Hamman W. Samuel

Fastest association rule mining algorithm predictor (FARM-AP)

Association rule mining is a particularly well studied field in data mining given its importance as a building block in many data analytics tasks. Many studies have focused on efficiency because the data to be mined is typically very large. However, while there are many approaches in literature, each approach claims to be the fastest for some given dataset. In other words, there is no clear winner. On the other hand, there is panoply of algorithms and implementations specifically designed for parallel computing. These solutions are typically implementations of sequential algorithms in a multi-processor configuration focusing on load balancing and data partitioning, each processor running the same implementation on it is own partition. The question we ask in this paper is whether there is a means to select the appropriate frequent itemset mining algorithm given a dataset and if each processor in a parallel implementation could select its own algorithm provided a given partition of the data.

PSST… privacy, safety, security, and trust in health information websites

Various newsworthy incidents typically include breaches of security, invasion of privacy, and harm caused by false information. In the e-health domain, there has been a lot of focus on ethical issues when dealing with electronic health records (EHRs) and patient medical records (PMRs). However, equally important are the myriad of health information websites that are being used to formally or informally get medical advice online. This study surveys related work on three popular and pertinent issues in health information websites: privacy, security, and trust. Our contributions include a succinct survey of different categories of popular health information websites (WebMD.com, MayoClinic.com, KidsHealth.org, PatientsLikeMe. com) to gauge existing methods for handling these issues. Moreover, an agenda is proposed for understanding the three issues orthogonally via access control. Other outcomes of the study include recommendations for open problems identified in health websites, including the need for fine-grained privacy, security and trust controls.

Findability in health information websites

In this study, we investigate how health information consumers locate content on health information websites. Preliminary results show that there is room for improvement in terms of finding specific content on health websites, that is, findability. We focus on and identify usability issues with three key aspects of health websites: search box, navigation menu, and home page. Results are based on a population sample of users with varied backgrounds, familiarity with medical terms, and a diversified range of question types. Consumer trends in looking up information demonstrate that using the search box is the method of choice, while navigation menus and links on the home page are not effectively being utilized. Ultimately, we propose possible solutions aimed at improving the overall quality of health information websites, such as faceted search, metaphor exploration, multi-dimensional views, and trending topics.

MedFact: Towards Improving Veracity of Medical Information in Social Media Using Applied Machine Learning.

Since the advent of Web 2.0 and social media, anyone with an Internet connection can create content online, even if it is uncertain or fake information, which has attracted significant attention recently. In this study, we address the challenge of uncertain online health information by automating systematic approaches borrowed from evidence-based medicine. Our proposed algorithm, MedFact, enables recommendation of trusted medical information within health-related social media discussions and empowers online users to make informed decisions about the credibility of online health information. MedFact automatically extracts relevant keywords from online discussions and queries trusted medical literature with the aim of embedding related factual information into the discussion. Our retrieval model takes into account layperson terminology and hierarchy of evidence. Consequently, MedFact is a departure from current consensus-based approaches for determining credibility using “wisdom of the crowd”, binary “Like” votes and ratings, popular in social media. Moving away from subjective metrics, MedFact introduces objective metrics. We also present preliminary work towards a granular veracity score by using supervised machine learning to compare statements within uncertain social media text and trusted medical text. We evaluate our proposed algorithm on various data sets from existing health social media involving both patient and medic discussions, with promising results and suggestions for ongoing improvements and future research.

Iron Mask: Trust-Preserving Anonymity on the Face of Stigmatization in Social Networking Sites

Social networking sites are pervasively being used for seeking advice, asking questions, giving answers, and sharing experiences on various topics including health. When users share content about sensitive health topics, such as sexual dysfunction, infertility, or STDs, they may wish to do so anonymously to avoid stigmatization and the associated negative effects on mental health. However, a user masking their name with a pseudonym may still be inadvertently exposing their identity because of various quasi-identifiers present in their profile. One such quasi-identifier that has not been investigated in literature is the content itself, which could be used for authorship identification. Moreover, an anonymous user’s credibility cannot be established because their profile is no longer linked with their reputation. This study proposes the Iron Mask algorithm for providing enhanced anonymity while preserving trust. Iron Mask improves anonymity by using a probabilistic machine learning approach based on whiteprint identification and inclusion of content as a quasi-identifier. Iron Mask also introduces the concept of a trust-preserving pseudonym which masks user identity without loss of credibility. We evaluate the proposed algorithm using datasets from Quora, a question-answering social networking site, and demonstrate the efficacy of our algorithm with satisfactory recall and survey feedback results.

