Today, @noemie, @chunhua and I were brainstorming a big idea, in part prompted by a good suggestion by @gowtham_rao: Imagine there was a graduate-level (masters and PhD) course offered in ‘healthcare data sciences’, which taught students from either the clinical sciences, public health, computer science, or statistics the applied skills to take clinical questions, translate into an observational study design, and execute the study to generate reliable evidence that can meaningfully inform medical decision making. Students would learn about healthcare data (claims, ehr, registries), would learn data wrangling (including database basics with sql) and would cover 3 flavors of analyses (descriptive predictive, and causal inference) with specific skills in r or python. We discussed the various modalities of such a course: you can do traditional on-campus lecture/lab format, you can do a mooc, you can offer video tutorials on youtibe, etc, each has their own merits and limitations. Imagine a different model, where universities partnered together to each offer the same course during the same calendar time/semester. There would be shared lectures/mooc-style lectures which taught principles and basics of healthcare data science using some common datasets (think synpuf, mimic, nhanes, hcup nis), but then on each campus, universities would also conduct labs on their campus-specific datasets to answer real clinical quesrions, either initiated by the students or prompted by the faculty. During the semester, students from across campuses could learn from each other via some common platforms, like a slack channel and github. Students’ final project would be to complete a study on their local data (hopefully resulting in a publication) but would also be provided the opportunity to submit their project across entire ohdsi network (ideally resulting in another publication).
So our questions to the community, before going too much further down this flight of fancy:
for our academic partners:
- do you have some course on your campus that already covers this content? if yes, what is it? if not, any reason why?
- would you or a colleague on your campus have interest in leading such a course within your university?
- do you have a omop cdm instance and associated computing infrastructure (database server and analysis server) which could be made accessible to students for coursework with the appropriate legwork? if yes, what’s your data? if not, what barriers do you have to establishing such an environment?
- how willing would your institution be in collaborating in a cross-institutional educational opportunity like this? would the novelty and impact of such an idea make it fly, or would the suggestion of collaboration create insurmountable administrative hurdles?
- on a scale from 1-10, how excited are you by this idea…1=i’ll never do it no matter what you say, to 10= awesome, sign me up for the first cohort of universities, I’m all in!
for our non-academic partners:
- if a network of academic centers decided to band together to offer a course in ‘healthcare data sciences’, how would you like to be involved? could you provide data to sites that need that? provide good clinical questions that impact public health and require reliable evidence? be a guest lecturer in the shared videos or on campus labs? provide funding for the effort required by faculty to make this vision a reality?
