3 ms·

@breck - I'm working on some projects that are tangential to this on both sided and I'd love to hear your thoughts on connecting healthcare IT to bio-research.

@breck - I'm working on some projects that are tangential to this on both sided and I'd love to hear your thoughts on connecting healthcare IT to bio-research. What kind of healthcare data would you be looking to receive on the bio-research side? and what features of "modern software updates" are most important to you (regularly scheduled like MS Patch Tuesday? seamless in the background like SaaS products?)? do you think of this data as inputs to your research or living next to your research? Thanks!

breck 8y ago
Well a large 1M Homo sapiens multiomics dataset with 100+ clinical variables would be great. :) ... the data gets more and more interesting the more variables you have, but generally all the government datasets reduce number of variables for privacy reasons. That’s why if perhaps there was a way to get access to the quantity of data on something like GEO or TCGA, but with full clinical, from healthcare providers, you might find some interesting things. I would like to see more research as dashboards more than research as reports. The book “The Half-life of Facts” is relevant: many bio conclusions we’ve made in the past hundred years seems to later be contradicted or at least tweaked as new tech and data come out. NIH and others now have big pushes for personalized medicine, like NCIs Precision Medicine Initiative, because it’s pretty obvious that a great medicine or piece of advice for one person is useless or even harmful to another. The amount of health data we’ve collected in all of human history will be dwarfed by the amount of data we will collect in 2028. Instead of looking back on data and making static conclusions, what if we wrote the code that adjusts the conclusions as more and more data come in. So the research findings will be “updated” continuously, like modern software, as opposed to shipping once in the form of a paper.

breck 8y ago

Well a large 1M Homo sapiens multiomics dataset with 100+ clinical variables would be great. :) ... the data gets more and more interesting the more variables you have, but generally all the government datasets reduce number of variables for privacy reasons. That’s why if perhaps there was a way to get access to the quantity of data on something like GEO or TCGA, but with full clinical, from healthcare providers, you might find some interesting things. I would like to see more research as dashboards more than research as reports. The book “The Half-life of Facts” is relevant: many bio conclusions we’ve made in the past hundred years seems to later be contradicted or at least tweaked as new tech and data come out. NIH and others now have big pushes for personalized medicine, like NCIs Precision Medicine Initiative, because it’s pretty obvious that a great medicine or piece of advice for one person is useless or even harmful to another. The amount of health data we’ve collected in all of human history will be dwarfed by the amount of data we will collect in 2028. Instead of looking back on data and making static conclusions, what if we wrote the code that adjusts the conclusions as more and more data come in. So the research findings will be “updated” continuously, like modern software, as opposed to shipping once in the form of a paper.