At What Price? Exploring the Potential and Challenges of Differentially Private Machine Learning for Healthcare

2023 | HICSS | Citations: 1

Authors: Aslan, Aycan; Matschak, Tizian; Greve, Maike; Trang, Simon; Kolbe, Lutz

Abstract: The increased generation of data has become one of the main drivers of technolog ... Expand

Abstract: The increased generation of data has become one of the main drivers of technological innovation in healthcare. This applies in particular to the adoption of Machine Learning models that are used to generate value from the growing available healthcare data. However, the increased processing of sensitive healthcare data comes with challenges in terms of data privacy. Differential privacy, the method of adding randomness to the data to increase privacy, has gained popularity in the last few years as a possible solution. However, while the addition of randomness increases privacy, it also reduces overall model performance, generating a privacy-utility trade-off. Examining this trade-off, we contribute to the literature by providing an empirical paper that experimentally evaluates two prominent and innovative methods of differentially private Machine Learning on medical image and text data to deepen the understanding of the existing potential and challenges of such methods for the healthcare domain. Collapse

Topics: privacy differential privacy healthcare data knowledge base data protection

Methods: experiment computational algorithm machine learning blood pressure PyTorch

Seeking Information Using Search Engines: The Impact of Negation on Judgments

2023 | Communications of the Association for Information Systems | Citations: 0

Authors: Kayhan, Varol; Warkentin, Merrill

Abstract: We examine how online searches lead to judgment formation in two contexts where ... Expand

Abstract: We examine how online searches lead to judgment formation in two contexts where statements are negated differently. In one context, a statement (of the form "if P then Q") is negated using "not" (as in "if P then not Q"), while in the other context, it is negated using a different term or phrase (as in "if P then R"). We show that online searches to find information about the validity of this statement generate different results in these contexts. Our findings contribute to extant work on online searches by suggesting that when individuals test the validity of a statement using search engines, their searches lead to categorically different results contingent on the search context. From a practical standpoint, we make recommendations to search engine developers and content providers so online searches generate more inclusive results. Collapse

Topics: search engine Google+ web page decision making accounting

Methods: experiment blood pressure experimental task statistical hypothesis test analysis of covariance

Does Culture Matter for the Design of Chatbots Promoting Blood Donation Behaviour? - The Difference in Perception of Culture-Tailored Conversation Styles

2023 | International Conference on Information Systems | Citations: 0

Authors: Müller, Helena; Pietrantoni, Nico; Reuter-Oppermann, Melanie; Greulich, Stefan

Abstract: While blood products are a critical resource in healthcare systems, providing s ... Expand

Abstract: While blood products are a critical resource in healthcare systems, providing sufficient blood products is a worldwide challenge, especially so since the COVID-19 pandemic. As easy and timely access to information is crucial to convince (potential) donors to change their behaviour and become regular donors, chatbots can offer fast and easy access to information whenever (potential) donors need it. Due to their human-like design, chatbots can help motivating and convincing users to donate blood regularly to work against the ongoing, post-pandemic challenges in providing sufficient blood supply. Based on previous findings, we assume that users' perception of a blood donation chatbot can vary worldwide, in relation to the incorporated design features. As part of a design science study, we conducted an online between-subject experiment with participants from USA, Germany, South Africa and India. We could show a significant negative moderating effect of horizontal individualism in terms of the chatbot's individualistic conversation style and the perceived similarity in social group membership, implicating the so-called "contribution conflict" with regard to IS and culture. Collapse

Topics: mobile application usability human computer interaction website design tablet

Methods: chatbot design science experiment cross sectional research between subject experiment

Theories: social identity theory theory of planned behavior cultural dimensions theory

Giving to Get Well: Patients’ Willingness to Manage and Share Health Information on AI-Driven Platforms

2023 | Communications of the Association for Information Systems | Citations: 0

Authors: Raddatz, Nirmalee; Kettinger, William; Coyne, Joshua

Abstract: The digitalization of healthcare makes for the widespread availability of patie ... Expand

