Mehedi Masud

An enhanced k -means clustering algorithm for pattern discovery in healthcare data

The huge amounts of data generated by media sensors in health monitoring systems, by medical diagnosis that produce media (audio, video, image, and text) content, and from health service providers are too complex and voluminous to be processed and analyzed by traditional methods. Data mining approaches offer the methodology and technology to transform these heterogeneous data into meaningful information for decision making. This paper studies data mining applications in healthcare. Mainly, we study k-means clustering algorithms on large datasets and present an enhancement to k-means clustering, which requires k or a lesser number of passes to a dataset. The proposed algorithm, which we call G-means, utilizes a greedy approach to produce the preliminary centroids and then takes k or lesser passes over the dataset to adjust these center points. Our experimental results, which were used in an increasing manner on the same dataset, show that G-means outperforms k-means in terms of entropy and F-scores. The experiments also yield better results for G-means in terms of the coefficient of variance and the execution time.

Spectro-temporal directional derivative based automatic speech recognition for a serious game scenario

Speech is one of the important modalities in a serious game platform. Serious game can be very useful for the rehabilitation of individuals with voice disorders. Therefore, we need an efficient and high-performance automatic speech recognition (ASR) system. In this paper, we propose a spectro-temporal directional derivative (STDD) feature that requires less number of computations in the modeling and yet gives high recognition accuracy in the ASR system. The proposed STDD feature is achieved by applying different directional derivative filters in the spectro-temporal domain. The feature dimension is then compressed by discrete cosine transform. The experiments are performed with voice samples of Arabic numerals spoken by persons with and without voice pathology. The experimental results show that the STDD feature outperforms the conventional mel-frequency cepstral coefficients both in clean and noisy environments.

Transaction processing in a peer to peer database network

This paper investigates a transaction processing mechanism in a peer to peer database network. A peer to peer database network is a collection of autonomous data sources, called peers, where each peer augments a conventional database management system with an inter-operability layer (i.e. mappings) for sharing data. In this network, each peer independently manages its database and executes queries as well as updates over the related data in other peers. In this paper, we consider a peer to peer database network where mappings between peers are established through data-level mappings for sharing data and resolving data heterogeneity. With regards to transaction processing in a peer to peer database network, we mainly focus on how to maintain a consistent execution view of concurrent transactions in peers without a global transaction coordinator. Since there is no global transaction coordinator and each peer executes concurrent transactions independently, different peers may produce different execution views for the same set of transactions. For this purpose, we investigate potential problems that arise when maintaining a consistent execution of concurrent transactions. In order to guarantee consistent execution, we introduce a correctness criteria and propose two approaches, namely Merged Transactions and OTM based propagation. We assume that one single peer initiates the concurrent transactions. We also present a solution for ensuring the consistent execution view of concurrent transactions considering the failures of transactions.

Acquaintance based consistency in an instance-mapped P2P data sharing system during transaction processing

The paper presents a transaction processing mechanism in a peer-to-peer (P2P) database environment that combines both P2P and database management systems functionalities. We assume that each peer has an independently created relational database and data heterogeneity between two peers is resolved by data-level mappings. For such an environment, the paper first introduces the execution semantics of a transaction and shows the challenges for concurrent execution of transactions, initiated from a peer, over the network. Later the paper presents a correctness criterion that ensures the correct execution of transactions over the P2P network. We present two approaches ensuring the correctness criterion and finally discuss the implementation issues.

UPDATE PROCESSING IN INSTANCE-MAPPED P2P DATA SHARING SYSTEMS

We consider the problem of update processing in a peer-to-peer (P2P) database network where each peer consists of an independently created relational database. We assume that peers store related data, but data has heterogeneity wrt instances and schemas. The differences in schema and data vocabulary are bridged by value correspondences called mapping tables. Peers build an overlay network called acquaintance network, in which each peer may get acquainted with any other peer that stores related data. In this setting, the updates are free to initiate in any peer and are executed over other peers which are acquainted directly or indirectly with the updates initiator. The execution of an update is achieved by translating, through mapping tables, the update into a set of updates that are executed against the acquainted peers. We consider both the soundness and completeness of update translation. When updates are generated and propagated in the network initiated from a peer, a tree is built dynamically called Update Dependency Tree (UDT). The UDT depicts the relationships among the component updates generated from the initial update. We also discuss the issues of the update propagation when a peer is temporarily unavailable or offline. Our propagation mechanism keeps track of a peer when the peer is not available for a certain period of time and once the peer comes back online the system propagates the updates destined to the returning peer to keep its database synchronized. Moreover, conflict detection and resolution strategies have been proposed for such a dynamic P2P database network. We have implemented and experimentally tested a prototype of our update processing mechanism on a small P2P database network. We show the results of our experiments.

