Appealing but Potentially Biasing - Investigation of the Visual Representation of Segmentation Predictions by AI Recommender Systems for Medical Decision Making

In this paper we provide a method that allows the visualization of similarity relationships present between items of collaborative filtering recommender systems, as well as the relative importance of each of these. The objective is to offer visual representations of the recommender system’s set of items and of their relationships; these graphs show us where the most representative information can be found and which items are rated in a more similar way by the recommender system’s community of users. The visual representations achieved take the shape of phylogenetic trees, displaying the numerical similarity and the reliability between each pair of items considered to be similar. As a case study we provide the results obtained using the public database Movielens 1M, which contains 3900 movies.

M3KGR: A momentum contrastive multi-modal knowledge graph learning framework for recommendation

Multi-armed Bandits (MAB), which is the short form for multi- armed bandit, is playing a significant role in varied spaces including online advertising, recommendation system, clinical trials and medical decision- making and even in 5G. The model aims to balance “choosing the option with the highest reward known so far” and “exploring the options unknown”. At last, it will reflect the results by a coefficient called regret. It is sometimes not suitable to some special situations. This article shows three of its classical algorithms: Explore Then Commit (ETC), Upper Confidence Bound (UCB) and Thompson Sampling (TS). ETC separates exploitation and commitments into two single parts but it will produce more regrets comparing to the other two algorithms. UCB selects arms based on the “average reward + upper confidence bound” metrics. It is one of the most popular algorithms. TS models uncertainty by leveraging Bayesian updates and random sampling. However, it is much harder than the others. They have different advantages and disadvantages. In summary, MAB has good application scenarios if people make good use of its different types of algorithms.

People use recommender systems to improve their decisions; for example, item recommender systems help them find films to watch or books to buy. Despite the ubiquity of item recommender systems, they can be improved by giving users greater transparency and control. This article develops and assesses interactive strategies for transparency and control, as applied to event sequence recommender systems, which provide guidance in critical life choices such as medical treatments, careers decisions, and educational course selections. This article’s main contribution is the use of both record attributes and temporal event information as features to identify similar records and provide appropriate recommendations. While traditional item recommendations are based on choices by people with similar attributes, such as those who looked at this product or watched this movie, our event sequence recommendation approach allows users to select records that share similar attribute values and start with a similar event sequence. Then users see how different choices of actions and the orders and times between them might lead to users’ desired outcomes. This paper applies a visual analytics approach to present and explain recommendations of event sequences. It presents a workflow for event sequence recommendation that is implemented in EventAction and reports on three case studies in two domains to illustrate the use of generating event sequence recommendations based on personal histories. It also offers design guidelines for the construction of user interfaces for event sequence recommendation and discusses ethical issues in dealing with personal histories. A demo video of EventAction is available at https://hcil.umd.edu/eventaction.

Modern medical decision making systems require users to manually collect and process information from distributed and heterogeneous repositories to facilitate the decision making process. There are many factors (such as time, volume of information and technical ability) that can potentially compromise the quality of decisions made for patients. In this work we demonstrate and evaluate a new medical decision making support system, called OMeD, which automatically answers medical queries in real time, by collecting and processing medical information. OMeD utilizes a natural-language-like user interface (for querying) and semantic web techniques (for knowledge representation and reasoning) to answer queries. We compare OMeD to a set of standard machine learning techniques across a series of benchmarks based on simulated patient data. The conventional techniques attempt to learn the answer to a query by analyzing simulated patient records. The sparsity of the simulated data leads conventional techniques to frequently misidentify the relationships between medical concepts. In contrast, OMeD is able to reliably provide correct answers to queries. Unlike conventional automated decision support systems, OMeD also generates independently verifiable proofs for its answers, providing healthcare workers with confidence in the system's recommendations.

The introduction of recommendation systems for the health sector has opened additional channels to improve the decision-making of physicians and administrators of the healthcare system. Web-based and mobile applications, social networks and virtual communities could improve these processes. The Internet provides a huge amount of information; therefore, the creation of intelligent mechanisms and tools, such as the ones proposed, allows physicians and the health sector to be better informed in the decision-making process while improving communication channels. In this work, a theoretical framework for the design and implementation of the so-called project MedicGate, a smart social network to optimize medical decisions using cognitive computing is presented. The expectation is that this initiative will have an impact on those responsible for the generation of health policies, by providing information and resources for decision-making through informed knowledge and feedback from the medical and academic sector.

In this study, the authors propose a generic architecture, associated terminology and a classificatory model for observing ICU patient’s health condition with a Content-Based Recommender (CBR) system consisting of K-Nearest Neighbors (KNN) and Association Rule Mining (ARM). The aim of this research is to predict or classify the critically conditioned ICU patients for taking immediate actions to reduce the mortality rate. Predicting the health of the patients with automatic deployment of the models is the key concept of this research. IBM Cloud is used as Platform as a Service (PaaS) to store and maintain the hospital data. The proposed model demonstrates an accuracy of 95.6% from the KNN Basic ‘\( ball\_tree \)’ algorithm. Also, real-time testing of the deployed model showed an accuracy of 87% while comparing the output with the actual condition of the patient. Combining the IBM Cloud with the Recommender System and early prediction of the health, this proposed research can provide a complete medical decision for the doctors.

