Human-Data Interaction in Healthcare

📝 Abstract

In this paper, we focus on an emerging strand of IT-oriented research, namely Human-Data Interaction (HDI) and how this can be applied to healthcare. HDI regards both how humans create and use data by means of interactive systems, which can both assist and constrain them, as well as to passively collect and proactively generate data. Healthcare provides a challenging arena to test the potential of HDI to provide a new, user-centered perspective on how data work should be supported and assessed, especially in the light of the fact that data are becoming increasingly big and that many tools are now available for the lay people, including doctors and nurses, to interact with health-related data.

💡 Analysis

In this paper, we focus on an emerging strand of IT-oriented research, namely Human-Data Interaction (HDI) and how this can be applied to healthcare. HDI regards both how humans create and use data by means of interactive systems, which can both assist and constrain them, as well as to passively collect and proactively generate data. Healthcare provides a challenging arena to test the potential of HDI to provide a new, user-centered perspective on how data work should be supported and assessed, especially in the light of the fact that data are becoming increasingly big and that many tools are now available for the lay people, including doctors and nurses, to interact with health-related data.

📄 Content

HUMAN-DATA INTERACTION IN HEALTHCARE Federico Cabitza and Angela Locoro Dipartimento di Informatica, Sistemistica e Comunicazione, Università degli Studi di Milano-Bicocca Viale Sarca 336 20126, Milano, Italy (Phone: +39-02-64487888); E-mail: {cabitza,angela.locoro}@disco.unimib.it ABSTRACT In this paper, we focus on an emerging strand of IT-oriented research, namely Human-Data Interaction (HDI) and on how this can be applied to healthcare. HDI regards both how humans create and use data by means of interactive systems, which can both assist and constrain them, as well as to passively collect and proactively generate data. Healthcare is a challenging arena to test the potential of HDI towards a new, user-centered perspective on how to support and assess data work, especially in current times where data are becoming increasingly big and many tools are available for the lay people, including doctors and nurses, to interact with health-related data. This paper is a contribution in the direction of considering healthcare data through the lens of HDI, and of framing data visualization tools in this strand of research, in order to let the subtler peculiarities among different kind of data and of their use emerge and be addressed accordingly. Our point is that doing so can promote the design of more usable tools that can support data work from a user-centered and data quality perspective. KEYWORDS Human-Data Interaction; Primary Use; Tertiary Data

1. INTRODUCTION Twenty-five years ago, medical informatics was defined as “dealing with the storage, retrieval and optimal use of biomedical data” (Shortliffe et al. 1990). At that time, little emphasis was put on the practices of data production, that is on how medical practice, and single stories of illness, care and recovery are represented, accounted and “datafied” in some objective manner. However, these practices, which include policies, rules, habits, conventions, tools and techniques, have always been intertwined with and affected by the available ITs, as well as by the expectations of the stakeholders on how to make sense and use of health-related data. Different perspectives on these expectations, and on what valuable health data are, lead to manifest chasms between primary use and other uses of health information, as often discussed in the specialist literature (Fitzpatrick and Ellingsen, 2013). To try to cross these chasms, we need to create the suitable language to describe the differences and give some operational definitions. We distinguish between three different macro-types of data and the related processes in which these data are either produced, processed or consumed: namely, primary data, which come from a broad range of sources and are produced within a caring process to make its unfolding seamless and smooth (Berg, 1999); secondary data that are derived from the primary data for purposes different than care, like accounting and medical billing (Abdelhak, Grostick, & Hanken, 2012); and tertiary data, that are produced from the secondary data for any unanticipated need of the potential consumers of health services (see Figure 1).
   To illustrate this tripartition, an analogy from the agriculture domain can be drawn (Locoro, 2016): primary data are like the produce of the land, which farmers grow for themselves as well as the external market. Secondary data are the product of a transformation of these primary data, like the one performed in food industry where vegetables are cleansed and chopped. Tertiary data are further transformed from secondary data to make them more easily consumable, that is suitable for and conveyed to a broader population of consumers in terms of information services, like fresh-cut vegetable products can be seen as the service to have vegetables already ready-to-eat. The definitions mentioned above shed light on the relationship between data and their uses, which cannot be overstated. The tripartition that we propose reflects the different uses and practices in which data are produced and consumed and it calls for a specific area of research focusing on how to support people in interacting with their data of concern and in gaining insight from them: Human-Data Interaction (HDI). HDI is both a phenomenon and a research field focusing on this phenomenon. As a phenomenon that is object of research it is the kind of action where, on the one hand data are produced, processed and exploited by humans; this encompasses, at the two extremes of the action spectrum, both the datification of facts, that is the process by which portions of the reality of interest are translated into the domain of words, symbols and numbers (through coding, classification, and measure1); and what we call data telling, that is the creation of accounts and stories that human can tell according to the data they make (a) sense of. On the other hand, HDI also regards actions in which humans are affected by data,

This content is AI-processed based on ArXiv data.