This session outlines the importance of data capture and using sound research methodology to determine the best use of your game-based learning intervention. We explore pertinent learning and behavioral theories, including outcome levels, elements of fidelity, and appraise different strategies that have both succeeded and failed in the use of games in healthcare education. More importantly, we explore the practical aspects of embedding data collection to prove the game’s impact to healthcare education and health.
Presiding Officer Training module 2024 lok sabha elections
Measuring Healthcare Outcomes using Serious Games, Gamification, and Virtual Reality
1. Measuring healthcare
outcomes using serious
games, gamification, and
virtual reality
Todd P Chang, MD
MAcM
serious play conference
2018.07.11
(Manassas, VA)
2. Disclosures / Conflicts of Interest
• American Heart Association
• National Board of Medical Examiners
• Oculus from FaceBook
• Independent Contractor with SBC Med Sim, LLC
2
5. Objectives
Map out
measurable
Outcomes for both
Mediating &
Moderating games
1
Define the 4
Outcome levels for
measuring
Effectiveness of a
Game
2
Discuss three case
studies measuring
effectiveness of
Games on patient
outcomes
3
28. • Barsuk et al. Crit Care Med 2009;37(10):2697-701.
• Barsuk et al. Acad Med 2010;85(10 Suppl):S9-12.
• Barsuk et al. MJ Qual Saf 2014;23(9):749-56.
30. Outcome A
Use
Scores
Outcome B
Knowledge
Gain
Outcome C
Prior
Knowledge
Outcome 01
Provider Action
(Behavior)
Outcome 02
Patient State
(Result)
Outcome 03
Prior Patient
State (Result)
Outcome 04
Health Metric
(Result)
31. CAN MY GAME PREDICT
HOW MUCH IMPROVEMENT
WILL OCCUR?
A question of
Validity
32.
33.
34. Discriminant Validity Testing
Expected Low
Performers
Expected High
Performers
Uncovers any bias due to:
• Differences in Video Game Experience
• “Game Cheating” behavior
• Lack of Functional Fidelity in a Game
• A different target audience
35. Gerard et al. Validity
Evidence for a Serious Game
to Assess Performance on
Critical Pediatric Emergency
Medicine Scenarios. Simul
Healthc 2018;13(3):168-80.
38. Background
• With higher patient volumes but fewer providers,
multi-patient care is now common
• Poor multi-patient care leads to increased errors
• Multi-patient care is not routinely taught nor
universally assessed
39. Purpose
To validate a serious games version of a pediatric ED
that measures multi-patient care based on experience:
1. Undergraduate students (jrs, srs)
2. Medical students (MS-II)
3. Resident physicians (PGY-1 through 3)
4. Attending / Fellow physicians (PGY-4+)
45. Outcomes & Analyses
Predictors
• Experience
• Multi-tasking ability
• Video Game frequency
Game Outcomes
• % Sentinel Orders
• Time to 1st Sentinel Order
• Time to 2nd Sentinel Order
• Time to Discharge
• % Patients Seen
• # correct Diagnoses
• # Differential Diagnoses
45
49. Results
• MANCOVA analyses revealed a statistically
significant difference in game performance after
controlling for video game experience & multi-
tasking ability.
• The two outcomes that the game was able to
distinguish expertise in were:
– # sentinel orders
– % correct diagnoses
50. 50
The one-way MANCOVA showed that there was statistically significant
difference (p=0.03) between the skill groups on the combined dependent
variables after controlling for game play frequency, F(21, 153) = 0.700, p =
.828, Wilks' Λ = .768, partial η2 = .084, and for MTAT placement score, F(21,
153) = 1.365, p = .144, Wilks' Λ = .610, partial η2 = .152.
Follow up univariate one-way ANCOVAs were performed. A Bonferroni
adjustment was made such that statistical significance was accepted at p <
.008. There were statistically significant differences in adjusted means for
total percentage of sentinel orders completed (F(3, 59) = 31.702, p < .0005,
partial η2 = .617), and number of correct diagnoses made per patient
seen(F(3, 59) = 21.441, p < .0005, partial η2 = .522).
