Earlier this month Shiv and Ryan published a piece in the Annals of Internal Medicine, entitled What Can Medical Education Learn from Facebook and Netflix? We chose the title because, as medical students, we realized the tools our classmates are using to socialize and watch TV use more sophisticated algorithms than the tools we use to learn medicine.
What if the same mechanisms that Facebook and Netflix use—such as machine learning-based recommender systems, crowdsourcing, and intuitive interfaces—could transform how we educate our health care professionals?
For example, just as Amazon recommends products based on other items that customers have bought, we believe that supplementary resources such as questions, videos, images, mnemonics, references, and even real-life patient cases could be automatically recommended based on what students and professionals are learning in the classroom or seeing in the clinic.
That is one of the premises behind Osmosis, the flagship educational platform of Knowledge Diffusion, Shiv’s and Ryan’s startup. Osmosis uses data analytics and machine learning to deliver the best medical content to those trying to learn it, as efficiently as possible for the learner.
Since its launch in August, Osmosis has delivered over two million questions to more than 10,000 medical students around the world using a novel push notification system that syncs to student curricular schedules.
Osmosis is aggregating medical school curricula and extracurricular resources as well as generating a tremendous amount of data on student performance. The program uses adaptive algorithms and an intuitive interface to provide the best, most useful customized content to those trying to learn.
Currently research is published in journals with expensive subscriptions, which hides knowledge behind paywalls away from the public who paid for it.
How we publish research is broken, but it can be fixed and the Open Access Button is here to help. Open Access has the potential to give everyone free access to the world’s cutting edge research. This will help your doctor save lives, families get access to the latest research on their loved one’s disease, researchers advance our understanding of the world, and students better further their education.
Most people I work with in medicine have never heard of GitHub .
For the unfamiliar, GitHub is an online repository, which is an essential tool used by computer programmers to store their programming code. It has a number of virtues, including giving users the ability to track multiple versions of their code (sort of like remembering all the track changes you ever made to your word document). This is an essential tool for programmers but its value goes beyond its function as a track changes repository, as it is a site that facilitates open source collaboration, given its “social” features, similar to social networks like Facebook or Twitter, in which you follow the content of others or others follow you.
The most amazing thing about GitHub is that many users post their code (their work, their blood, sweat, and tears) publicly on their GitHub profile. Individuals will comment on others code, providing valuable input that the owner of the code can use to improve their work. In addition, can “fork” another person’s code repository, and work directly on the code in their own Github profile to make changes or improvements, sort of like a tag team collaboration. GitHub is the tool to help facilitate large-scale open source collaboration for the software/web programming world (such as that which lead to the Linux revolution).
By early 2012 there were apparently 1.2 million users hosting over 3.6 million repositories. Now that’s collaboration to scale!
So again, you may ask, why should physicians or medical researchers care about GitHub? Because it can have broader application beyond the software/web programming world, as shown by its use among non programmers, who are currently repurposing Github to advance collaborate in their own respective fields. They are posting book projects and transcripts of talks on the site, to encourage conversation and collaboration. One user even published his personal DNA information to encourage development of open-source DNA analysis. It has been suggested that Github could even be used by US citizens to “fork” the law so that they can propose their own amendments to their elected officials.
How might we use Github to democratize the world of medical research?
As researchers we do so many different activities that we perform in isolation, which forces us to “reinvent the wheel” constantly, from drafting of ethics board applications, to creation of research protocols, to the writing of snippets of statistical code or code for web programs.
Open publication adds to the flood of information already available to most doctors, placing a burden on them to search and filter it. IBM’s Watson is one famous attempt to approach the ideal where the doctor would be presented right at the point of care with exactly the information he or she needs to make a better decision. Elsewhere, I have reported on a proposal to help experts doctors filter and select the important information and provide it to their peers upon demand–a social networking approach to evidence-based medicine.
Not only published papers, but the data that led to those research results should be published online, to help researchers reproduce the results and build on them to make new discoveries. I report in an earlier article on this site about the work of Sage Bionetworks to get researchers to open their data. Of course, putting up raw data leaves many challenges: one has to be careful to deidentify it according to accepted standards. One has to explain the provenance of the data carefully: how it was collected and massaged (because data sets always require some culling and error-correction) so it can be understood and properly reused. Finally, combining different data sets is always difficult because they are collected under different conditions and with different assumptions.
What if you had access to all of the medical research in the world? Or better yet, what if the physician treating your particularly complex or rare condition had access to the latest research? Or what if a public health organization in your community could access that research to inform policymakers of measures to advance public health?
“Wait,” you may think, “can’t they already access that research? Doesn’t the Internet make that possible?” While unfortunately the answer to the first question is “No,” fortunately the Internet can make such access possible. As it is today, most physicians and public health professionals have very limited access to health research, almost all of which is published online. Only about a quarter of the research published today ends up being available to those working outside of universities, where libraries subscribe to a good proportion of the research journals.
So, what are these health professionals missing? What difference to their work would access to research make? Cheryl Holzmeyer, Lauren Maggio, Laura Moorhead and I seek to answer these questions with a new National Science Foundation study for which we are currently recruiting physicians and staff of public health NGOs.
We seek to demonstrate the difference it makes to the daily work of these health professionals to have easy electronic access to all the biomedical and public health research – or at least that large proportion held by Stanford University Library – for a period of eleven months (with one month of limited access as a control). To assess the impact of this access, we provide participants with a special portal to the research literature and track when and what research is viewed, while following up with interviews on the use and value of this access.