INTRO highlights from four panelists.
Geomedicine is emerging. Location data needed “to make medical records more enriched”. Environment is a critical player, environment is related to the geography significantly. New medical framework needed.
Place information to be brought into the story. Bring mobility into the story, as a location traces. Location over time (physical activity) matters. How I communicate chronical disease. Traces of daily lives. Pulling features out of daily traces. Looking for standardised platforms to use. Looking for mobilized innovations in healthcare.
Intersection of mobile activity data to deconstruct human genome why people do and how they do. Get data from telecoms, twitter, facebook and analyze the data. Track diseases from social and location networks. Incorporate into doctors’ diagnosis tools.
Tools for healthy behavior changes. How tools change daily behavior patterns. Places, environment, time etc matters. How to utilize technology in healthcare. Location data seems very attractive for unlocking hidden patterns in disease and people data.
APPLICATIONS of new approaches to healthcare.
Eating. If smbd prohibited to eat fast food, the app will discover smbd goes to Taco Bell and warns to not eat there. May be app will tell the doctor about that patient violating the rules of nutrition. Therapy tools. Tools keep track and make recommendations. Tool knows you spent 5h in shopping mall and how it affects insomnia. There are 20,000 toxic places in the states. 10M pounds chemicals per day produced… it is issue for environment. Public health. You are informed, you make decisions. Further: how we bring other specific points of data into the framework?
BARRIERS in geomedicine
Privacy in healthcare is a huge barrier. Overwhelm with information. He-he, BigData comes to healthcare:) Sensemaking out of the data. Profession of data scientist comes to healthcare. Ecosystem needed.
Location is an opportunity. In long term – understanding human activity at the level of context to encourage doing smth. Change people awareness, people attitude. Autonomous tracking of behavior patterns. Sensors do not really working. Where do you go? Whom you contact with? You can understand people, their health related behavior. E.g. if you eat in front of TV, how often do you eat home, how
fast do you eat, time of eating, gym facilities. Interventions are bound to the context – what exactly to push to the patient. In-door locations within hospitals and clinics. Geodistance of your day. Those things impact medication. analysis of daily and weekly traces give answers. Parkinson disease is location specific. Self-diagnosis via summarizing information. For medical tools: “one size does not fit all anymore”. Different personalities, research done.
conclusion for mHealth – Public Health, Disease Management, Mobile EMR confirmed and distinguished categories (from other ~10 categories) as mHealth solutions.
Twitter Geo Stack
@RAFFI Director of Platform Services from Twitter presented their geo stack.
at high-level Twitter’s geo infrastructure looks like this:
* Ruby in Rails frontend
* Scala-based “Rockdove” service using JTS
* Cassandra as the persistent store
* Thrift as RPC mechanism
There were details about the data format for Cassandra (place, edit and id), index nodes sharded by place id, Lucene used for search. There is Gazetter component interacting with Indexes. Gazetter is also connected to Cassandra.
my question was “what was a consideration to prefer Cassantra to MongoDB or HBase?”. The answer was “it is historical, at Twitter it Cassandra is historical. There are 5-6 Cassandra committers. So it’s historical.”