IDMVis: Temporal Event Sequence Visualization for Type 1 Diabetes - - PowerPoint PPT Presentation

▶
idmvis
SMART_READER_LITE
LIVE PREVIEW

IDMVis: Temporal Event Sequence Visualization for Type 1 Diabetes - - PowerPoint PPT Presentation

Oct 28, 2022 490 likes •1.17k views

IDMVis: Temporal Event Sequence Visualization for Type 1 Diabetes Treatment Decision Support Yixuan (Janice) Zhang Kartik Chanana Cody Dunne Contributions Contributions IDMVis a temporal event sequence visualization tool to support

slide-1
SLIDE 1

IDMVis:

Temporal Event Sequence Visualization for Type 1 Diabetes Treatment Decision Support

Kartik Chanana Cody Dunne Yixuan (Janice) Zhang

slide-2
SLIDE 2

Contributions

slide-3
SLIDE 3
  • IDMVis – a temporal event sequence visualization tool

to support diabetes treatment decision

  • Hierarchical Task Abstraction

Contributions

slide-4
SLIDE 4
  • IDMVis – a temporal event sequence visualization tool

to support diabetes treatment decision

  • Hierarchical Task Abstraction

Hierarchical Task Analysis

Contributions

slide-5
SLIDE 5
  • IDMVis – a temporal event sequence visualization tool

to support diabetes treatment decision

  • Hierarchical Task Abstraction

Hierarchical Task Analysis Task Abstraction

Contributions

slide-6
SLIDE 6
  • IDMVis – a temporal event sequence visualization tool

to support diabetes treatment decision

  • Hierarchical Task Abstraction

Hierarchical Task Analysis Task Abstraction Design

Contributions

slide-7
SLIDE 7

Imagine a 10-year-old kid, who has been diagnosed with type 1 diabetes…

slide-8
SLIDE 8

Imagine a 10-year-old kid, who has been diagnosed with type 1 diabetes…

Continuous Glucose Monitor (CGM)

slide-9
SLIDE 9

Imagine a 10-year-old kid, who has been diagnosed with type 1 diabetes…

Glucose Meter Continuous Glucose Monitor (CGM)

slide-10
SLIDE 10

Imagine a 10-year-old kid, who has been diagnosed with type 1 diabetes…

Glucose Meter Continuous Glucose Monitor (CGM) Insulin Injection

slide-11
SLIDE 11

Imagine a 10-year-old kid, who has been diagnosed with type 1 diabetes…

Continuous Glucose Monitor (CGM) Glucose Meter Insulin Injection Exercise Sickness Food Diabetes Logbook

slide-12
SLIDE 12

During a clinical visit …

slide-13
SLIDE 13

During a clinical visit …

slide-14
SLIDE 14

During a clinical visit …

slide-15
SLIDE 15

During a clinical visit …

slide-16
SLIDE 16

How to help diabetes clinicians make treatment decisions? What are the tasks?

slide-17
SLIDE 17

Hierarchical Task Analysis

slide-18
SLIDE 18

Hierarchical Task Analysis Task Abstraction Design

Task 1 Task 2 Task 3 Task 4 Task analysis:

slide-19
SLIDE 19

Ultimate Goal Task 1 Task 2 Task 1.1 Task 1.2

Hierarchical Task Analysis Task Abstraction Design

Task analysis: ^ Hierarchical

slide-20
SLIDE 20

Hierarchical Task Analysis Task Abstraction Design Develop a treatment plan and educate patients

slide-21
SLIDE 21

0. Develop a treatment plan and educate patients Hierarchical Task Analysis Task Abstraction Design

slide-22
SLIDE 22

0. Develop a treatment plan and educate patients

1. Collect and display the patient’s data

Increasing Task Specificity

Hierarchical Task Analysis Task Abstraction Design

slide-23
SLIDE 23

0. Develop a treatment plan and educate patients

1. Collect and display the patient’s data 2. Overview the patient’s data

Increasing Task Specificity

Hierarchical Task Analysis Task Abstraction Design

slide-24
SLIDE 24

0. Develop a treatment plan and educate patients

1. Collect and display the patient’s data 2. Overview the patient’s data 3. Reason about patient blood glucose levels

Increasing Task Specificity

Hierarchical Task Analysis Task Abstraction Design

slide-25
SLIDE 25

0. Develop a treatment plan and educate patients

1. Collect and display the patient’s data 2. Overview the patient’s data 3. Reason about patient blood glucose levels 4. Educate patients and caregivers

