Machine learning (ML) is increasingly used in healthcare practices, due to its potential to support personalization, diagnostic and prediction, automatization, and increase effectiveness. In physiotherapy, most existing ML solutions suggest replacing the physiotherapist, neglecting the complexity of their skills and practice. We articulate an alternative to the design of ML technology for physiotherapy: one that emphasizes the relational aspects of the practice and offers personalized support to physiotherapists and patients alike. Based on domain studies and design explorations with physiotherapists, interaction designers and ML experts, we present 1) insights on physiotherapy's in-clinic and out-of-clinic looped structure, 2) opportunities and requirements to integrate ML in that loop, and 3) a conceptual interactive ML-based infrastructure that exploits those opportunities. Our work widens current ML developmental aims for physiotherapy, proposing a vision that encodes sustainable sociotechnical relationships in healthcare practices.

The paramount measure of success for a machine learning model has historically been predictive power and accuracy, but even a gold-standard accuracy benchmark fails when it inappropriately misrepresents a disabled or minority body. In this work, we reframe the role of machine learning as a provocation through a case study of participatory work co-creating exergames by employing machine learning and its training as a source of play and motivation rather than an accurate diagnostic tool for children with and without Sensory Based Motor Disorder. We created a design probe, Cirkus, that supports nearly any aminal locomotion exergame while collecting movement data for training a bespoke machine learning model. During 5 participatory workshops with a total of 30 children using Cirkus, we co-created a catalog of 17 exergames and a resulting machine-learning model. We discuss the potential implications of reframing machine learning’s role in Assistive Technology for values other than accuracy, share the challenges of using “messy” movement data from children with disabilities in an ever-changing co-creation context for training machine learning, and present broader implications of using machine learning in therapy games.

Body maps are a popular tool in body-centric design, facilitating a sensitization and expression of felt sensations and emotions. Yet, they also bring forth assumptions about the body and our somatic experience. Based on an open and exploratory design ideation inquiry, we have started to explore how body maps could be advanced so as to cater to a plurality of bodies and aspects that shape somatic experiences. We present an annotated portfolio featuring six design themes (temporality, sociality, representativeness, granularity, context, focus). These themes help us examine implicit assumptions of current body maps, and offer possible alternatives for what future body maps could become. We contribute our themes, inspirational design ideas and practical design techniques to help craft novel body maps. Our contributions can serve as inspiration to others, towards advancing body maps as a research tool for body-centric interaction design.

Electrically conductive fluid injected during the operation of hydraulic fracturing can be mapped by anomalous electromagnetic (EM) fields under the excitation of controlled EM sources on the surface. Steel casings, existing in the area of fracturing operation and sources of strong EM responses, need to be considered during survey design and data interpretation. We develop a novel numerical approach that efficiently simulates the effect of steel casings by assigning lumped conductive properties to mesh edges, so finely discretized cells for casings are avoided. In our numerical study, a multistage fracturing along a horizontal well is considered, and the operation is monitored by two survey configurations: long-offset excitation and near-well excitation. Time-lapse analysis shows that the steel casings always have nonnegligible effect in EM field data, even for the long-offset excitation with a distant EM source; the near-well excitation, compared with other conventional configurations, has stronger EM responses and can maintain a high level of sensitivity throughout the entire fracturing procedure. Our results help the design of a controlled source EM survey for a more effective fracturing monitoring campaign.