Visual information is an important factor in recommender systems. Some studies have been done to model user preferences for visual recommendation. Usually, an item consists of two fundamental components: style and category. Conventional methods model items in a common visual feature space. In these methods, visual representations always can only capture the categorical information but fail in capturing the styles of items. Style information indicates the preferences of users and has significant effect in visual recommendation. Accordingly, we propose a DeepStyle method for learning style features of items and sensing preferences of users. Experiments conducted on two real-world datasets illustrate the effectiveness of DeepStyle for visual recommendation.

The appropriate tagging of questions on Q&A networks like StackExchange is important for organizing the content of these platforms. Accordingly, recommender systems can help the users by suggesting relevant tags for their questions. The data used as input to these recommender systems can be manifold like headlines of the questions or their entire content. In this work, we explore whether mathematical formulas alone are sufficient for predicting tags of mathematical questions. To achieve this goal, we employ and compare well established methods on both textual and visual representations of formulas. To the best of our knowledge, we are the first to use this kind of limited content as input. Our results show that formulas are helpful for finding proper tags while visual representations outperform textual ones.

Even though mobile tourism analysis is developing, there is still a lack of literature regarding its formation patterns which makes the picture of its evolution incomplete. The evaluation in this study was carried out in general by emphasizing the evolution of exploratory analysis on the theme of "mobile tourism". This article aims to determine the increasing development of mobile tourism in terms of the number of publications, journals, authors, and the most relevant citations. Then analyze the itinerant tourism distribution map based on grouping, with merging, joint citation, collaboration network analysis, and joint word assessment, and interpret the conceptual structure of the country's co-occurrence and collaboration network regarding its visual representation. The method used is a descriptive approach through bibliometric analysis. An in-depth search of the Scopus database and RStudio software was conducted, screening, analyzing, and interpreting 752 relevant articles from 2014 to 2023. The findings show that an increase in the development of mobile tourism is identified from an increase in the number of publications, the most relevant journals, and authors, as well as the increasingly diverse origins of the authors even though the number of annual average citations has decreased slightly. Recommended six themes in mobile tourism studies, namely Tourism, Technology Acceptance Models (TAM), Mobile Applications, Recommendation Systems, Loyalty, and Cultural Tourism, and produced Technology Acceptance Models and Tourism as important studies with great prospects in the future.

Recommender systems (RS) are an important filtering tool for discovering services in a personalized way to guide users from the large space of possible options. Over the past decades, recommendation systems in healthcare application have become popular due to the exponential increase in health data available on various platforms. The chapter begins with an overview of recommender systems, discussing the purpose, functionality, and key components. It highlights the significance of recommender systems in healthcare, where the abundance of data and the complexity of medical decisions necessitate personalized recommendations to enhance the quality of care. Additionally, it introduces the concept of big data and its relevance in healthcare recommender systems, emphasizing the wealth of information available from various sources such as electronic health records, wearable devices, and social media. Moreover, the chapter addresses the challenges associated with the implementation of recommender systems in the healthcare domain.

Sentiment analysis has attracted increasing attention recently due to its potential wide applications in opinion analysis, recommendation system, etc. Visual-textual sentiment analysis aims to improve the performance of sentiment analysis by leveraging both visual and textual signals. In this paper, we address the visual-textual sentiment analysis in product reviews. Our main contributions are two-fold. First, instead of crawling data from Flickr or Twitter with positive and negative labels in existing works, we introduce a new dataset for visual-textual sentiment analysis, termed as Product Reviews150K (PR-150K), which is collected from the product reviews of online shopping websites. Second, we propose a deep Tucker fusion method for visual-textual sentiment analysis, which efficiently combines visual and textual deep representations based on the Tucker decomposition and a bilinear pooling operation. Extensive experiments on our PR-150K, MVSO, and VSO datasets show that our method outperforms several state-of-the-art methods.

This article describes a personal music recommender system designed to overcome the ‘cold start’ problem that affects the visibility of recent works of composition and improvisation outside popular music. The approach builds upon our experimental results that showed increased levels of engagement by subjects with unfamiliar music through a focus on acoustic properties and emotional response. Set in the context of a digital music library, the developed recommender component provides two visual acoustic representations of the piece, and additionally allows users to record their own continuous affect responses. These interface elements are designed to invoke the behaviours observed in our experiments, meaning visitors stay engaged with the online resource for longer and so experience a broader range of music.

Multimodal Knowledge Graphs (MKGs), which organize visual-text factual knowledge, have recently been successfully applied to tasks such as information retrieval, question answering, and recommendation system. Since most MKGs are far from complete, extensive knowledge graph completion studies have been proposed focusing on the multimodal entity, relation extraction and link prediction. However, different tasks and modalities require changes to the model architecture, and not all images/objects are relevant to text input, which hinders the applicability to diverse real-world scenarios. In this paper, we propose a hybrid transformer with multi-level fusion to address those issues. Specifically, we leverage a hybrid transformer architecture with unified input-output for diverse multimodal knowledge graph completion tasks. Moreover, we propose multi-level fusion, which integrates visual and text representation via coarse-grained prefix-guided interaction and fine-grained correlation-aware fusion modules. We conduct extensive experiments to validate that our MKGformer can obtain SOTA performance on four datasets of multimodal link prediction, multimodal RE, and multimodal NER. Code is available in https://github.com/zjunlp/MKGformer.

Visual content-enhanced sequential recommendation with feature-level attention