Increasing Task Specificity

Hierarchical Task Analysis Task Abstraction Design

slide-26
SLIDE 26

0. Develop a treatment plan and educate patients

1. Collect and display the patient’s data 2. Overview the patient’s data 3. Reason about patient blood glucose levels 4. Educate patients and caregivers 5. Make a treatment plan

Increasing Task Specificity

Hierarchical Task Analysis Task Abstraction Design

slide-27
SLIDE 27

0. Develop a treatment plan and educate patients

1. Collect and display the patient’s data 2. Overview the patient’s data 3. Reason about patient blood glucose levels 4. Educate patients and caregivers 5. Make a treatment plan

Increasing Task Specificity

3.1 Examine post-event glucose level

Hierarchical Task Analysis Task Abstraction Design

slide-28
SLIDE 28

0. Develop a treatment plan and educate patients

1. Collect and display the patient’s data 2. Overview the patient’s data 3. Reason about patient blood glucose levels 4. Educate patients and caregivers 5. Make a treatment plan

Increasing Task Specificity

3.1 Examine post-event glucose level 3.2 Examine the interplay between events

Hierarchical Task Analysis Task Abstraction Design

slide-29
SLIDE 29

0. Develop a treatment plan and educate patients

1. Collect and display the patient’s data 2. Overview the patient’s data 3. Reason about patient blood glucose levels 4. Educate patients and caregivers 5. Make a treatment plan

Increasing Task Specificity

3.1 Examine post-event glucose level 3.2 Examine the interplay between events

Hierarchical Task Analysis Task Abstraction Design

slide-30
SLIDE 30

3.1 Examine post-event glucose level

Hierarchical Task Analysis Task Abstraction Design

slide-31
SLIDE 31

3.1 Examine post-event glucose level

Hierarchical Task Analysis Task Abstraction Design

Examine related data to understand

  • bservation
slide-32
SLIDE 32

Design Requirements

Hierarchical Task Analysis Task Abstraction Design

slide-33
SLIDE 33
  • DR1. Composite Visualization of Integrated Data

Design Requirements

Hierarchical Task Analysis Task Abstraction Design

slide-34
SLIDE 34
  • DR1. Composite Visualization of Integrated Data
  • DR2. Visualization of Folded Temporal Data

Design Requirements

Hierarchical Task Analysis Task Abstraction Design

slide-35
SLIDE 35
  • DR1. Composite Visualization of Integrated Data
  • DR2. Visualization of Folded Temporal Data
  • DR3. Align and Scale Temporal Data

Design Requirements

Hierarchical Task Analysis Task Abstraction Design

slide-36
SLIDE 36
  • DR1. Composite Visualization of Integrated Data
  • DR2. Visualization of Folded Temporal Data
  • DR3. Align and Scale Temporal Data
  • DR4. Summary Statistics

Design Requirements

Hierarchical Task Analysis Task Abstraction Design

slide-37
SLIDE 37

Design of IDMVis

Hierarchical Task Analysis Task Abstraction Design

slide-38
SLIDE 38

Hierarchical Task Analysis Task Abstraction Design

14-Day Overview Detail View Summary Statistics Panel

slide-39
SLIDE 39

Use small multiples to partition data folded by days

14-Day Overview

Hierarchical Task Analysis Task Abstraction Design

slide-40
SLIDE 40

14-Day Overview

Hierarchical Task Analysis Task Abstraction Design

slide-41
SLIDE 41

Hierarchical Task Analysis Task Abstraction Design Events from logbooks with blood glucose readings Glucose Meter

slide-42
SLIDE 42

Hierarchical Task Analysis Task Abstraction Design Events from logbooks with blood glucose readings Glucose Meter Continuous Glucose Monitor (CGM)

slide-43
SLIDE 43

Hierarchical Task Analysis Task Abstraction Design Events from logbooks with blood glucose readings Glucose Meter Continuous Glucose Monitor (CGM)

normal range (70-180 mg/dL) above range (>180 mg/dL) below range (<70 mg/dL)

slide-44
SLIDE 44

Events from logbooks with blood glucose readings CGM above range (>180 mg/dL) CGM normal range (70-180 mg/dL) CGM below range (<70 mg/dL)

