Background
Related work
Scope
Review of similar solutions in the healthcare context
Assessment methods
Summary
Key contribution
Evaluating visual data analysis and reasoning
Contextualizing the insight-based methodology
From bioinformatics to clinical data
Methods
Hypothesis
Health timeline
Graphical representation and interactivity
Design rationale and requirements
-
Web-based visualization software
-
Acquisition of data from Medicare and PBS
-
Cloud-hosted application software
-
A clean and simple user-interface. For the purpose of the study, no tutorials or explanations were given to the participants. This was done to “stress test” the design of the graphical interface.
-
Interactivity for data exploration (time adjustment via panning and zooming)
Data representation baseline
Insight-based methodology
-
Observation: The observation or finding provided by the participant during the process of analyzing the data via a representation.
-
Time: The amount of time taken to reach the insight.
-
Domain Value: The value, importance, or significance of the insight.
-
Hypotheses: Some insights enable users to identify a new relevant hypothesis.
-
Directed versus Unexpected: Directed insights are those that answer specific questions. Unexpected insights are those that were not considered in the design of the study.
-
Correctness: Insights can be correct or incorrect depending on the data represented in the visualization. Some insights are incorrect conclusions that result from misinterpreting the data visualization. For our study, the insights formulated by the participants need to be clinically valid assessments on the patient’s condition.
Domain value criteria and insight value coding
Value | Criteria | Example |
---|---|---|
1 | Describe the data. No patterns or periodicity spotted. Values described as “low” or “high”. No awareness of the times an event appears in the dataset. | “this is a patient on injectable antipsychotic medications” |
2 | Describe periodicity or frequency of an event. Found patterns, irregularities and amount of repetitions. No conjectures or assumptions. | “the patient has quite a number of GP visits in 2011, 2012 and 2013” |
3 | Requires a conjecture or assumption. Try to explain why an event or value is repeating, missing, following a pattern. Speculation on the treatment, status, follow-up, treatment or behavior of the patient. Prediction on the future status of the patient based on a single event. No correlation with other events. Single conjecture explaining one phenomenon | “patient also has an investigation suggesting that he has some metabolic disease” |
4 | Conjecture or assumption about two or more events. Explanation of probable cause and effect. Ties two events or phenomena together with a probable reason or explanation. Not all the elements in the dataset are explained some relationships remain unknown to this insight. | “in summary I think it is a patient with psychotic illness” |
5 | Hypothesis that ties together the discovered elements and events into one possible explanation. Explains relationship between events. Explains probable underlying reasons of the events. Ties together all the events mentioned beforehand. | “overall this patient presents quite of a complex picture of mainly depression probably complicated with psychotic component anxiety” |
Thinking aloud process
Clinical data
Patient data selection
Clinical data included in the study
Experiment protocol
Visualization presentation and order
Time constraints
Sample size
Participants
Data analysis
-
The experimentId refers to the sequential identifier number of the assessments.
-
The patientId refers to an identifier created for the sole purpose of this study and is used to link together the assessments with the de-identified patient data. The identifier does not contain clinical or personal information.
-
The assessmentId refers to the assessment during which the participants used either the Timeline or the baseline to conduct their observations.
-
The visualization refers to either the Health Timeline (referred to as timeline) or the baseline (referred to as table).
-
The time refers to the time at which the insight began to be voiced by the participant. In some cases, multiple insights share the same starting time indicating that they were given by the participant during the same sentence (thought process).
-
The insight refers to the textual insight as transcribed from the audio.
-
The value refers to the domain value of the insight which follows the criteria detailed in this article.
Metrics
-
Metrics per assessment: For each assessment we calculated the average number of insights, as well as their average and cumulative value.
-
Number of insights: the total count of insights across all the assessments. We also counted the number of insights separated by their value (from one to five points).
-
Time to first insight: the time required by the participant to formulate the first insight regardless of its value. Additionally, we also calculated separately the insight time for each of the values.
Statistical tests
Results
Metrics per assessment
Metric | Baseline | Health Timeline | p-values |
---|---|---|---|
Number of insights | min=6 | min=5 | 0.70 |
max=42 | max=42 | ||
μ=23.04 | μ=22.32 | ||
Md=23 | Md=21 | ||
σ=±10.43 | σ=±9.12 | ||
Cumultive value | min=8 | min=10 | 0.01 |
max=45 | max=56 | ||
μ=24.96 | μ=34.68 | ||
Md=24 | Md=38 | ||
σ=±10.96 | σ=±11.87 | ||
Mean value | min=0.21 | min=0.78 | 0.01 |
max=2.17 | max=2.71 | ||
μ=1.26 | μ=1.70 | ||
Md=1.21 | Md=1.60 | ||
σ=±0.52 | σ=±0.57 |
Number of insights
Metric by Insight Value | Baseline | Health Timeline | Ratio | p-values |
---|---|---|---|---|
Any | 576 | 558 | 1.16 | 0.70 |
1 | 175 | 114 | 0.65 | 0.03 |
2 | 98 | 178 | 1.82 | 0.01 |
3 | 71 | 78 | 1.10 | 0.53 |
4 | 10 | 32 | 3.20 | 0.01 |
5 | 0 | 7 | — | Non-significant |
Metric | Baseline | Health Timeline | p-values |
---|---|---|---|
Insight Value | min=0 | min=0 | <0.01 |
max=4 | max=5 | ||
μ=1.0833 | μ=1.55 | ||
Md=1 | Md=2 | ||
σ=±1.10 | σ=±1.25 |
Time to first insight
Insight Value | Baseline | Health Timeline | Ratio | p-values |
---|---|---|---|---|
Any | μ=7 | μ=13.16 | 1.88 | <0.01 |
σ=±10.02 | σ=±14.24 | |||
>1 | μ=20.36 | μ=22.24 | 1.09 | 0.17 |
σ=±16.66 | σ=±18.48 | |||
>2 | μ=40.01 | μ=51.21 | 1.28 | <0.01 |
σ=±29.96 | σ=±45.98 | |||
>3 | μ=92.83 | μ=68.44 | 0.74 | <0.01 |
σ=±51.27 | σ=±59.88 | |||
>4 | μ=0 | μ=63.50 | — | <0.01 |
σ=0 | σ=±60.06 |