Re: example github

22 views

Skip to first unread message

Benjamin West

unread,

Apr 20, 2013, 12:30:35 PM4/20/13

to Elsa Kaminsky, Joyce Lee, MeDevice List, Peter Martini

Thanks!

Damon and Jana have mentioned they are interested in helping with the data format issues. The medevice projects list actually links to Damon's project http://thedamon.github.io/opendatabetes/ as the canonical place to discuss the data formatting issues.

I've tried to take some notes here: https://gist.github.com/bewest/4965993

It turns out there are some industry standard formats, and we basically need a standardized toolset to make it easy to transform the data into the standard formats.

Seems to hinge around encoding each metric with an industry standard code, and then there are a bunch of different serializations to choose from. There is a json ccda linter on github, which looks promising as it's maintained by the same guys doing Indivo and the SmartPlatforms stuff.

The wiki also has some discussion, and Bernard's diabetes format wiki on pbwiki also has a bunch of discussion on the formats in use, I'm not sure if all the fields have been inventoried nicely in one place anywhere. I noticed 4 separate time values, and I'm not sure how to represent thing like estimated unabsorbed doses of insulin on board at a moment in time. In addition, there are differences in basic bolus definitions between pumps; see Bolus Wizards by Zisser et al.

So medevice as an overview project links to the opendatabetes project, which is scoped tightly to the topic of data formats. Each link on the medevice page points to a project in a similar way... what I've tried to do by providing these links is to enumerate all of the technologies that diabetics are already customizing using open source techniques in order to improve their care. It turns out that there is such widespread interest that there is a project for nearly every kind of tool and technology out there. I think that's amazing.

The projects are somewhat organized by topic under each heading, but the headings may not make sense to people, and the list of "other projects" is mostly a simple search on github and google. Most of the source has been made public/written in the last two years, another aspect of all this I find incredible.

Clearly there are a lot of people motivated to do *something* in order to improve their care or the care for someone they care about. However, there is no forum that can make their contribution available for others to inspect, criticize, or reproduce. Medevice is a kind of place holder "github organization" in order to provide a place to weave the collective contributions of individuals so that their work in solving the problem of diabetes is not done in isolation. A rising tide lifts all ships.

-bewest

On Sat, Apr 20, 2013 at 8:53 AM, Elsa Kaminsky <elsa.k...@gmail.com> wrote:

Hi Ben,

This is awesome. I am going to go ahead and try to mock up another more useful version of the poster to serve as a map for parts of the "landing site" - totally love the Mondrian idea and am going to incorporate more tiling in the poster to simplify the questions / tasks and fit them more legibly, ideally on something that would fit on a page if it works as sort of an interface. In the shorter term, I can edit the poster a little bit just to add in topics like formats, security concerns, existing visuals, existing projects, etc. (gleaned from looking at medevice in part). And I'll definitely join the group!

I am also going to introduce you to Peter Martini, peterc...@gmail.com. We met today and talked about some issues related to data formats, actually, and played around with logging Medtronic pump to Carelink uploading, which I know you have done a lot with.

It sounds like it would be relevant to have a single topic page that unites the information on data formats in one place, to link to from a landing site of sorts. Do you suggest this one for formats?

Okay, talk soon!

Elsa

On Fri, Apr 19, 2013 at 2:41 PM, Benjamin West <bew...@gmail.com> wrote:

https://gist.github.com/bewest/5422223 - Here's how I might translate something like that.

Basically, all I did was use "markdown" to and add some "#" marks before each heading, which is what tells markdown that it's a heading.

If you fork https://github.com/medevice-users/diabetes#projects this project. You should be able to "add a file" using the + sign I think. You could add your proposal. After you fork, you can go to prose.io and edit files through their interface also. The README.markdown file has headings that roughly correspond to the kind of topical areas you were showing me.

I'd love to get a picture of your poster up on the web, even as it is. In the future, as you get used to editing content online, I will create a kind of topical infographic like your poster or like Mondrian or something, and we can use that as a kind of entry point. http://www.compositionwithjavascript.com/

<3 Mondrian. What would be really cool is to have the size and shape and quantity of tiles be generated from contribution stats to the various projects. A visualisation to document the vitality of the community, of sorts. Maybe too meta?

