Director of Learning and Technology MedU MedU We build Virtual Patient VP cases 501c3 non profit Hanover NH 14 people 29000 new students each year 1000000 cases viewed year Disclosure ID: 807528
Download The PPT/PDF document "Engagement 2.0 Bryant Patten" is the property of its rightful owner. Permission is granted to download and print the materials on this web site for personal, non-commercial use only, and to display it on your personal computer provided you do not modify the materials and that you retain all copyright notices contained in the materials. By downloading content from our website, you accept the terms of this agreement.
Slide1
Engagement 2.0
Bryant Patten
Director of Learning and Technology
MedU
Slide2MedU
We build Virtual Patient (VP) cases
501(c)(3) non profit
Hanover, NH
14 people
29,000 new students each year
1,000,000+ cases viewed / year
Slide3Disclosure
Employee of MedU
Slide4Slide5Slide6By engagement we mean:
" the extent to which students are willing and
able to take on the learning task at hand.”
or
"the learner actively engages in cognitive processes for learning.”
Our focus is cognitive engagement
Slide7History pre 2013
Tracked time on case and performance
never showed performance
Time on card / all cards not meaningful
Slide8Why
Committed to continuous improvement of our metrics
Research
Slide9Engagement Score Algorithm V1.0 2013
Components
answerScore
timeScore
toolbarScore
summaryScore
Equally weighted
Slide10Summary Statement / Machine Learning
Lightside System
Current version is 0 or 1
Code directly embedded in CASUS display system
Slide11BACKGROUND
Isolated online environments require learner autonomy and may not inherently foster learner
cognitive engagement
.
Many clerkship directors are using time on case as an indicator of engagement, but empirical evidence suggests this approach is not optimal.
Little is known about the factors in a virtual patient (VP) case that will promote
cognitive engagement
, which we define as “
the degree to which students are willing and able to take on the learning task at hand
.”
OBJECTIVES
To develop and validate a
computer-generated dynamic engagement score
based on student interactions with MedU VP cases.
METHODS
Engagement Score Development
We developed an engagement score that includes
four equally weighted components
based on student interactions with the case, each of which is tracked by the VP software.
A
scoring algorithm
and preliminary
cut-points
for determining low, moderate or good engagement were developed after reviewing log data from 20 randomly selected students.
Engagement Score ValidationContent: Six medical educators were surveyed to establish content validity of the score components. Response process: Four faculty members reviewed log data for 10 cases and scored student engagement as either low, moderate or good. We then assessed rater agreement with the empirically derived scoring cut-points using Pearson correlation, and we assessed inter-rater reliability using intra-class correlation for these ratings.Consequence: We displayed the engagement score to students as a routine aspect of MedU case use.
Development and Validation of an Engagement Metric for Virtual Patient Cases
Norm Berman, MD, Anthony Artino, PhD
Geisel School of Medicine at Dartmouth and Uniformed Services University of the Health Sciences
RESULTSEngagement Score ComponentsTime on page: > 20 secondsMCQ answer accuracy: cumulative percentUse of clinical reasoning toolbar: scaled score (0–12)Summary statement automated analysis and case match: binary score (0, 1)Total Score Values: Red < 0.3; Yellow 0.3–0.5; Green > 0.5Validity EvidenceContent: All educators agreed that the components of the score reflect engagement.Response process Mean Pearson correlation = 0.98Mean inter-rater reliability = 0.98Consequence: Display of engagement score to students impacts their behavior. Good engagement increased from 72% in week 1 to 86% in week 5.
CONCLUSIONS
A machine-generated engagement metric, based on student actions in a VP case, is feasible. Validity evidence suggests these scores may reflect important aspects of students’ cognitive engagement with the VP cases. IMPLICATIONSThe engagement score appears to be a good indicator of student interaction with MedU cases, and may be better than time on case.The engagement score, as an indicator of cognitive engagement, can serve as an important outcome measure in efforts to improve the design of VPs.The next step in collecting validity evidence for our engagement scores will include correlating these scores with students’ self-reported cognitive engagement using a survey instrument that is currently being validated.
REFERENCES
Artino AR. Think, feel, act: motivational and emotional influences on military students online academic success.
J Comput High Educ (2009) 21:146–166 Berman NB, Fall LH, Smith S, Levine DA, Maloney CG, Potts M, Siegel B, Foster-Johnson L: Integration strategies for using virtual patients in clinical clerkships. Acad Med 2009, 84(7):942–949.
BACKGROUND
METHODS
RESULTS
OBJECTIVES
Engagement
CONCLUSIONS
REFERENCES
IMPLICATIONS
Slide12Response from educators
Strong favorable response
Some summative use
Request for more student information
Slide13Response from students
Anecdotal negative reaction when used summatively
Intense interest in algorithm when used summatively
We do NOT tell students score components
Odd combination of Digital Native and Immigrant
Slide14Performance of students
Close to 90% getting green engagement
80% are getting credit on summary statement
65% correct answers
Slide15Engagement Score Algorithm V1.5
CORE Components
answerScore
timeScore
clicks (hyperlinks & images)
Equally weighted
Not Yet Displayed
Slide16Concerns
Students gaming system
Allowing faculty full access via publication
Tension between faculty & students
requiring engagement vs spying on me
Slide17Engagement Score Algorithm V2.0
Currently Under Development
Components
multi-part answerScore
timeScore
toolbarScore (with semantic analysis)
click tracking
multi-part summaryScore
targeted engagement tools
Slide18Summary Statement / Machine Scoring V2.0
iParadigms System (turnitin)
Five values of 0, 1 or 2
Accuracy
Narrows DD
Semantic qualifiers
Transformative language
Global summary
Web services model
Slide19Possible Future Directions
Next Generation Machine Learning
Parameterized Algorithm
Feed our LA work
Standards and Collaboration
Slide20Engagement Standard
Does one exist?
Is it time to create one?
Slide21tin can
Slide22IMS Caliper Analytics™ Interoperability Standards Reach Candidate Final Release Status
- May 6, 2015
Slide23Slide24bryant.patten@med-u.org
http://www.med-u.org