ALBECS-2024: Workshop on Algorithmic Behavior Change Support

10th April 2024

This workshop is part of the 19th International Conference on Persuasive Technology 2024

In-person in Wollongong, Australia & with a hybrid morning session

Register for joining online

Motivation

Working out more, reducing food waste, being kinder to others, ... - there are many behaviors people wish to change. Yet, while many behavior change applications have been developed to support people, users commonly do not adhere to these applications or abandon them entirely. To address the evident mismatch between what the applications offer and what users need, a variety of algorithmic approaches has been developed to adapt what the applications offer, when, how, and with whom. The term "algorithm" or "algorithmic approach" is thereby meant broadly to capture any computerized procedure for adapting behavior change support. These algorithmic approaches thereby draw from insights in areas such as nudging, persuasive technology, eCoaches, and conversational agents and use various algorithmic techniques from recommender systems to machine and reinforcement learning. Placing the human at the center and combining the strengths of humans and technology (i.e., augmented or hybrid intelligence) often is an important design guideline, as is accounting for ethical and societal values.

This Workshop

This workshop brings together researchers, designers, developers, practitioners, and educators who are interested in the concept, development, evaluation, and impact of algorithmic behavior change support. Thus, while the Persuasive Technology conference considers all forms of technology, the focus of this workshop is on algorithms (e.g., based on logistic regression, reinforcement learning, recommender systems) that support behavior change. We are thereby interested in algorithms at both the macro level (i.e., which are part of behavior change applications) and the micro level (i.e., which support forms of behavior change as part of a technology that has a goal other than behavior change). We explicitly invite participants from various backgrounds such as artificial intelligence, human-computer interaction, psychology, medical practice, and ethics of technology to contribute their perspectives and experiences.

The broader objective of this workshop is to strengthen the community of people working on adaptive support for behavior change. To this end, the workshop aims to create a lively exchange of ideas that benefits both the current and future research of the individual workshop participants. Specifically, the workshop's aim is to a) learn about each other's work, b) jointly work on problems of the workshop participants, and c) establish a vision for future work on algorithmic behavior change support.

Below is a non-exhaustive list of possible topics in the context of algorithms for adaptive behavior change support:

  • Methods and guidelines that can be used when designing algorithms (e.g., accounting for ethical and societal values, augmented or hybrid intelligence)
  • Novel algorithms
  • Application of algorithms in behavior change contexts (e.g., education, health, sustainability)
  • FAIR implementation of algorithms
  • Evaluation methods for algorithms (e.g., micro-randomized trials)
  • Effectiveness of (components of) algorithms in simulated, experimental, or in-the-wild settings

If you are unsure whether your topic fits the scope of the workshop, please do not hesitate to ask us (see Contact-section).

Call for Contributions

Submissions to this workshop are in the form of papers describing both the authors' work and a problem that they would like to work on together with the other workshop participants. Working with other workshop participants on one's own problem is a unique opportunity to get input from a large number of experts. Examples include:

Questionnaire creation:
-> creating question formulations in small groups, discussing if questions measure intended constructs
Study design:
-> getting feedback on one's draft experimental design for evaluating an algorithm
Instructions for study:
-> letting workshop participants follow an initial instruction set to see where they get stuck, which things are unclear, or which results they would get
Applying design guidelines:
-> after having developed design guidelines, asking workshop participants how they would apply the guidelines to their own research or how they (dis)agree with the guidelines
Algorithm design:
-> based on an initial algorithm idea, asking workshop participants for important concepts or guidelines to account for when further developing or implementing the algorithm
Confirming guidelines for algorithm design:
-> based on an initial set of guidelines one wants to use for one's algorithm design, asking workshop participants how they (dis)agree with these guidelines to establish community agreement
Establishing codes for a thematic analysis:
-> based on free-text responses providing feedback on an algorithm used in a previous study, asking workshop participants to in small groups come up with codes describing the concerns of study participants
Establishing themes for a thematic analysis:
-> based on codes describing the concerns of study participants about an algorithm, asking workshop participants to discuss possible themes

The authors should clearly describe how they think that working on their problem with the workshop participants could benefit their work.

We have already accepted 4 submissions for in-person presentations (i.e., at least 1 author should attend the workshop in person). Now, we are additionally looking for online presentations (i.e., at least 1 author should attend the workshop online). At least 1 author should be ready to 1) present their work in a short 7-10 minute presentation and 2) work on their problem with the workshop participants for about 30 minutes. On rejection, you are also welcome to still join our workshop. To attend the workshop, participants must be registered for the workshops/tutorials at Persuasive Technology 2024.

All submissions should adhere to either the CEUR Workshop Proceedings guidelines or the Springer LNCS guidelines. Papers describing the authors' work should be written in English and have either 2-6 (short paper) or 7-12 pages (long paper), excluding references. Workshop proceedings will be published in CEUR under Creative Commons License Attribution 4.0 International (CC BY 4.0), though authors can choose to opt out from this. This means that the final accepted papers will be published in the CEUR format. In addition to the papers describing the authors' work, the submission should include in an appendix the description of the problem the authors would like to work on together with workshop participants. This description does not count toward the page limit and will not be published. Please indicate in your submission whether you want to present in-person or online (i.e., a hybrid setting).

Please anonymize your submission.

You can submit through easychair.

