Updates


Update for May

posted May 15, 2013, 6:05 PM by Dmitri Boulanov

There haven't been updates on this blog for a bit, but that doesn't mean the team hasn't been busy with euMetrica work.
At the moment we are working on:
  • Testing out the backend script that will run on EC2 to process the data - this has been tested and needs to be integrated into the entire flow
    • S3 and SQS read/write functions have been tested together; SNS has been tested separately - all work as expected
  • Incorporating the feedback of the cardiologist at Tufts Med reviewing this project. We got some good feedback in late April, including feedback to the Mozilla judges, which we plan to share in Chicago in June. To incorporate this feedback, we will need to test and have the best peak and trough detection algorithms we can establish. This work will be split from the development work and will most likely be developed in Matlab, before being ported to Java or Javascript. 
  • Requirements for a more basic UI have been completed and development is underway for a simple 6 screen demo version of the app, which should have less to explain to folk and better demonstrate the server functionality
  • To better demonstrate the graphing functionality, we will stick with only the static graphs from our previous demo. We have experienced some bugs and challenges with the dynamic graphs, so we are putting those on hold for now - placing it in future enhancement status.
  • We are awaiting the API for the spirometer hardware, which is ready to go, prototype wise. Once we have this and the new UI developed, we can easily make a call to the API to grab a measurement from the spirometer. Unfortunately, we do not believe that the spirometer hardware will be fully calibrated before June, but are still hopeful to have a demo at the US Ignite conference
  • We are to meet a couple more doctors, including a general physician and a pulmonologist, in the area, to get feedback on the demo and devices
In other news, I came across this article with a citation of Google's Android chief on future of the platform:
"For example, you are going to have computing which can potentially warn you before you have a heart attack."

Nice! 

Update for March

posted Apr 9, 2013, 11:24 PM by Dmitri Boulanov

As mentioned before, we have been keeping all of the communication in our team emails. So let me share a few updates for March.
  • A prototype of a mobile digital spirometer has been designed; tweaking and manufacturing is under way
  • Came across original research done at U. Pittsburg (please see updated Research section) which suggests significant value in monitoring vital signals real-time and notifying medical staff of any deteriorating condition per thresholds based on those signals (through the composition of a single index physiological variable) - some deterioration in wellness detected six hours in advance! Please check the papers for the setup and methods - of the research by Dr. Pinsky.
  • Peak detection script for EKG signals, HR calculation script completed in Python
    • Found issues with existing data, so will need to take additional readings, but script works fine
  • New back-end architecture is described by a diagram on the Back-end page; the backend stack is completely based in Amazon Web Services
    • Originally set up hosting on Heroku, but ditched in favor of all the available AWS functionality and interfaces (below)
    • SQS for a queue, S3 for storage, EC2 for processing (Linux, Python script) and SNS for notifications (SMS; platform independent)
    • Python interface to AWS - boto was used. Here is an example of a full Python interface to AWS services we used.
    • Complete end-to-end testing is in progress; Queue, peak-detection and SNS has been proven successful individually
    • Will likely switch to email notifications in future testing, considering the price for SMS. However, SMS arrives quicker than emails do.

  • Next iteration: front-end update requirements are being drafted - to be wrapped up by weekend of 4/12. We will be simplifying the EKG workflow.
  • Working on requirements for charting and data storage for the spirometer data (for COPD cases)
  • Working on specifying the alerts/thresholds that will set off notifications from SNS
  • Once complete end to end workflow is developed and tested (May), we can move on to creating a copy on iOS (June) and developing out a database to store all of the test information (June, July)
  • Once iOS workflows are available, we can do our first round of user and doctor testing (as docs tend to prefer iOS) this summer, along with doing a round of surveys during the testing, in order to get feedback on the functionality and get into the next round of server and UI development
  • Finally, we aim to keep an eye out for additional EKG and other (pulse oximeter using a Node) devices to integrate into the system
Please see the pitch for Mozilla Ignite Round 3 in the About section, for more info on our plans for Round 3 and beyond.