As far as setting a narrative or breaking things down into stages, what if we tell the narrative of trying to do everything we can do to improve "fidelity of care?" Step 1 is survey the resources and environment available to us, and begin attacking the problems we can immediately address... at the moment this includes all the things you mentioned, a community for people to build a static web app that can import simply csv and show some visualisations, projects to get data from devices, etc. Step 2 is to weave those disparate projects together so that everyone is benefitting from each other's work, and step 3 is to make new technologies, like a pure open insulin pump, and an open artificial pancreas, and open glucometers. Right now, those new devices are what I consider a "stage 3" kind of thing, while getting data off of meters and insulin pumps needs to be a stage 0 kind of thing. I wonder if that kind of timeline makes sense?

1.) Work with every technology available to better adapt it to current use.
2.) Validate crowd-sourcing medical technology works
3.) Create new open medical devices, force out of date vendors to adopt our safer technology.

-bewest

On Fri, Apr 19, 2013 at 10:59 AM, Elsa Kaminsky <elsa.k...@gmail.com> wrote:

Hey! awesome chatting with you too! Thanks :)

I'm getting into github at the moment and below I am attaching a visual sort of mind map, similar to the one I submit as part of the Sanofi submission (but more mapping questions rather than solutions, so better for this I think) and copy pasting what I have recorded as my Sanofi submission (kind of irritating that you cannot go back into your submission and see the exact final wording...but I think this should suffice for a decent overview).

Would love to hear your thoughts. Am going to meet with a data engineer from Bloomberg today who is part of a NY-based Meetup around Type 1 diabetes software and tools.

Email

elsa.k...@gmail.com

Phone Number

(804) 497-0692

Affiliations

Parsons the New School for Design, JDRF Central Virginia Chapter

Concept Name

Intelligent Information Platform for Type 1 Diabetes Self Care

Concept Headline

(15 words max)

Data design with patients in mind: integrating physiological and human factors for more intelligent care.

Concept Description

(100 words max)

We aim to generate, in a year or less, a meaningful improvement in the software used to track data, identify patterns, and make better decisions in the complex arena of self-management via CSII. By uniting and leveraging existing knowledge, technology and interests, we can go far beyond the capabilities of existing applications, and add value and rigor to data management for patients and stakeholders. By bringing together expertise, tools, and resources at our disposal today, we can improve care outcomes before the release of developing technologies (e.g., the artificial pancreas) and inform the development of care technologies going forward.

Concept Maturity

Early Prototype

Evidence-Based Health Outcomes

Describe how your concept provides an evidence-based way to improve the outcomes and/or experience of people living with diabetes in the US. Be specific about the data-driven improvement to outcomes and/or experience (250 words max)

Type 1 diabetes management is quantified via core metrics such as hbA1C, estimated average blood glucose, standard deviation or variability of blood glucose values over time, and frequency of acute hypoglycemia and emergency events. On the most basic level, we propose to assess the effectiveness of our intervention based on historical versus user tester scores on these metrics. In addition, we will evalute our project on metrics of quality of life, as judged by factors such as amount of time spent on care, efficiency and effectiveness of tracking, analysis, and representation work, and reduction of tradeoffs between lifestyle flexibility, therapy effort, and health status. By developing a fundamentally data-focused system that patients will use to track, analyze, identify patterns, and develop useful personal strategies through iterative interaction with their data and collaborate with their physicians, we will observe changes in these concrete healthcare metrics as well as more subjective measures (relating to quality of life and and usage experience versus existing solutions). More adequate models of not only the physiological but also human (lifestyle, psychological, interface) factors in self management using diabetes technology will also enable us to quantitatively and qualitatively assess usability. 70 percent of patients currently do not download their data, and the majority of patients on insulin pumps have hbA1Cs of 8.0% or higher. By creating a tool that enables users to see their data integrated with contextual information and obtain actionable insight into their disease, we believe there is clear opportunity for measurable improvement.

Decision-making

Based on your target audience, how does your concept enable better decision-making? What types of decisions does your concept inform? Across the spectrum of type 1 and/or type 2 diabetes (lifestyle and environmental factors to diagnosis, treatment, maintenance, and beyond), when does your concept make the greatest impact and how? (250 words max)

The target audience for this project can be broken down to three categories of beneficiaries. Those for whom our project will produce tangible benefits include: first, type 1 diabetics, primarily those using insulin pumps; second, their physicians, with whom patients need to collaborate to achieve care optimization; and third, researchers and device companies involved in in silico and outpatient trials who can benefit from greater understanding of human factors in type 1 management via technology. Our team aims to bridge the gaps most often occurring in software projects aimed at addressing issues in diabetes, such as a. lack of lead user (i.e., patient and clinicial) knowledge, b. inability to design and refine for the human user, c. outdated concepts of the user as having unlimited time and/or rationality, and d. crossfunctional yet narrow views of the problem at hand. We are able to bridge and identify, translate, and coordinate knowledge and expertise between relevant stakeholders including patients, doctors, researchers, device makers, charities, and independent experts. Our knowledge management protocols are key to this process.

