Updates‎ > ‎

From old blog: Oct 9th - Midweek update

posted Feb 9, 2013, 7:40 PM by Dmitri Boulanov   [ updated Jun 16, 2013, 3:40 PM ]
1) Came across an inspiring article in the LA Times, originally found through YCombinator's Hacker News, which featured a number of applications using smartphones for medical / healthcare purposes. Cardiio, which I saw give an awesome presentation last week at Living Well Through Data conference in NYC, and Cellscope, which I came across last week while doing some research are featured for using the camera capability. Interesting developments for mHealth - there's plenty more to come - through RockHealth, HealthBox, BluePrint Health, and other non-accelerator sources.

2) Figured out the issue with the Node that we were having. It's actually a smartphone issue - my HTC Thunderbolt doesn't find the Bluetooth device (Android Gingerbread OS), while a newer Droid Razr (Android Ice Cream Sandwich OS) does find it and pairs with it very quickly. The device works like a charm once paired - the temperature reading is a bit off (reading 35 degrees Celcius for what should be 37.7 degrees Celcius for normal body temp), but I could see there being some calibration issues to be resolved - it's a prototype unit from what I understand. Thus, I'll need to look for either A) a new phone or B) an Android tablet (Kindle Fire maybe?), where I could test the latest code, as my phone is unsuitable/outdated after only a year. (this raises a lot of question about continuing to develop for Android as a platform).

3) Learned a bunch from the guys at the AT&T Mobile Hackathon. Ended up not participating, but working on the project a bit, as no-one thought the concept was cool enough (a lot of folk ended up going to try to work for the prize money offered for apps with Sphero). Two projects that stood out: a Kinect-based system for telling whether the patient is still in a hospital bed (posture alarm system) and an app for notifying a caregiver or provider about whether the patient is taking the right meds, using picture messaging / computer vision (which did not get developed). Tried out Parse and Apigee App Services for a backend, but was not impressed with the limitations. Was asked to try out AT&T backend for mobile health, but was not impressed either. I think we're going to go for Amazon Web Services (AWS) after all. Tried out the demo app - installed and run one on my Droid - worked like a charm (accepts inputs for a list of high scores, sends this data to the cloud (SimpleDB)), then retrieves it to show in the list).

4) Next step is to create a new Android app, euMetrica1 and integrate the TripNotify (from Hackanooga) code into it. Then hook it up with Amazon's SimpleDB to store some user info, based on the example app, and then try to hook up with the Node for temperature readings and try to store those. Next step after that is try to store fall events with the Node's accelerometer and send out SMS notifications based on those (can only do once I buy a new device).

5) Once we have a more developed front-end and back-end in the cloud for storage, we can start talking analytics on the back-end and what can we do with that. That is, analytics, complex event handling and any machine learning will most likely come as part of Development Round 2 development of this system. As well as the integration of the EKG device and heart signal readings, as they have yet to ship us the prototype device and necessary paperwork.

6) There's a lot of work to be done in general. This week, I hope we can get some sort of idea for how we're going to structure the initial data. Then, we could diagram out the back-end and diagram this data structure, so we can successfully iterate from there forth. I think that an Android developer and a back-end developer could really help here (one back-end developer is in talks with us now). I'll put up a form for people to leave their information to contact us in the About Us section. Then, there's the application, for which we have gathered a lot more relevant backround publications and info.

7) Thinking about a visit to a local assisting-living home community, where I could learn more about their needs. Interviews and user observation are a part of need gathering, which I'm learning about in my online Human-Computer Interaction class at Stanford (Coursera). Perhaps we could learn about user hacking, lead users or even extreme users of medical devices at these places and incorporate some concepts/hacks they utilize into our system.