Hierarchical Task Analysis Task Abstraction Design

slide-45
SLIDE 45

A pattern of high blood glucose after lunch shown by purple circles Single-event alignment

Hierarchical Task Analysis Task Abstraction Design

slide-46
SLIDE 46

A pattern of high blood glucose after lunch shown by purple circles Single-event alignment

3.1 Examine post-event glucose level

Hierarchical Task Analysis Task Abstraction Design

slide-47
SLIDE 47

Zoom-in window to show event details

3.1 Examine post-event glucose level

Hierarchical Task Analysis Task Abstraction Design

slide-48
SLIDE 48

Hierarchical Task Analysis Task Abstraction Design

3. Reason about patient blood glucose levels 3.1 Examine post-event glucose level

…

Zoom-in window to show event details

slide-49
SLIDE 49

Stretch time scaling

Events from logbooks with blood glucose readings CGM above range (>180 mg/dL) CGM normal range (70-180 mg/dL) CGM below range (<70 mg/dL)

Dual-event alignment

Hierarchical Task Analysis Task Abstraction Design

slide-50
SLIDE 50

Left-justified time scaling

Events from logbooks with blood glucose readings CGM above range (>180 mg/dL) CGM normal range (70-180 mg/dL) CGM below range (<70 mg/dL)

Dual-event alignment

Hierarchical Task Analysis Task Abstraction Design

slide-51
SLIDE 51

Right-justified time scaling

Events from logbooks with blood glucose readings CGM above range (>180 mg/dL) CGM normal range (70-180 mg/dL) CGM below range (<70 mg/dL)

Dual-event alignment

Hierarchical Task Analysis Task Abstraction Design

slide-52
SLIDE 52

Detail View Hierarchical Task Analysis Task Abstraction Design

slide-53
SLIDE 53

Qualitative Study

slide-54
SLIDE 54

Dietitians Certified diabetes educators (CDEs)

Participants

slide-55
SLIDE 55

Average years of work experience:

17.2 years

Dietitians Certified diabetes educators (CDEs)

Participants

slide-56
SLIDE 56

Day-by-meal table

B C A

Exploration using IDMVis Semi-structured interviews

Methodology

slide-57
SLIDE 57

Results

slide-58
SLIDE 58

Superimposed detail view helps identify issues of data quality (e.g., missing or conflicting data)

So sugar-to-treat [blood glucose] should have gone up from here, not down. It went down. Kept going down. Sugar to treat should be here, before this curve comes back up. That’s my concern. It’s missing something here.

” “

Results

Missing something?

slide-59
SLIDE 59

Sentinel event alignment allows exploration of event sequence relationships

Results

slide-60
SLIDE 60
  • Use single-event alignment to look for event consistency

Sentinel event alignment allows exploration of event sequence relationships

Results

slide-61
SLIDE 61
  • Use single-event alignment to look for event consistency
  • Use dual-event alignment to examine variability of patterns

Sentinel event alignment allows exploration of event sequence relationships

Results

slide-62
SLIDE 62

I like being able to see how you could separate and see between the length, the time between meals... You can’t tell them to eat three times a day at the same time. So it’s just sort of helpful to see the variability... It would help you plan for it in the fact that you might reduce his basal based on the fact that he’s an erratic eater.

” “

Results

slide-63
SLIDE 63

Conclusion

slide-64
SLIDE 64
  • IDMVis – a temporal event sequence visualization
  • Novel techniques for temporal folding
  • Aligning by dual sentinel events & scaling the intermediate timeline

Conclusion

slide-65
SLIDE 65
  • IDMVis – a temporal event sequence visualization
  • Novel techniques for temporal folding
  • Aligning by dual sentinel events & scaling the intermediate timeline
  • Hierarchical task abstraction

Hierarchical Task Analysis Task Abstraction Design

Conclusion

slide-66
SLIDE 66

bit.ly/IDMVis

For more information, please visit

B C A

Kartik Chanana Cody Dunne Yixuan (Janice) Zhang