Healthcare Enterpreneurship on Coursera

posted Apr 9, 2013, 10:56 PM by Dmitri Boulanov   [ updated Apr 9, 2013, 10:57 PM ]

Me and Chris enrolled in this six week class on Healthcare Innovation / Entrepreneurship class on Coursera. Should be great for the project long term. Let's see how it goes!

Colbert Report! Smartphone sensors & Dr Topol

posted Apr 9, 2013, 10:54 PM by Dmitri Boulanov

Awesome video clip coverage of Dr. Topol of West Wireless Health Institute demonstrating a mobile EKG and other smartphone-linked medical devices on the Colbert Show. Stephen Colbert's character is pretty patient for his usual self. This kind of exposure should get folk more interested in exploring and adopting wireless health! Upvote!

Principles behind euMetrica coming up elsewhere!

posted Apr 9, 2013, 10:49 PM by Dmitri Boulanov   [ updated Apr 9, 2013, 10:50 PM ]

It's very reassuring to see some of the themes we mention while developing euMetrica coming up elsewhere. Means we're heading in the right direction - at least if the problems we thought were problems are actually being discussed by the users. Let's take a look!

"...While there are many applications for dedicated devices, Estrin said she is particularly interested in converting phone apps into data streams..."

Dr. Eric Topol diagnoses Atrial Fibrillation (current euMetrica focus!) on a flight, using the AliveCor mobile EKG reader - see the EKG graph here

Now, wouldn't it be great if you had one of the smartphone EKG devices and could connect to the WiFi network, sharing your data with a physician for analysis just like this, but all in real time over the net? That's one of our goals at euMetrica! (note: if there were other device players on the market, the devices could be made more affordable through competition too!)

On slide 27, a quote by Qualcomm staff mentions the lack of interoperability between current wireless medical devices and the inconvenience it causes for the caretakers and patients...Wouldn't it be great if I could see all of my body measurements in a single place? We're aiming for that one with euMetrica as well!


Key points from articles

posted Apr 9, 2013, 10:37 PM by Dmitri Boulanov   [ updated Apr 9, 2013, 10:38 PM ]

Your phone will know you're sick before you do
"...Usability is key: People want something intuitive and simple. Healthcare solutions must be designed with the user in mind, enabling people to arrive at actionable conclusions based on their personal data..."

"...Dr. Deborah Weiner, a pediatric emergency physician at Boston Children’s, created a mobile app for communications in the ED and a platform to help deliver laboratory values, diagnostic test results and other pertinent information back to doctors more quickly so they can make medical decisions earlier. A pilot of the app found that lab results came back 28 percent faster with the app than without it, according to Fried..."

"...The data gathered by these apps is analyzed, displayed and can be shared with physicians. And most of the apps even will suggest behavioral changes to improve test results...'This app revolution is going to enable more continuous health care,” Lai said. “ So we are keeping people healthy instead of treating them when they’re sick.' "

^^ This article above suggests that doctors may be part of a feedback loop for patients, using the apps as a connecting point. Hmm... interesting thought. But how about having the systems act as a closed loop or a physician's assistant commenting on your latest readings, much like a personal trainer in a gym?

Device ecosystem update

posted Apr 9, 2013, 10:27 PM by Dmitri Boulanov

It has been quite a bit since our last update - we have been using email as the main means for threads, correspondence and news. It's time some of that material got out to the blog. I will start off with a device update!

Breathometer - I wonder if this had an API, would it possibly be integrated into the system. Could future ones identify the health state of your stomach flora?

QMedic bracelets - could these be used to track vital signs and be worn decoratively? Will there be an API for these?

Sensiotec - very cool news. These guys are developing contactless (that's right! no touch!) sensors for measuring vital signs and digitizing the data. Could be a huge leap ahead for digital health. Would this have an API we could access? I wonder how big the device would be? Could it be minituarized? These guys are competing for the Tricoder prize and integrate data from other sensors... could we exchange ideas?

"Smartphone physical"! 
An entire array of smartphone-linked health devices is discussed in this TEDMED blog entry. It's obvious that individuals will be monitored, head to toe, soon. This is just the first batch of such devices. What capabilities are missing? How is the data stored? Can it be shared? What are some signals that are not being monitored and what are the hurdles to such functionality? Can euMetrica act to link all the data that these devices gather through their APIs? One app to sync with all of them and keep your body data all in one place? Can we make these cheaper using existing technologies?