Abstract: The digitalization of healthcare makes for the widespread availability of patient-provided data. Artificial Intelligence (AI) relies on this data. In this information-intensive environment, it is imperative to understand the contributing factors of an individual's willingness to manage and share personal health information (PHI). Drawing from the health belief model, we identify the factors that motivate individuals to manage and share their PHI in an AI-driven health platform to obtain its intended benefits. We recognize security risks and present the use of a blockchain database as a representative means of securely managing and controlling an individual's PHI. Data collected from a nationally representative sample of allergy sufferers indicate that the health belief model strongly predicts willingness to share PHI on a personalized AI-supported platform. Our study makes significant contributions by investigating the factors that motivate patients to use an AI-driven health platform to manage their health.Artificial intelligence (AI) advances are taking shape and promising to transform the healthcare industry. According to a study by Accenture, AI in the U.S. healthcare industry is expected to have a compound annual growth rate of 46.1% and reach $95.65 billion by 2028 (Vantage Market Research, 2022). AI can offer a more personalized healthcare experience that optimizes the time it takes to gather information for more efficient decision-making, ultimately allowing better patient care. AI is making its presence known across the medical profession, and medical professionals, researchers, patients, and healthcare providers that do not embrace this technology could be left behind.Ultimately, the success of AI-based platforms for medical purposes depends heavily on individual willingness to use the technology. 1 AI can collect, combine, and analyze large amounts of data from various sources, increasing information-gathering and processing capabilities tremendously (Deane, 2018). This capability offers numerous benefits but also potential risks. For example, according to Rocher, Hendrickx, and de Montjoye (2019), AI can profile, identify, and track individuals by de-anonymizing once anonymized information. Some AI only needs 15 data points about an individual to identify the person. Because of security risks, people may be reluctant to share their personal health information (PHI) with AIbased platforms (Deane, 2018).Achieving the potential benefits of AI from patient platforms will depend on (1) the development of intelligent systems that can collect and analyze vast datasets and (2) the willingness of data owners to share and manage their data with these intelligent systems (Agarwal & Khuntia, 2009). Within healthcare, these conditions are (1) the development of systems that can collect and process PHI and provide or augment diagnostics as/more efficiently and effectively than a traditional doctor-patient profile and (2) patients' ability and willingness to manage and share their own PHI. AI developers work to satisfy the first condition but often take the second condition for granted with the attitude, "If we build it, they will come." However, the first condition is certainly not sufficient to satisfy the second. Without convincing many individuals to share their PHI, an AI-driven health platform cannot serve its purpose. As a result, the question arises as to how to increase individuals' willingness and ability to manage and share their PHI with AI-powered health platforms. Thus, the purpose of our investigation is to discover how to begin to satisfy this second condition by addressing the following questions:Are patients willing to share their PHI with AI-driven health platforms despite security risks? Are patients empowered and ready to manage their PHI, thus, in large part, bringing the control of information to the individual?While numerous studies focus on the design aspects of AI-based predictive models, a significant lack of studies investigate the behavioral aspects pertaining to the usability of these models by patients and clinicians. We adapt Rosenstock's (1974) health belief model (HBM) and expand it to identify the determinants of our test subjects' willingness to share PHI with an AI-driven health platform that uses a blockchain-based database. We apply the HBM to a patient profile in which patients do not solely rely on traditional healthcare provider information channels to get diagnostic and treatment information. Instead, they are empowered to use their PHI to get diagnostic and treatment information from an AI platform. As a result, our application of the HBM moves beyond a patient's willingness to follow a proactive health behavior to considering how patients might proactively use (manage and share) their personal health information to help them receive health-related guidance and insights. This patient information management and information sharing behavior represent a future path of the healthcare industry, and we wish to determine whether this well-regarded health-centered model explains behavior under this new "informationally empowered" patient health profile.Because we care about the practical usability of AI-driven health platforms rather than purely hypothetical usability, we model our research instrument on an actual AI-driven health platform (i.e., doc.ai). 2 This platform uses AI for diagnostic and treatment information and a blockchain-based database to store PHI in 1 We define an AI-based healthcare platform as a software application that (1) collects and stores data for analysis by artificial intelligence (e.g., machine learning) and (2) makes the results of that data analysis available to patients and healthcare professionals in order to guide diagnostic and treatment decisions. We base this definition on an actual AI-based healthcare platform (doc.ai) that we describe in section 2. 2 We provide more specific information regarding this health platform in sections 2 and 3. Collapse

