Data Mining Practical Machine Learning Tools and Techniques

User Centered Design

Although rapid advances in machine learning have made it increasingly applicable to expert decision-making, the delivery of accurate algorithmic predictions alone is insufficient for effective human-AI collaboration. In this work, we investigate the key types of information medical experts desire when they are first introduced to a diagnostic AI assistant. In a qualitative lab study, we interviewed 21 pathologists before, during, and after being presented deep neural network (DNN) predictions for prostate cancer diagnosis, to learn the types of information that they desired about the AI assistant. Our findings reveal that, far beyond understanding the local, case-specific reasoning behind any model decision, clinicians desired upfront information about basic, global properties of the model, such as its known strengths and limitations, its subjective point-of-view, and its overall design objective--what it's designed to be optimized for. Participants compared these information needs to the collaborative mental models they develop of their medical colleagues when seeking a second opinion: the medical perspectives and standards that those colleagues embody, and the compatibility of those perspectives with their own diagnostic patterns. These findings broaden and enrich discussions surrounding AI transparency for collaborative decision-making, providing a richer understanding of what experts find important in their introduction to AI assistants before integrating them into routine practice.

https://dl.acm.org/doi/pdf/10.1145/3359206

"Hello AI": Uncovering the Onboarding Needs of Medical Practitioners for Human-AI Collaborative Decision-Making

Data-driven decision-making consequential to individuals raises important questions of accountability and justice. Indeed, European law provides individuals limited rights to 'meaningful information about the logic' behind significant, autonomous decisions such as loan approvals, insurance quotes, and CV filtering. We undertake three experimental studies examining people's perceptions of justice in algorithmic decision-making under different scenarios and explanation styles. Dimensions of justice previously observed in response to human decision-making appear similarly engaged in response to algorithmic decisions. Qualitative analysis identified several concerns and heuristics involved in justice perceptions including arbitrariness, generalisation, and (in)dignity. Quantitative analysis indicates that explanation styles primarily matter to justice perceptions only when subjects are exposed to multiple different styles --- under repeated exposure of one style, scenario effects obscure any explanation effects. Our results suggests there may be no 'best' approach to explaining algorithmic decisions, and that reflection on their automated nature both implicates and mitigates justice dimensions.

'It's Reducing a Human Being to a Percentage': Perceptions of Justice in Algorithmic Decisions

Designing interfaces or applications that move beyond the bounds of a single device screen enables new ways to engage with digital content. Research addressing the opportunities and challenges of interactions with multiple devices in concert is of continued focus in HCI research. To inform the future research agenda of this field, we contribute an analysis and taxonomy of a corpus of 510 papers in the cross-device computing domain. For both new and experienced researchers in the field we provide: an overview, historic trends and unified terminology of cross-device research; discussion of major and under-explored application areas; mapping of enabling technologies; synthesis of key interaction techniques spanning across multiple devices; and review of common evaluation strategies. We close with a discussion of open issues. Our taxonomy aims to create a unified terminology and common understanding for researchers in order to facilitate and stimulate future cross-device research.

/pdf/cross-device-taxonomy-survey-opportunities-and-challenges-of-2rubom7yex.pdf

Cross-Device Taxonomy: Survey, Opportunities and Challenges of Interactions Spanning Across Multiple Devices

The ever-increasing number of devices quantifying our lives offers a perspective of high awareness of one’s wellbeing, yet it remains a challenge for personal informatics (PI) to effectively support data-based reflection. Effective reflection is recognised as a key factor for PI technologies to foster wellbeing. Here, we investigate whether building tangible representations of health data can offer engaging and reflective experiences. We conducted a between-subjects study where n = 60 participants explored their immediate blood pressure data in relation to medical norms. They either used a standard mobile app, built a data representation from LEGO® bricks based on instructions, or completed a free-form brick build. We found that building with instructions fostered more comparison and using bricks fostered focused attention. The free-form condition required extra time to complete, and lacked usability. Our work shows that designing instructional physicalisation experiences for PI is a means of improving engagement and understanding of personal data.