Why euMetrica? CNN, RockHealth on Mobile Monitoring

posted Feb 27, 2013, 7:55 PM by Dmitri Boulanov

Maria brought up this CNN article a couple of days back - this is excellent coverage of the direction and angle we have been pursuing since August of last year. It's great to see that others are recognizing the potential of remote monitoring using mobiles coupled with sensor hardware as well. Key quotes from the article:

"...Healthcare solutions must be designed with the user in mind, enabling people to arrive at actionable conclusions based on their personal data..."

"...in order to ultimately take off, digital health services need to be designed so that they embed into current human behaviors. Devices that are uncomfortable to wear or don't provide simple explanations of data will not be adopted..."

A recent post in RockHealth's blog has also kept us inspired -- a post called "Ideas We'd Like to Fund", which explicitly covers two subject areas we are pursuing with euMetrica:

"Chronic disease management: Chronic conditions account for 75% of health spending in the Unites States. Better managing this population, particularly those at the highest level of risk for hospitalization, is the single largest opportunity to bend the cost curve...."

"Mega-platforms...We want a platform for sensors, to aggregate the data and make it meaningful for consumers and practitioners. We want a platform for digital health companies transforming care management in the preventive, acute, and chronic spaces to sell their clinically-validated outcomes to payers for distribution to consumers..."

Unrelated: Another post on what the team is working on (server-side code, peak-detection for arrhythmia and related alerts, messaging platform) soon...

From old blog: Oct 27th - The ecosystem - more players

posted Feb 9, 2013, 7:46 PM by Dmitri Boulanov   [ updated Feb 9, 2013, 7:46 PM ]

A quick update on the ecosystem. Remote monitoring for infants and for the elderly are the players in the remote-monitoring space represented here, each bringing a device and software (limited to that device or set of devices) into play:

Exmobaby

Rest Devices

BeClose

Each seems to have a developed product and a suite of software for it. I wonder if we can partner with each one of these ventures, in our quest for a unified monitoring and analytics system, which can integrate with each one of these sets of hardware, thus allowing the consumer to choose our system and one of the sensor equipment products that best suits their needs.

From old blog: Oct 21st - App Mockups and Update for Mozilla Ignite Round 1

posted Feb 9, 2013, 7:43 PM by Dmitri Boulanov   [ updated Jun 16, 2013, 3:38 PM ]

Posting an update on our work now. We completed several mockups, all of first draft nature.
  • Check out my prototype made in Justinmind prototyper here. You can really see the user flow for the main screens we seek to use - the ones included in the napkin sketch.
  • Dimitry's mockup in Fluid


I will be continuing to work in Fluid, as we can collaborate online and it's usable on Android/iPhone. Will post a version once it's done tonight.

As for the code, the Android version is available via GitHub - just download here and compile it in Eclipse.

Now, to address the questions made necessary by the Mozilla Ignite:

How's it going? How's your app progressing?

  • It's going well, although we could be advancing faster with development (that's if we had a bigger team and were doing this full-time). I feel we have done a good amount of research on what we're trying to do and are now in the development phase, exploring the process and prototyping several versions, before sticking with one (parallel iteration is the approach my Human-Interaction Class at Stanford emphasized as the winning approach to developing websites and apps). I also went to a Business Pitch competition at Boston University (Amr joined in at the end), where I found some find legal and business advice for our project. The judges emphasized verticals -- trying to stick with one use-case first (what's the population with a problem - tackle one well at a time, don't develop a poor general solution). We're supposed to get some business and legal mentorship during business hours at the School of Management at Boston University, which is cool. Will post more as results of the competition are announced
  • The development is going a bit slow, as we only have two part-time developers (me and Dimitry - with me performing administrative tasks and doing research and getting publicity - like the competitions and seeking advisors) - we could certainly use more Android developers. We have a few screens done and hope to wrap up the AWS (Amazon Web Services) integration tonight. I'll be pulling a few nights of coding to get our code looking like the mockups presented above. Check out a screenshot of a working graph that Dimitry sent to me here (we're using Android 4.0):

What's been done?

