Optimising communication for clinical decision-making - perspective from an expert non-pathologist

Research output: Contribution to conferenceAbstract

Abstract

Decision making in medicine is largely influenced by the veracity and communication of the clinical information being reported and presented. This paper does not address the problem of veracity (accuracy) but focusses on the methods and principles that can be used to optimize the interaction design of electronic reporting as well as the communication and presentation of the report that is to be subsequently interpreted by clinical staff other than pathologists.

Approximately there are 82 million pathology and cytology reports made every year in the US alone [1]. These reports are particularly critical to decision making in oncology given the high incidence rate of various cancers and since the report itself directly informs which patient treatment pathway the clinician may prescribe. However, surgeons have been reported not to understand 30% of the pathology reports they receive [2]. Nevertheless, the current best practice for cancer reporting is arguably electronic synoptic reporting where reports are structured and generated from checklists and interactive form elements. Studies have shown that electronic synoptic reporting is easier to use and increases efficiency and data input quality in comparison to conventional clinical narration using dictation [3-4]. In [3] computerized synoptic reports resulted in the collection of 99% of the required data whereas standard reporting resulted in the collection of 45.9% of the data. It has also been reported that it only takes 6 minutes on average to complete an electronic synoptic report [3]. In addition to the obvious benefits, having data collected in a structured manner provides promising opportunities for artificial intelligence by allowing machine learning models to be trained using high quality clean data that uses a standard taxonomy.

Synoptic structured reporting can also be enhanced by intelligent branching where form options and questions only appear when they are relevant to the answers provided by the user from the prior questions and checkboxes. This clearly saves time by removing the need to cognitively process questions to only then provide the standard answer of ‘non-applicable’ to these irrelevant questions. Moreover, to aid decision making during digital synoptic reporting, smart prompts and tailored assistance can be provided at each stage of the workflow. Many of these prompts or decision support aids can be adaptive, hence they are only shown when needed by the user, e.g. some aids could be triggered from user tracking data or the user themselves.

Whilst electronic synoptic reporting is currently the best model, other forms of human-computer interactions are now possible. This includes conversational forms where a chatbot [5] is used to elicit answers from the user. This conversational reporting method is natural and akin to human-to-human interaction. Chatbots use artificial intelligence to allow the user to have a text or voice-base dialogue with a computer. Chatbots are particularly useful for collecting standard structured data and can intuitively guide the user through the reporting procedure without the use of a cumbersome user interface elements or cumbersome form elements that often comprise of distractions and information that is irrelevant to the report in many contexts.

Whilst efforts have been made to streamline synoptic reporting, little attention has been given to the communication, interaction design and formatting of pathology reports [1]. Powsner et al. [2] and other studies in various disciplines report that formatting style can influence report comprehension, even if the content is identical. With this in mind, my presentation will provide an overview of key principles that can be used to optimise the communication of pathology reports. These principles originate from the human-computer interaction discipline and other sub-disciplines including usability engineering, cognitive science, graphic design and Gestalt psychology. This presentation will include clear descriptions and application of Fitts’ Law, Hicks’ Law, cognitive affordances, the human perception system, models of the human mind (i.e. dual process theory of cognition), the human’s cognitive limitations (Miller’s 7±2 rule [6] and serial position effects) and the importance of visual hierarchy and information architecture [7]. In addition, we will discuss how user experience design methods can be used to enhance pathology reporting, and how usability engineering methods can be used to measure the usability and user experience of reporting and report interpretation [8-9].

Given the effectiveness of electronic synoptic reporting, we will also discuss the guidelines on interactive form design and associated experiments undertaken in the human-computer interaction research community [10-13]. This will include recommendations on presenting required and optional form elements as well as layout, user feedback and error reporting. Studies have shown that users are more effective in completing forms when the form design uses the form design guidelines [10-11]. This presentation will also provide a description of the principles of ‘usability’, including learnability, efficiency, memorability, errors and satisfaction [9]. We will also look at the role of delightful and desirable user interfaces in medicine, and in particular Norman’s theory that attractive products work better due to the subtle positive emotions induced from perceiving attractive visual designs, which in turn provides the user with a better mental state towards using that system [14].

This presentation will allude to a toolkit and the need for a design audit tool to ensure pathology reports adhere to a standard set of design principles. If we cannot standardize reporting, then we can at least standardize the design guidelines for reporting and perhaps even the methods for report usability testing and report design auditing practices. However, there are many options to consider such as whether one should split report forms over multiple screens, or whether to use subtle micro-animation in the interface, or to use smart prompts or the extent of interactive data visualization, especially for cancer genomic reporting [15].

Finally, this presentation will discuss the research undertaken by Bond et al. [16-21] on decision support systems, clinical decision making and eye tracking in medicine. Most of his work has been carried out in the computerized electrocardiology domain but is transferable to pathology. Other concepts will be discussed including ethical-by-design methods, form accessibility and the recent recognition of automation bias resulting from the introduction of smart digital technologies in medicine.
Original languageEnglish
Publication statusAccepted/In press - 10 Aug 2018
EventEuropean society of pathology congress - Spain, Bilbao, Spain
Duration: 8 Sept 201812 Sept 2018
http://www.esp-congress.org/

Conference

ConferenceEuropean society of pathology congress
Country/TerritorySpain
CityBilbao
Period8/09/1812/09/18
Internet address

Keywords

  • synoptic reporting
  • UX
  • pathology
  • medical reporting
  • structured reporting
  • electronic reporting
  • usability
  • cognition ergonomics
  • human factors

Fingerprint

Dive into the research topics of 'Optimising communication for clinical decision-making - perspective from an expert non-pathologist'. Together they form a unique fingerprint.

Cite this