Community question retrieval in health forums

Community Question Answering (CQA) has emerged as a popular type of service enabling users to ask and answer questions, and access the existing knowledgebase. CQA archives contain a lot of useful user-generated content and have been recognized as important information resources for the web. To improve accessibility to this body of knowledge in CQA archives, effective and efficient question retrieval is required. Question retrieval in a CQA archive aims to identify and retrieve existing questions that are relevant to new user questions. The objective of this study is to develop a question retrieval system that can sift through such forums and identify existing questions which are most similar to the user-provided question. We focus on health forums, and propose a CQA system using weighted TF-IDF, relevance heuristics, and term expansion. We compare our proposed algorithm against other well-known methods, and demonstrate that our method outperforms the Latent Dirichlet allocation (LDA) topic model, Latent Semantic Indexing (LSI), language modelbased information retrieval, BM25, vector space, Word2Vec, and semantic similarity approaches. Our initial experiments use datasets from the IEEE Healthcare Data Analytics Challenge 2015, and we also present our efforts towards development of a Bronze Standard for question similarity evaluation using self-annotations and annotations provided by affiliates of Mayo Clinic.

BubbleNet: An innovative exploratory search and summarization interface with applicability in health social media

We analyse the application of various interfaces to facilitate exploratory search and summarization of documents, especially BubbleNet, an innovative interface for summarizing corpus that also allows discovery of new knowledge that the user may not have previously been looking for. BubbleNet is a visual force-directed graph that displays an interactive and dynamic network of topics, semantic relationships, and related documents based on a corpus. Our experimental results show that BubbleNet gives a better user experience and faster performance in comparison with other exploratory search and summarization interfaces such as query-based search, word clouds, hierarchical directories, and topic graphs. We also explore the applicability of BubbleNet to Cardea, a health portal under development for patients and medics.

Golden Retriever: Question Retrieval System

Duplicate questions get posted on Q&A online forums because users may not be aware of similar questions. Our proposed system, Golden Retriever, can recommend existing questions that are semantically related to incoming questions. Compared with other existing techniques such as Latent Semantic Indexing, Language Model and Semantic Similarity, our approach shows good results for the ICHI Healthcare Data Analytics Challenge dataset using normalized TF-IDF, relevance heuristics, and semantic relatedness.

A Repository of Codes of Ethics and Technical Standards in Health Informatics

We present a searchable repository of codes of ethics and standards in health informatics. It is built using state-of-the-art search algorithms and technologies. The repository will be potentially beneficial for public health practitioners, researchers, and software developers in finding and comparing ethics topics of interest. Public health clinics, clinicians, and researchers can use the repository platform as a one-stop reference for various ethics codes and standards. In addition, the repository interface is built for easy navigation, fast search, and side-by-side comparative reading of documents. Our selection criteria for codes and standards are two-fold; firstly, to maintain intellectual property rights, we index only codes and standards freely available on the internet. Secondly, major international, regional, and national health informatics bodies across the globe are surveyed with the aim of understanding the landscape in this domain. We also look at prevalent technical standards in health informatics from major bodies such as the International Standards Organization (ISO) and the U. S. Food and Drug Administration (FDA). Our repository contains codes of ethics from the International Medical Informatics Association (IMIA), the iHealth Coalition (iHC), the American Health Information Management Association (AHIMA), the Australasian College of Health Informatics (ACHI), the British Computer Society (BCS), and the UK Council for Health Informatics Professions (UKCHIP), with room for adding more in the future. Our major contribution is enhancing the findability of codes and standards related to health informatics ethics by compilation and unified access through the health informatics ethics repository.

On Management of the Health Content Lifecycle

The Internet is an ideal tool for promoting public health goals of prolonging life, health, and improving the quality of life. There are many websites with health-related information where one can go to as an information source, for health advice, or selfdiagnosis. However, these health websites require a more acute awareness of ethical issues due to potential life threatening risks from misuse of information. Providing disclaimers and accreditation logos only goes so far in covering potential legal conflicts, but fulfilling ethical obligations for non-maleficence requires more action on our part. As such, the content lifecycle of these websites requires greater emphasis on privacy, security, and trustworthiness. We propose and give a high-level description of a Health Content Management System (HCMS) that addresses both the managerial, as well as the ethical issues with health content. Surveys of existing health websites and content management systems demonstrate the need for the proposed system. Moreover, the novelty of the proposed HCMS is appraised and asserted in comparison with similar health framework concepts. Our contributions include survey results of more than 50 health websites, taxonomy of health websites’ characteristics, discussion about legal versus ethical obligations, and a blueprint for typical and novel features for health websites. Moreover, this study presents a new approach to analysing health content via lifecycles. Keywords-Ethics; Trust; Medical; Websites; CMS; CMF; Review; Disclaimer; Liability