/pdf/how-instructional-data-physicalisation-fosters-reflection-in-2n58idt4.pdf

How Instructional Data Physicalisation Fosters Reflection in Personal Informatics

The prospect of machine consciousness cultivates controversy across media, academia, and industry. Assessing whether non-experts perceive technologies as conscious, and exploring the consequences of this perception, are yet unaddressed challenges in Human Computer Interaction (HCI). To address them, we surveyed 100 people, exploring their conceptualisations of consciousness and if and how they perceive consciousness in currently available interactive technologies. We show that many people already perceive a degree of consciousness in GPT-3, a voice chat bot, and a robot vacuum cleaner. Within participant responses we identified dynamic tensions between denial and speculation, thinking and feeling, interaction and experience, control and independence, and rigidity and spontaneity. These tensions can inform future research into perceptions of machine consciousness and the challenges it represents for HCI. With both empirical and theoretical contributions, this paper emphasises the importance of HCI in an era of machine consciousness, real, perceived or denied.

/pdf/do-you-mind-user-perceptions-of-machine-consciousness-18lhohao.pdf

Do You Mind? User Perceptions of Machine Consciousness

Playing together is crucial to the unique and invaluable bond between grandparents and grandchildren. However, co-located interactions and play can be limited due to time, distance, or pandemic-related restrictions. To facilitate distributed play, we developed MagiBricks, a system comprised of 3D-printed smart toy bricks and baseplates that provide feedback regarding their placement. The familiarity and appeal of toy bricks to both older adults and children make them ideal for intergenerational play. We conducted a within-subjects study with six grandparent-grandchildren pairs. We compared the interactions and perceived connectedness of the pairs while playing over a distance with either i) MagiBricks or ii) identical regular toy bricks. We found that MagiBricks affected communication dynamics, role taking, nature of play, and perception of connectedness during playtime compared to regular bricks, and were unanimously preferred. We contribute design implications for future systems leveraging (smart) tangibles and fostering intergenerational connectedness.

MagiBricks: Fostering Intergenerational Connectedness in Distributed Play with Smart Toy Bricks

Physical activity is entering the virtual realm. Zwift is an at-home cycling system that is enjoying increasing popularity, yet the specifics of the experience of a virtual cyclist have not been studied to date. Building virtual sports systems can make physical activity accessible to more diverse user groups. To understand how and why users engage in virtual cycling, we conducted n=22 interviews with Zwift users. Through charting the motivations behind using Zwift, we determined that it allowed users to engage in a range of cycling activities traditionally reserved for professional cyclists. Our work reports on key motivations and identifies five key strategies which Zwift uses to create an engaging virtual sports experience. Further, we discuss how Zwift creates a world of virtual professionalism. Our findings offer a structured understanding of the experience of Zwift which can be used to inspire the design of future virtual amateur sports systems.

http://www.alexandria.unisg.ch/268347/1/Zwift-Bentvelzen.pdf

Tailor My Zwift

The world’s population is growing and ageing. Consequently, the demand for health care services increases, which leads to the need for health care providers to find new ways to provide services. Video consultations (VCs) are a cost-efficient way to address this problem. However, they can only work if both health professionals and patients trust and make use of VC systems. We describe our work on determining the reasons behind why the use of VC declined after an initial period of intensive use at a local Dutch medical centre. Questioning patients why they no longer contact their doctors via video was not possible as it would violate privacy law. We propose an alternative method that combines interviews with medical professionals and vignette studies, which can be used to follow up with patients potentially using VCs. We provide a set of lessons learned, which can support future inquiries in the medical domain.

Paweł W. Woźniak

Papers

How Instructional Data Physicalisation Fosters Reflection in Personal Informatics

Do You Mind? User Perceptions of Machine Consciousness

MagiBricks: Fostering Intergenerational Connectedness in Distributed Play with Smart Toy Bricks

Tailor My Zwift

Understanding Decline in Medical Video Consultation: A User-Centred Approach