Topics: artificial intelligence self efficacy healthcare data blockchain database system

Methods: survey survey design SEM tool chi squared test machine learning

Theories: health belief model technology acceptance model

A State of the Art Overview on Biosignal-based User-Adaptive Video Conferencing Systems

2023 | International Conference on Business Informatics | Citations: 0

Authors: Seitz, Julia; Krisam, Chiara; Benke, Ivo

Abstract: Video conferencing systems are widely used in times of distributed teams since ... Expand

Abstract: Video conferencing systems are widely used in times of distributed teams since they support flexible work arrangements. However, they have negative impacts on users, such as lacking eye gaze or zoom fatigue. Adaptive interventions in video conferences based on user behavior provide interesting solutions to overcome these challenges, for example, by alerting users when looking tired. Specifically, biosignals measured by sensors like microphones or eye-trackers are a promising basis for adaptive interventions. To provide an overview of current biosignal-based user-adaptive video conferencing systems, we conducted a systematic literature review and identified 24 publications. We summarize existing knowledge in a morphological box and outline further research directions. Thereby, a focus on biooptical signals is visible. Current adaptations target audience feedback, expression understanding and eye gaze mostly by image and representation modifications. In future, we recommend including further biosignals and addressing more diverse problems by investigating adaptation capabilities of further software elements. Collapse

Topics: artificial intelligence database system video conference human computer interaction usability

Methods: biometric measurement structured literature research literature filtering design artifact testing within subject experiment

Healthier together: Social support, self-regulation and goal management for chronic conditions in online health communities

2023 | Information & Management | Citations: 0

Authors: Song, Jiahe; Xu, Pei

Abstract: We investigate the effectiveness of goal management in online health communities ... Expand

Abstract: We investigate the effectiveness of goal management in online health communities (OHCs) in light of social support and self-regulation for patients with chronic diseases. With a five-year longitudinal data set, we found that informational support contributes to both goal progress and self-reflection. Emotional support encourages patients to practice self-reflection if the goals are set for non-mental diseases. We also find that patients’ selfreflection promotes goal progress. Finally, the lack of attention control negatively affects goal progress and self-reflection. Grounded in social cognitive theory, this study is among the first to empirically examine the effect of self-regulation and social support in OHCs. Collapse

Topics: online community website self efficacy affordance electronic health

Methods: qualitative content analysis longitudinal research literature study instrumental variables estimation descriptive statistic

Theories: social cognitive theory

A Human-in-the-Loop Segmented Mixed-Effects Modeling Method for Analyzing Wearables Data

2023 | ACM Transactions on Management Information Systems | Citations: 0

Authors: Srinivasan, Karthik; Currim, Faiz; Ram, Sudha

Abstract: Wearables are an important source of big data, as they provide real-time high-re ... Expand

Abstract: Wearables are an important source of big data, as they provide real-time high-resolution data logs of health indicators of individuals. Higher-order associations between pairs of variables is common in wearables data. Representing higher-order association curves as piecewise linear segments in a regression model makes them more interpretable. However, existing methods for identifying the change points for segmented modeling either overfit or have low external validity for wearables data containing repeated measures. Therefore, we propose a human-in-the-loop method for segmented modeling of higher-order pairwise associations between variables in wearables data. Our method uses the smooth function estimated by a generalized additive mixed model to allow the analyst to annotate change point estimates for a segmented mixed-effects model, and thereafter employs Brent's constrained optimization procedure to fine-tune the manually provided estimates. We validate our method using three real-world wearables datasets. Our method not only outperforms state-of-the-art modeling methods in terms of prediction performance but also provides more interpretable results. Our study contributes to health data science in terms of developing a new method for interpretable modeling of wearables data. Our analysis uncovers interesting insights on higher-order associations for health researchers. Collapse