Data damage assessment and recovery algorithm from malicious attacks in healthcare data sharing systems

In a data sharing system in a cloud computing environment, such as health care system, peers or data sources execute transactions on-the-fly in response to user queries without any centralized control. In this case confidential data might be intercepted or read by hackers. We cannot consider any centralized control for securing data since we cannot assume any central third party security infrastructure (e.g., PKI) to protect confidential data in a data sharing system. Securing health information from malicious attacks has become a major concern. However, securing the data from attacks sometimes fail and attackers succeed in inserting malicious data. Hence, this presents a need for fast and efficient damage assessment and recovery algorithms. In this paper, we present an efficient data damage assessment and recovery algorithm to delete malicious transactions and recover affected transactions in a data source in a health care system based on the concept of the matrix. We compare our algorithm with other approaches and show the performance results.

Design and evaluation of hospital-based business intelligence system (HBIS)

This paper describes the development of a hospital-based business intelligent system (HBIS) based on a novel developmental methodology, called the design science research methodology (DSRM), and implemented in a regional general hospital in Taiwan. A design science research methodology is adopted to cover six activities: problem identification and motivation, definition of solution objectives, design and development, demonstration, evaluation, and communication. Based on the DSRM developmental method, HBIS was successfully developed and deployed in the hospital case, and a survey of users shows positive results. In addition, the support and involvement of top management in HBIS development is found to be a critical success factor, and system implementation allowed the hospital to significantly improve performance of managerial indicators for the three abovementioned dimensions. This study contributes a novel developmental methodology from the Information Systems (IS) field as a reference model for future HBIS development, along with the integration of indicators from three major managerial dimensions - NHI, hospital accreditation, and healthcare quality. A hospital-based business intelligent system (HBIS) is presented.Reporting the development and assessment of a HBIS system.Identifying the critical factors to the development of HBIS.Proposing tools that improve the decision-making of health system administrators.

Well-formed semantic model for co-learning

A collaborative learning approach is presented for e-learning systems.A well-formed semantic model for co-learning is proposed.Developed a tool that represents course content graphically with semantic meaning.Demonstrated the feasibility of the proposed approach through experiments. Co-learning of a course by students in an educational institute is becoming a common practice due to the bulk of resources available in the Web, existence of a large number of textbooks, and other offline materials. However, sometimes, students are mystified due the existence of different styles of presentations, definitions, terminologies and examples of a common subject in those sources. This is also true for professors who want to design a course material and teach students in a standard way. Considering the need of well-formed and standard teaching and co-learning materials, in this paper we propose a model that assists professors to design a course. We develop a tool that represents course content graphically with illustrations and semantic meaning. The proposed model is an automated semantic e-learning system based on BNF rules and the OWL ontology language that is capable of representing course contents using ontology. We also demonstrate the feasibility of this model through experiments using the BNF grammar for a programming language as a studying course.

PDST: a peer database simulation tool for data sharing systems

At present there are many simulation tools developed in order to simulate a peer-to-peer (P2P) system. All the tools are dedicated to P2P content distribution systems, simulate network systems for measuring the efficiency of the networks, and file sharing systems. In the last few years, steady progress has been made in research on various issues related to peer database systems. However, there is no software tool for evaluating a peer database system in a large P2P network. In this paper, we present a software tool that can simulate a peer database system in a large P2P network. The tool provides different facilities, for example, generates peers, databases with synthetic data, acquaintances, and mappings between peers. The tool also provides a general framework for executing queries and updates in a peer database system.

Tracking and repairing damaged healthcare databases using the matrix

In a distributed mobile e-health care system, e-health service providers exchange data on the fly in response to user queries without any centralized control. Local databases in e-health service providers might be intercepted during the exchange of data and read by intruders; and malicious transactions may damage data that is highly confidential. In this case any centralized control for securing data cannot be assumed to protect confidential data. Therefore, securing health information from malicious attacks has become a major concern. Although prevention techniques are available, the history of system break-ins guarantees that there is no foolproof technique that totally eliminates security loopholes in a computer system. Hence, efficient damage assessment and recovery techniques are needed. Traditional methods require scanning the entire log from the point of attack to the end which is a slow procedure. In this paper, we present an efficient damage assessment and recovery algorithm to recover the database from malicious transactions. The algorithm is based on data dependency and uses a single matrix. The results of this work prove that our algorithm performs better than the other algorithms in both the damage assessment and the recovery stages.