Important Dates

  • Submission deadline: 22nd February 2024 20th March 2024
  • Notification deadline: 29th February 2024 22nd March 2024
  • Camera Ready: 3rd April 2024
  • Workshop: 10th April 2024

Keynote Speakers

Nina Deliu

Sapienza University of Rome, University of Cambridge

Designing experiments via bandit algorithms: modeling considerations for better outcomes

Abstract. The multi-armed bandit (MAB) framework holds great promise for designing adaptive experiments with outcomes and resource (e.g., cost, time, or sample size) benefits. For example, it can result in better participant outcomes and improved statistical power at the end of a trial. However, due to mathematical and computational aspects, most MAB variants have been developed and are implemented under binary or normal outcome models. In this talk, guided by three case studies we have designed, I will illustrate how traditional statistics can be integrated within this framework to enhance its potential. Specifically, I will focus on the most popular Bayesian MAB algorithm, Thompson sampling, and on two types of outcomes: (i) rating scales, increasingly common in recommendation systems, digital health and education, and (ii) zero-inflated data, characterizing mobile health experiments. Theoretical properties and empirical advantages in terms of balancing exploitation (outcome performance) and exploration (learning performance) will be presented. Further considerations will be provided in the unique and challenging case of small samples.

Deborah Richards

Macquarie University

Human-Empowered Computer-Mediated Behavior Change

Abstract. This talk will present a range of different applications and studies focused on behavior change in contexts such as incontinence and sleep disorders in children, ethical decision-making by cybersecurity professionals, adherence to treatment advice, reducing study stress, emotion regulation, and self-management of stroke recovery. Underlying the design of each application are principles ensuring that the application is evidence-based, empowering, empathic, and ethical. To achieve this, we will consider the process/workflow, components, use of knowledge bases as well as algorithms and dialogue templates, and adaptation to the user. The talk will conclude with a consideration of ethical issues in the design and use of such systems.

Bio. Deborah Richards is a Professor in the School of Computing at Macquarie University. Whilst working in the ICT industry, she completed a BBus (Comp and MIS) in 1989 and MAppSc (Info Studies) in 1995. Deborah completed a PhD in artificial intelligence on the reuse of knowledge at the University of New South Wales and joined academia in 1999. She is currently the Director of the Virtual Reality Laboratory and Director of Industry and External Relations for the School of Computing. She has been an artificial intelligence researcher since 1993, initially focused on the acquisition and reuse of knowledge and knowledge-based systems. In 2003, she moved to agent-based systems with a current emphasis on intelligent virtual agents and intelligent virtual worlds. The ethical use of these technologies has been a primary concern for over a decade. With 20 years in industry prior to joining academia, she is specifically interested in applying technology to improve current shortcomings and overcome barriers faced by stakeholders in education, training, health and well-being. She has over 400 refereed publications and over $5M in competitive external grant funding.

Organizers

Nele Albers

Delft University of Technology

Amal Abdulrahman

Delft University of Technology

Deborah Richards

Macquarie University

Caroline Figueroa

Delft University of Technology

Bibhas Chakraborty

National University of Singapore, Duke University

Ananya Bhattacharjee

University of Toronto

Linwei He

Tilburg University

Mark A. Neerincx

Delft University of Technology, TNO

Joseph Jay Williams

University of Toronto

Nezih Younsi

Sorbonne University

Tibor Bosse

Radboud University

Annemiek Linn

University of Amsterdam

Crystal Smit

Erasmus University Rotterdam

Willem-Paul Brinkman

Delft University of Technology

Tentative Program

This workshop consists of two keynotes by Nina Deliu and Deborah Richards and five paper presentations. Each of the five paper presentations consists of a short presentation and a collaborative session in which workshop participants together work on a problem presented by the authors.

To receive the link for joining the hybrid morning session online, join our Google group. We will share the link with group members at least 30 minutes before the start of the first keynote.

Time Activity
8h30 Registration & morning tea
Morning session (hybrid)
9h00 Keynote 1: Designing experiments via bandit algorithms: modeling considerations for better outcomes (Nina Deliu)
10h00 Break
10h15 Session 1

Personalizing eHealth applications
Integrating Digital Calendars with Large Language Models for Stress Management Interventions
Pranav Rao, Sarah Yi Xu, Ananya Bhattacharjee, Yuchen Zeng, Alex Mariakakis and Joseph Jay Williams
Expert Insights on Conversational AI Systems as an Information Intermediary for Patients and Healthcare Providers for Diabetes Lifestyle Change
Pei-Yu Chen, Sophie van Gent, M. Birna van Riemsdijk, Myrthe Tielman and Tjeerd Schoonderwoerd
Explanation Patterns for The Sleep Adherence Mentor (SAM)
Amal Abdulrahman, Deborah Richards, Patrina Caldwell and Karen Waters
12h30 Lunch
Afternoon session (in-person only)
13h30 Session 2

Guidelines for algorithm design
Advancing Ethical and Inclusive Algorithm Design: Collaborative Strategies for Bias Detection and Mitigation
Adeel Ahmed, Ali Husnain and Abdul Wahid Toor
Algorithmic Support for Health Behavior Change: A Scoping Review Protocol
Diederik Heijbroek, Nele Albers and Willem-Paul Brinkman
15h15 Afternoon tea
15h30 Keynote 2: Human-empowered computer-mediated behavior change (Deborah Richards)
16h30 Reflection & closing remarks
Evening program
17h00 Conference welcome reception
19h00 Optional dinner with workshop participants at restaurant TBD.

Registration

Please register through the Persuasive Technology conference website. Note that registration fees are in Australian Dollar (AUD).

To receive the link for joining the hybrid morning session online, join our Google group. We will share the link with group members at least 30 minutes before the start of the first keynote.

Contact

For any question related to the workshop please contact Nele Albers:

N dot Albers at tudelft dot nl (also hyperlink available at Organizers section)