Topics: smart wearable device mobile system analytical information system human computer interaction human in the loop

Methods: design science wearables based research conceptual modelling hierarchical linear modeling computational algorithm

MDSCAN: AN EXPLAINABLE ARTIFICIAL INTELLIGENCE ARTIFACT FOR MENTAL HEALTH SCREENING

2023 | European Conference On Information Systems | Citations: 0

Authors: Tutun, Salih; Bhattacherjee, Anol; Topuz, Kazim; Tosyali, Ali; Li, Gorden

Abstract: This paper presents a novel artifact called MDscan that can help mental health ... Expand

Abstract: This paper presents a novel artifact called MDscan that can help mental health professionals quickly screen a large number of patients for ten mental disorders. MDscan uses patient responses to the SCL-90-R clinical questionnaire to create a full-color image, similar to radiological images, which identifies which disorder or combination of disorders may afflict a patient, the severity of the disorder, and the underlying logic of this prediction, using an explainable artificial intelligence (XAI) approach. While prior artificial intelligence (AI) tools have seen limited acceptance in clinical practice because of the lack of transparency and interpretability in their "black box" models, the XAI approach used in MDscan is a "white box" model that elaborates which patient feature contributes to the predicted outcome and to what extent. Using patient data from a mental health clinic, we demonstrate that MDscan outperforms current (expert-based) clinical practice by an average of 20%. Collapse

Topics: explainable artificial intelligence shapley additive explanation local interpretable model agnostic explanation IT security artificial intelligence

Methods: computational algorithm machine learning data transformation t-distributed stochastic neighbor embedding survey

Agile CRISP-DM for Analytics Projects: Understanding Diabetes Treatment Using fsQCA

2023 | Americas Conference on Information Systems | Citations: 0

Authors: Weng, Qin

Abstract: Despite the increasing popularity and demand for analytics projects, the uncert ... Expand

Abstract: Despite the increasing popularity and demand for analytics projects, the uncertainty in goals and dynamic processes involved in these projects behoove IS researchers to further understand the project management techniques applicable to these projects. In this paper, I illustrate the process of carrying out an analytics project using an agile project management approach. In this project, following the project management concept of "tailoring", fsQCA is used to analyze the effective treatment configurations of traditional Chinese medicine and Western medicine for diabetes. This illustrated process could be potentially used in various small-sized analytics projects carried out by IS researchers, and R&D efforts. The practical implications of the analysis for medical providers are also discussed. Collapse

Topics: project management healthcare data agile software development

Methods: Cross-Industry Standard Process for Data Mining blood pressure descriptive statistic design methodology machine learning

Narrowing the age-based digital divide: Developing digital capability through social activities

2023 | Information Systems Journal | Citations: 0

Authors: Zhao, Yang; Zhang, Tao; Dasgupta, Rohit K.; Xia, Renpin

Abstract: Healthcare information technologies (HIT) have shown great potential for improvi ... Expand

Abstract: Healthcare information technologies (HIT) have shown great potential for improving the effectiveness and quality of healthcare services. However, the inequal ability of older adults to use HIT may limit their exploitation of these benefits. To narrow the age-based “digital divide”, this research further develops the concept of digital capability and emphasises the link between older adults and their social context. Based on a qualitative inductive study of 33 participants, who included Chinese patients and their family members, we generate a novel theoretical model for understanding the process by which social activities may shape older adults' digital capabilities. Based on the model, we suggest two strategies that might encourage older adults to engage with HIT. This research contributes to the information systems (IS) literature by strengthening digital capability as a conceptual lens to investigate individuals' engagement with information communication technologies (ICTs). It also extends research on the social context for ICT use by revealing how social processes at multiple levels influence digital capability development. Finally, this study offers practical implications for governments and private sectors to encourage and promote ICT use by older adults. Collapse

Topics: information technology capability information system use social network digital divide smartphone

Methods: qualitative interview field study business process modeling theory development blood pressure