Simulation for Health Technology Assessment
Post-event report
What is DICE?
From the 4th to 9th of January 2021, HIPER hosted an online course on DICE simulation with guest lecturers Adjunct Professor Jaime Caro (McGill University) and Mr Jörgen Möller (EVIDERA). The course was a unique learning opportunity as instructors are pioneers of the DICE method. DICE, which stands for Discretely Integrated Condition – Event simulation, presents a transparent and elegant alternative to traditional modelling methods.
As an innovative approach to modelling in HTA, a DICE model does not limit the user to health states and transitions between them and thus facilitates problem design. Furthermore, time in DICE is managed through a series of nested loops, instead of relying on Excel rows to represent time. Such features allow users to build up the model quickly, make changes easily, and overcome the complexity associated with relaxing the memoryless property and conducting sub-group analysis in Markov models. In short, DICE is not only conveniently transparent for reviewers, but also makes the lives of modellers easier.
Overview of the course
The module kicked-off with a welcome and introduction by Assistant Professor Wee Hwee-Lin and quickly proceeded to delve into the nuts and bolts of DICE. Over the next several days, students were presented with a blueprint and learned how to build a Markov model in DICE, conduct uncertainty analysis, and debug their models. At the same time, participants became acquainted with the deeper inner-workings of modelling for HTA, and learned about hazard functions, probability distributions, and conditional survival. Next, another layer of complexity was added to the base case example by introducing elements of time-to-event modelling and parametric survival analysis. This enabled us to “individualise” the cohort model by introducing some degree of heterogeneity. As a final step, the module exposed students to the importance of validation and made a case for open-source models.
In terms of the general course structure, theoretical concepts were introduced by Professor Caro in the early afternoon, and later demystified through practical application in Excel. Having the hands-on experience in the form of group workshops has significantly enhanced the value of the course and equipped students with the necessary skills and confidence to apply DICE. The module culminated in the submission of a group DICE model and an online assessment testing students’ understanding of the relevance and credibility of evidence from a modelling study.
Who participated?
Participants came from diverse professional backgrounds and included clinicians, the pharmaceutical industry, and academia. The diverse perspectives enriched class discussions and group work sessions. Technical assistants successfully navigated the varying levels of participants’ prior knowledge of decision-analytic modelling by explaining the concepts in lay terms and finding relevant examples from each discipline.
When asked to share their thoughts and impressions of the seminar, here is what participants had to say:
”The faculty, including the teaching assistants, are knowledgeable and help the students a lot with their learning. ”
”I really enjoyed the group work sessions, the lectures were clear, and all questions were answered, the TAs provided a lot of guidance and balanced well between letting us figure it out on our own and step in.”