  • Administrative items, such as this website, domain, user accounts and the brainstorm round and development round applications, as well as applications to two pitch competitions for funding (to pay for the team and hardware expenses), as well contracts with hardware prototype vendors (obtained two hardware prototypes - for sensing body signals)
  • Attended Living Well Through Data conference (will post notes shortly) - learned about our competition and potential partners in physiological sensing technologies
  • Attended Hackanooga (organized by Kelly and Mozilla in Chattanooga, TN) in September - met with Mozilla organizers, other developers - got the experience of hacking a prototype for a fall-detection app over the course of 48 hours (see TripNotify), which got some press. We got some feedback from the locals and other devs, as well as the chance to mentor students at the local STEM high-school. Finally, we met with Dr. George Yu at Variable Technologies and agreed to borrow a prototype of his Node sensor device. We met Dr. Mina Sartipi (professor of Computer Science at Univ. Tennessee at Chattanooga), who later met with us and described to us the studies she's doing for a fall-detection system (our first focus for the system - an alert mechanism - for cell-phone and sensor based fall-detection for the elderly)
  • Attended AT&T Mobile Hackathon, in the hopes of learning more back-end technologies and finding more talent - pitched the idea there (got no more team members). Learned about Parse vs Apigee vs AWS for web-services (met with an Apigee rep, who demo'ed the functionality and spoke with a professional back-end developer). Picked AWS as the back-end technology.
  • Built out team to include a senior software dev, a biomedical engineer and three consultants with good Computer Science and engineering experience (well educated, experience in healthcare included)... need to find more front-end and back-end developers, as well as one more partner for the venture
  • Research into the first few use cases (asthmatics, diabetics, elderly populations, populations with heart conditions)
  • Feedback from a few doctors (check out the spirometer use-case, more coming), interview with an assisted living home CEO (why TripNotify (fall-detection app on a smartphone) won't succeed and what's actually needed instead), feedback from a few business professors (focus on verticals - check out COPD as a use-case, more coming), connected with a few medical students at Tufts Medical School recently (haven't met them for feedback / insights just yet)
  • Initial mockups on the web (see above) - available for feedback - to be provided to MDs and nurses for feedback in the future
  • Initial few screens in Android are done (open source on Github) - with graphing capabilities, Amazon Web Service integration, and Node (infrared thermometer and acceloremeter) and EKG integration coming very shortly (end of the week). We got the AWS (cloud back-end) and Node example apps working on an Android 4.0 phone (Samsung SIII Galaxy). Learning Android (with my two years of Java experience as a junior software engineer) and Amazon Web Services as we go - trying to find more frameworks.

 

What are some challenges you are experienced / have experienced?

