Exploring the Relationship Between Eye Movements and Electrocardiogram Interpretation Accuracy

Abstract

Interpretation of electrocardiograms (ECGs) is a complex task involving visual inspection. This paper aims to improve understanding of how practitioners perceive ECGs, and determine whether visual behaviour can indicate differences in interpretation accuracy. A group of healthcare practitioners (n = 31) who interpret ECGs as part of their clinical role were shown 11 commonly encountered ECGs on a computer screen. The participants’ eye movement data were recorded as they viewed the ECGs and attempted interpretation. The Jensen-Shannon distance was computed for the distance between two Markov chains, constructed from the transition matrices (visual shifts from and to ECG leads) of the correct and incorrect interpretation groups for each ECG. A permutation test was then used to compare this distance against 10,000 randomly shuffled groups made up of the same participants. The results demonstrated a statistically significant (α 0.05) result in 5 of the 11 stimuli demonstrating that the gaze shift between the ECG leads is different between the groups making correct and incorrect interpretations and therefore a factor in interpretation accuracy. The results shed further light on the relationship between visual behaviour and ECG interpretation accuracy, providing information that can be used to improve both human and automated interpretation approaches.

How different visualizations affect human reasoning about uncertainty: An analysis of visual behaviour

“…To understand the reasoning process, we considered both which items of information participants fixate on, and, crucially, how they compare these items, which can be ascertained from how they transition between them. Transition frequencies were represented as probability distributions, and a permutation test was used to assess whether the difference, calculated using a distance metric (Davies et al 2016;Kübler, Kasneci, and Rosenstiel 2014), between those making correct inferences and those making incorrect inferences was significantly greater than the difference between groups created at random.…”

An investigation of the effects of n-gram length in scanpath analysis for eye-tracking research

“…We chose the "Tree" subset because this representation shows complete information, whereas the Venn representation showed only part of the data; thus, the former was more suitable for our purpose of investigating how people combine different pieces of information during reasoning tasks. The second dataset ("ECG") was taken from a study investigating the relationship between ECG (electrocardiogram) interpretation and eye movements [Davies et al 2016]. A group of 43 medically trained participants were asked to look at an ECG of a patient and infer the underlying medical condition.…”

“…It follows that the possible n-grams of the scanpath ABC are A, B, C, AB, BC and ABC; whereas the possible subsequences of this scanpath include all the n-grams above plus AC; thus, omitting the central letter B. Frequency-based methods count frequencies of n-grams and compare the distributions of these in different groups [Kübler et al 2017]. A distance measure can be used to find significant differences between groups in terms of distributions of n-gram frequencies [Davies et al 2016;Kübler et al 2014Kübler et al , 2017. These techniques have been applied more or less successfully to viewing artwork, search tasks, video game play, driving tasks, and medical image interpretation [Davies et al 2016[Davies et al , 2017Kübler et al 2014Kübler et al , 2017 Hidden Markov Modelling, another frequency-based method, works by generating probability distributions for sequences of AOI transitions.…”

“…A distance measure can be used to find significant differences between groups in terms of distributions of n-gram frequencies [Davies et al 2016;Kübler et al 2014Kübler et al , 2017. These techniques have been applied more or less successfully to viewing artwork, search tasks, video game play, driving tasks, and medical image interpretation [Davies et al 2016[Davies et al , 2017Kübler et al 2014Kübler et al , 2017 Hidden Markov Modelling, another frequency-based method, works by generating probability distributions for sequences of AOI transitions. However, the composite probabilities generated using these models may not represent the aggregate subsequences across a group of scanpaths due to the fact that often they do not go beyond first or second order Markov chains [Goldberg and Helfman 2010a].…”

Conflicts – Oil Exploration and Water

“…Oil contaminants have been detected in virtually all water media used by the public for different purposes, from surface water to rainwater and from hand-dug wells to borehole groundwater sites. [15,23,[53][54][55][56] The social and economic consequences have been devastating for the local population and are further discussed in the next section.…”