid raft domains of nervous cells during scrapie, suggesting that they may exist in a notable viscous liquid-ordered state. Such physicochemical alteration would have a profound impact on the raft thermodynamic properties, its spatial organization, and signal transduction, all potentially relevant for prion generation.

Friend Recommendation in a Social Bookmarking System: Design and Architecture Guidelines

Social media systems allow users to share resources with the people connected to them. In order to handle the exponential growth of the content in these systems and of the amount of users that populate them, recommender systems have been introduced. As social media systems with different purposes arose, also different types of social recommender systems were developed in order to filter the specific information that each domain handles. A form of social media, known as social bookmarking system, allows to share bookmarks in a social network. A user adds as a friend or follows another user and receives updates on the bookmarks added by that user. In this paper, we present an analysis of the state-of-the-art on user recommendation in social environments and of the structure of a social bookmarking system, in order to derive design guidelines and an architecture of a friend recommender system in the social bookmarking domain. This study can be useful for future research, by highlighting the aspects that characterize this domain and the features that this type of recommender system has to offer.

Using new communication technologies and social media interaction to improve the motivation of users to exercise

The number of communication technologies and devices from which users can access information has rapidly increased. Moreover, users now have the chance to interact through social media channels, in order to share what they like and what they are experiencing in their everyday life. Both these aspects influence the design and development of Human-Computer Interaction applications that aim at motivating users to exercise more. In fact, the possibility to manage the exercising activity from different types of devices and the possibility to interact with other users, can increase the motivation of a user to exercise more. This paper presents a persuasive web application for sport and health, designed to motivate people in their exercising activity. The innovative aspect of our application is the possibility to use on a web browser some features previously available only through a mobile application. Moreover, it allows a richer interaction with the Facebook social network. Preliminary results show how different types of devices and new communication networks can be integrated, in order to improve the user experience and motivate users to a more active lifestyle.

The role of social interaction on users motivation to exercise: A persuasive web framework to enhance the self-management of a healthy lifestyle

Abstract The current research guidelines of the European community suggest the importance of the development of systems that help users manage their health themselves. The increasing amount of communication technologies and devices from which users can access information, and the possibility to interact through social media channels, play an important role in this scenario. Based on these considerations, in this paper we present an innovative persuasive web application, designed both to exploit social networking sites and to cooperate with a mobile application that already operates in the e-health and motivational domains. In particular, the innovative aspects introduced by the web application are the possibility to access also from a web browser some features previously available only through a mobile application and a more direct and user-friendly integration of social network sites. Indeed, thanks to an extensive interaction with the Facebook social network, users are allowed to share their experience with the application. This generates a strong social influence effect, which inspires and motivates other users to improve their exercising activity. Experimental results put in evidence that our web application, also thanks to social interactions, is favoring an enhancement of users’ motivation to a more active lifestyle. This is mainly due to its capability to have an impact on the other users thanks to the posts generated on the Facebook social network.