I used to develop solutions for clinical trial management. The problem set is similar to EMR. I noticed that the data schema requirements were incredibly flexib

I used to develop solutions for clinical trial management. The problem set is similar to EMR. I noticed that the data schema requirements were incredibly flexible by necessity. A health practitioner is basically gathering ad-hoc information from their patient. "How do you feel?" "Where does it hurt?" "When does it hurt?" The schema to store that information, when it's formalized, resembles the gramar rules for a spoken language, and is challenging to code, and even more challenging to query. I'm not saying this is why health care IT is so screwed up, but it's something that drove me away. There are standards for data storage, like HL7 and CDISC. Now, add on the fact that you have to formally test the living crap out of anything healthcare related, or people can die. The bureaucracy can be a lot for some people to put up with. It's difficult to attract good developers, when there are so many other fields to work in that are more free-wheeling. Also, healthcare IT is not very conducive to a lean startup culture, because of the domain expertise and labor involved to get a product out the door.