  • Finding developers has been extremely tough. Now I understand what people speak/blog about when they say the opportunity/opening for a software developer versus developer available to be recruited ratio is about 10 to 1. Most developers who did express initial interest (for example, at the AT&T Mobile Hackathon at MIT) have instead joined other, more 'fun' projects (such as making a DJ game using Sphero (I did buy one) - the robotic ball and making a list of the hottest bars in Boston at any given moment using real time data). Fun as they are, I don't believe them to be as impactful in terms of changing the world for the better (go BU Engineering!) as a healthcare system, based on sensors, would be.
  • We are struggling with the team size, as I have been taking care of most of the administrative items and doing research, meeting with business / medical folk, and trying to develop all at the same time, while having a full-time job. It'd be helpful to have a larger team, if we are all part-time based, or raise enough funds to go full-time with the project. I would love to have more developers on-board, as well as a few front-end designers perhaps. Most of all, I'd like to find other passionate people to help me with the direction, brainstorming, and representation of the project (these would be partners of a venture).
  • Neither Amr nor myself have significant development experience, besides a semester of MATLAB class. I have two years of Java experience (a summer at a health-IT company where I did some Windows Forms in C#), but am just starting in Android and in front-end development. Amr has some C++ and a bit of Objective-C experience, but is also just learning - so the prototype we made at Hackanooga in 48 hours (with a bunch of hours spent mentoring the STEM students) was pretty good :) Dimitry brings more software development experience to the table, but I'm not as familiar with his background as he just joined the team and am hoping to get familiar with his skillbase first hand, as we go forth. Tunde, Adam, and Chris have contributed by consulting about a few items for us, but are not working on development at this point in the game (are more part-time consultants).
  • We are still defining the direction for the project, in terms of the use-cases, so it's tough to design and code for something general (the code we have now does exactly this), where we're not sure of who will be the target user. This should be alleviated shortly however, as we do more research and identify our first medical condition vertical. Once that's done, we can sketch out multiple mockups of a specific user flow with the help of designers, write out specifications, and get the developers working on a beta-ready system prototype
  • Our first prototype, TripNotify, when pitched an idea to actual potential users (an assisted living home) flunked, because of a lack of research about the users. It's a great learning experience, as we didn't invest much time and effort into the app - but I have to mark this down as a challenge, considering we totally missed the mark here - the elderly population in a living home is not likely to use a smartphone in this day and age - and their caregivers will not take the responsibility of using one to take medical measurements for them (that's a hospital's or nurse's job) - thus we can rule out some potential users right after this exploration
  • The development of this project has been organic, rather than focused. I believe that adding a focus (through business and healthIT mentorship we seek to gain) will get us growing much faster and help us develop a product in a direction aimed at a beta-test much quicker. This is a challenge / pain that should be alleviated fairly soon, as we gain Boston University (BU) and Mozilla mentorship.

What are you planning on tackling next?

  • Per advice of BU School of Management faculty, we will further research several use-cases for our system (COPD, asthmatics, other populations with chronic conditions) - figure out the problems each population faces, how much does that cost them and the healthcare system, and how we could digitally address those problems with our system and available hardware (and some hardware we could partner with manufacturers on). Continue development of the idea and prototypes, tailoring it to suit the feedback we're getting. Pick one use-case and develop a scalable (for future use-cases), yet specific system to address the problems that population is having with their medical conditions through a mobile, physiological sensor and Web interface - by providing preventative care (actionable items) and event notification capabilities
  • Complete the development of our first prototype, with the messaging, cloud-storage, multiple-device, threshold-based alerts, and separate doctor/patient logins (involving user authorization)... not all of these have to be included at once, but most of them should be - by the application for Round 2 in November. I'm forking what Dimitry has to date (a few screen we have done) and completing a portion of these
  • Find a business mentor at BU, find a biomedical engineering or electrical engineering (for physiological signal processing feedback) professor at BU for mentorship, find an MD for medical expertise consulting (that should be fairly easy in Boston - considering Mass General (the #1 hospital in the US for 2012) is here, as well as Harvard Medical School and countless other healthcare providers) and find some experienced Health-IT entrepreneurs to guide us through the ecosystem
  • Attract more talent: front-end and back-end developers (Android, Java, AWS). Another significant part (or full) time partner. Healthcare IT developers
  • Find additional graphing, cloud, computing and physiological related frameworks for Android
  • Work with Dr. George Yu (Variable Technologies) to get other sensors working on the Node. There is a serious possibility of partnering on the design a peak flow meter (spirometer) sensor for the Node, thus giving us a cheap future spirometer (way to measure how well the lungs are working) to integrate with our system (one which we provided the specifications for). CO2 and some existing optical sensors could be adapted for healthcare purposes with further research.
  • Get more feedback - from medical and healthcare community (schedule more meetings about use-cases), biomedical engineering community (on sensors - plenty of expertise at Boston University), healthcare IT (me and Amr both have some local contacts we are yet to utilize) and general IT communities (on available technologies and solutions to our challenges)
  • With all of the feedback, figure out where some computation heavy or high availability use-cases in healthcare might be (so we can utilize the full potential of the Gig!). Real-time preventative care will involve doing some heavy computation on real-time and historical data, providing actionable items to the caregiver and patients - users of our system - in the hopes of delivering better healthcare
  • Pitch the idea to the judges at Mozilla, folk at BU, and the wider business community (including healthcare organizations)!

1-10 of 24