52. Results
• Linear Regression for these two outcomes showed
that Expertise accounts for:
– 64.3% of the variation in # sentinel orders
– 50.4% of the variation in % correct diagnoses
53. Next Steps
• VitalSigns will emphasize the two outcomes that
the game was able to distinguish expertise in
– # sentinel orders
– % correct diagnoses
• The UI needs to be re-examined to determine why
time-to-critical action did not differ
– New input methods (e.g. mouse vs. finger vs. keyboard)
55. Background
• Low-frequency, high-stakes medical events
– Need trained, competent health providers
– Are too infrequent and too high-stakes to allow
trainees to ‘practice’
• Immersive Virtual Reality is an emerging, unproven
simulation technology for low-frequency, high-stakes
events
56. Purpose
To describe ED provider stress physiology during
actual & Virtual Reality resuscitations by measuring
Heart Rates
To determine performance differences between
residents & attendings within the VR
63. Results - Performance
Resident Attg
Variable n Mean
Std
Dev
n Mean
Std
Dev
p-value
Sc.4 Cricothyrotomy (sec) 4 90.50 39.32 6 31.83 16.99 0.019*
64. Results - Performance
Resident Attg
Variable n Mean
Std
Dev
n Mean
Std
Dev
p-value
Sc.4 Cricothyrotomy (sec) 4 90.50 39.32 6 31.83 16.99 0.019*
# lorazepam ordered 15 4.67 1.84 19 2.63 1.57 0.003*
# levetiracetam ordered 15 0.93 0.88 19 0.89 0.94 0.864
# fosphenytoin ordered 15 1.73 0.80 19 1.00 0.67 0.014*
65. Next Steps
• VR scenarios will emphasize the 2 outcomes that
the VR was able to distinguish expertise in
– # anti-epileptic drugs
– Time-to-advanced airway
• The UI needs to be re-examined to determine why
other time-to-critical action did not differ
– Easier inputs and assistance
• Target audience needs to be more novice
67. Purpose
To determine if a PBL gamification system (points-
badges-leaderboards) improves self-initiated simulated
CPR practice frequency and skill
68. Background
• Healthcare providers do not routinely provide high-
quality chest compressions
• CPR skills begin to decay at 3 months
• Strategies to increase CPR practice frequency are
have yet to be developed
Cheng A et al. JAMA Pediatrics 2015. Sutton R et al. Pediatrics 2011.
72. 72
Chang TP et al. Leveraging Quick Response Code Technology to facilitate Simulation-based Leaderboard Competition. Sim
Healthc 2018.
73. Outcomes & Analyses
Predictors
• Access to a Selfie-based
Leaderboard
Game Outcomes
• Frequency of self-motivated
CPR practice
• CPR performance
73
76. 76
Control Intervention All access Total
Number of
participants
319 600 95 1014
% 31.4 59.2 9.4 100.0
Control Intervention All access Total
Number of
plays
685 1165 143 1993
Plays per
participant
2.14 1.94 1.51 1.97
77. Adjusted (for Crossover effect) results
77
Control Intervention p-value
Number of
participants
319 600
Mean plays per
person (95%CI)
1.62
(1.36 to 1.88)
1.71
(1.46 to 1.96)
0.6
Mean score
(95%CI)
89.3
(87.6 to 91.0)
90.7
(89.4 to 92.0)
0.19
• No Primary Effect
• No Crossover Effect
• No change in CPR score
78. • Halo-seeking effect & Satiety
• Insubstantial or Divisive
competition
• PBL as insufficient gamification
mechanics
• Short-term vs long-term
expectations
79. Circling Back
Map out
measurable
Outcomes for both
Mediating &
Moderating games
1
Define the 4
Outcome levels for
measuring
Effectiveness of a
Game
2
Discuss three case
studies measuring
effectiveness of
Games on patient
outcomes
3
80. Contact information
• Todd P Chang, MD MAcM
Children’s Hospital Los Angeles / Keck School of Medicine at the
University of Southern California
80
Kirkpatrick’s 4 level model for training evaluation, which is the most extensively used system for assessing educational interventions. I’ll discuss each level and the method by which my research project assesses that outcome.
Kirkpatrick’s 4 level model for training evaluation, which is the most extensively used system for assessing educational interventions. I’ll discuss each level and the method by which my research project assesses that outcome.
Kirkpatrick’s 4 level model for training evaluation, which is the most extensively used system for assessing educational interventions. I’ll discuss each level and the method by which my research project assesses that outcome.
Kirkpatrick’s 4 level model for training evaluation, which is the most extensively used system for assessing educational interventions. I’ll discuss each level and the method by which my research project assesses that outcome.
Kirkpatrick’s 4 level model for training evaluation, which is the most extensively used system for assessing educational interventions. I’ll discuss each level and the method by which my research project assesses that outcome.
Competency Assessment of IJ / SC CVC insertion of all residents before ICU (2010 – 2012). 3.82 --> 1.29 per 1000 catheter days: 74% reduction in CLABSI.
The QCPR simulator was equipped with the standard 45kg spring with sensors within to capture compressions in real time.