Buket Kaya

Unsupervised link prediction in evolving abnormal medical parameter networks

The saying “treat the disease, not the symptoms” is widespread, a cliche for eliminating or repairing the root of a problem rather than mitigating the negative effects. It is taken for granted that prevention is the best course of action. It is ironic, then, that many of today’s best “disease treatments” are actually symptom suppressors. The prediction of abnormal medical parameters based on the past patient medical history revealed efficacious in foreseeing medical signs a patient could likely be affected in the future. In this paper, we predict the onset of future signs on the base of the current health status of patients. For this purpose, we first construct a weighted abnormal medical parameter network considering the relations between abnormal parameters. Then, we propose an unsupervised link prediction method to identify the connections between abnormal parameters, building the evolving structure of abnormal parameter network with respect to patients’ ages. To the best of our knowledge, this is the first attempt in predicting the connections between the results of laboratory tests. Experiments on a real network demonstrate that the proposed approach can reveal new abnormal parameter correlations accurately and perform well at capturing future disease signs.

A recommendation method based on link prediction in drug-disease bipartite network

Link prediction is one of the most important research topics in social network analysis. It estimates of possible future connections between nodes in the network taking advantage of networks current state. The link prediction also provides useful information to make comments about the future. In this study, a method for link prediction in the disease-drug network is proposed. Sofar, the most of studies done is usually based on connection prediction in single mode networks. This method has been applied on a bipartite such as disease-drug network, as apart from single mode networks. The results obtained from experiments by unsupervised prediction demonstrate that the proposed method has a good percentage of success.

A novel multi-objective genetic algorithm for multiple sequence alignment

There is no well-accepted theoretical model for multiple sequence alignment. An algorithm is accepted as a good method for multiple sequence alignment if it produces better fitness scores with respect to the benchmark datasets. For this purpose, we propose an efficient method using multi-objective genetic algorithm to discover optimal alignments in multiple sequence data. The main advantage of our approach is that a large number of trade-off alignments can be obtained by a single run with respect to conflicting objectives: alignment length minimisation and similarity and support maximisation. We compare our method with the four well-known multiple sequence alignment methods, MUSCLE, ClustalW, SAGA and MSA-GA. The first two of them are progressive methods, and the other two are based on evolutionary algorithms. Experimental results on the BAliBASE 2.0 database demonstrate that our method found better solutions than the others for most of the cases in terms of accuracy.

Prediction of Symptom-Disease Links in Online Health Forums

Social networks are structures that are used to model complex networks in different environments. The desire of reaching useful information by studying complex networks has made social network analysis an important research topic today. One of the most interesting subjects of social network analysis is link prediction. It estimates potential future connections by using the current state of the network. In this study, link prediction is made in bipartite graphs which is one of the social network structures. For this study, the network is first constructed with question and advise (generally disease) made on online health forum sites. Online health forum is a place where user can get his/her medical questions answered by real health professionals and other forum members. Then, symptoms of diseases are obtained by analyzing questions on online forum sites. Thus, a bipartite network consisting of questions and diseases corresponding to the obtained symptom data is constructed. In this network, link prediction has been made with internal links method. The results of the method have been compared with four of the other link prediction methods and it was found that this method has better performance than the other methods.

A social network based optimal tariff recommender for a group of GSM users

GSM is a mobile technology that allows people to communicate with one another. The technology enables people to call others over the phone with a GSM number and a certain tariff for communication. An interpersonal communication network is established by means of calls and text messages. The purpose of the present study is to develop a system that recommending the most appropriate tariff to each GSM user through social network modelling and thus to consider the total utility of the participants in the communication network. To the best of our knowledge, this is the first study using social network modelling to recommend the most appropriate tariff to a specific group of GSM users. The results of the experiment on a synthetic communication network comprised of users with various tariffs suggest that the method is practical and able to yield accurate results.

Evaluating reliability of question-disease relations in online health forms: A link prediction approach

Abstract The Internet has become an indispensable part of human life in today. People can now easily find answers to questions they are curious about via the internet. The short, effortless and free way that the Internet provides is extremely attractive for people to have an idea in subjects they wonder related to their health. There are many online health forums where people can ask questions answered by health professionals. Every day, people ask thousands of questions on these sites and get answers about which diseases their complaints may be related to. The frequent use of online forum sites by people has led to the selection of these forums as data source for this study, and analysis of reliability. Firstly, in this study, link prediction in bipartite social networks, where intensive works have been done and it is applied on many areas nowadays, is tried to be carried out on question-disease bipartite network constructed with data obtained from analysis of online health forums whose use rate increase substantially. For this purpose, a novel link prediction method called as intensive link prediction is proposed, and prediction success of this method is compared with five of similarity-based link prediction methods. Better results have been obtained with the proposed method than the other methods. Then, the accuracy of the answers given to the users on online health forums which received intense interest are tested. The reliability of online health forums is measured by the accuracy analysis performed.

Extracting Relations Between Symptoms by Age-Frame Based Link Prediction

The saying “treat the disease, not the symptoms” is widespread, a cliche for eliminating or repairing the root of a problem rather than mitigating the negative effects. However, the effort to prevent the negative effects which may be reason of a disease is the best course of action. The prediction of symptoms based on the past patient medical history revealed efficacious in foreseeing symptoms (abnormal parameters) a patient could likely be affected in the future. In this paper, we predict the onset of future symptoms on the base of the current health status of patients. For this purpose, we first construct a weighted symptom network considering the relations between abnormal parameters. Then, we propose an unsupervised link prediction method to identify the connections between parameters, building the evolving structure of symptom network with respect to patients’ ages. Experiments on a real network demonstrate that the proposed approach can reveal new abnormal parameter correlations accurately and perform well at capturing future disease risks.

A link prediction approach for drug recommendation in disease-drug bipartite network

Social networks we have encountered in different areas and in different forms have a dynamic structure because the relationships they define constantly change. Link prediction is an important and effective solution to understand this dynamic nature of networks and to identify future relations. It estimates of possible future connections between nodes in the network taking advantage of networks current state. In this study, a method for link prediction in the disease-drug network is proposed. Sofar, the most of studies done is usually based on connection prediction in single mode networks. This method has been applied on a bipartite such as disease-drug network, as apart from single mode networks. To compare performance of the proposed method, four of similarity based link prediction methods has been also applied to the network. The results obtained from experiments show that the proposed method has a good percentage of success than the other similarity based link prediction methods.

An intelligent method for optimization of tariffs in GSM networks

GSM is a mobile technology that allows people to communicate with one another. The technology enables people to call others over the phone with a GSM number and a certain tariff for communication. An interpersonal GSM network is established by means of calls and text messages. This paper proposes an approach to recommend optimal tariffs to GSM users to maximize the total utility of individuals in the GSM network. However, finding optimal tariffs for very large GSM networks is computationally intractable by standard methods. For this purpose, we present a novel multi-agent based algorithm to recommend the most appropriate tariffs to a specific group of GSM users. The results of the experiment on a synthetic GSM network comprised of users with various tariffs suggest that the method is practical and able to yield accurate results.