DATA SCIENCE

How did you arrive at the creation of your concept from a design and development perspective? What analysis methods (e.g., baseline knowledge models, evidence-based practice, predictive analysis) and data sets (e.g., your own, government, specific industry) did you utilize or generate to manifest your concept? Are you combining data sets to create new methods of analysis? (250 words max)

We arrived at our concept through a. deep dives into personal experiences and datasets; b. large scale micronarrative research involving intensive mining of social media conversations as well as in-depth interviews with diverse stakeholders including friends, family, physicians, researchers, trial participants, and funding organizations; and c. quantitative data available from sources such as T1d Exchange, key technology presentations and meetings, and literature. Additionally, we have consulted and learned from prototypes of data tracking and analysis applications, information visualization generated through statistical coding, information design, and user and task analysis. For our concept, we are using patient data sets of our network of prototype testers collected through their devices as well as via self-tracking tools we are including as part of the platform. Ultimately, we hope to proceed, in a stepwise fashion, from integrating device data and contextual lifestyle and health data in the first generation, to integrating data from other applications of the users’ choice, from bike computers, to fit bits, to nutrition-tracking apps, and finally, to prototype a system for portable information upload and mobile access to realtime data and self-experimentation.

On Thu, Apr 18, 2013 at 3:20 PM, Benjamin West <bew...@gmail.com> wrote:

Really cool chatting with you!

-bewest

---------- Forwarded message ----------
From: Benjamin West <bew...@gmail.com>
Date: Thu, Apr 18, 2013 at 12:17 PM
Subject: example github
To: Elsa Kaminsky <elsa.k...@gmail.com>, "Lee, Joyce" <joy...@med.umich.edu>

Howdy,

Here's a terse example, will explain more later:

http://docs-v06.smartplatforms.org/reference/data_model/ - polished web page version

https://github.com/chb/smart-docs/blob/master/reference/data_model/index.md - github version

Anyone who edits the github version, or creates a fork on github and then edits that, is essentially suggesting a contribution to the website. This allows anyone to correct the prose which dominates most of the website.

The same thing is true of the laws, or anything else. You can edit the prose quite easily through prose.io or directly on github and then re-run the tool to get a new website.

Github makes it easy to discuss the changes being suggested -- this is actually where the "complexity" comes from.

Here's another example.

http://bewest.github.io/glucose-html5/wiki/Home/ - "web page" naive, not very polished experiment

The content came from here: https://github.com/medevice-users/diabetes/wiki - The wiki homepage.

The wiki can also go on the website.

-bewest

Reply all

Reply to author

Forward

0 new messages

Email	elsa.k...@gmail.com
Phone Number	(804) 497-0692
Affiliations	Parsons the New School for Design, JDRF Central Virginia Chapter
Concept Name	Intelligent Information Platform for Type 1 Diabetes Self Care
Concept Headline (15 words max)	Data design with patients in mind: integrating physiological and human factors for more intelligent care.
Concept Description (100 words max)	We aim to generate, in a year or less, a meaningful improvement in the software used to track data, identify patterns, and make better decisions in the complex arena of self-management via CSII. By uniting and leveraging existing knowledge, technology and interests, we can go far beyond the capabilities of existing applications, and add value and rigor to data management for patients and stakeholders. By bringing together expertise, tools, and resources at our disposal today, we can improve care outcomes before the release of developing technologies (e.g., the artificial pancreas) and inform the development of care technologies going forward.
Concept Maturity	Early Prototype
Evidence-Based Health Outcomes Describe how your concept provides an evidence-based way to improve the outcomes and/or experience of people living with diabetes in the US. Be specific about the data-driven improvement to outcomes and/or experience (250 words max)	Type 1 diabetes management is quantified via core metrics such as hbA1C, estimated average blood glucose, standard deviation or variability of blood glucose values over time, and frequency of acute hypoglycemia and emergency events. On the most basic level, we propose to assess the effectiveness of our intervention based on historical versus user tester scores on these metrics. In addition, we will evalute our project on metrics of quality of life, as judged by factors such as amount of time spent on care, efficiency and effectiveness of tracking, analysis, and representation work, and reduction of tradeoffs between lifestyle flexibility, therapy effort, and health status. By developing a fundamentally data-focused system that patients will use to track, analyze, identify patterns, and develop useful personal strategies through iterative interaction with their data and collaborate with their physicians, we will observe changes in these concrete healthcare metrics as well as more subjective measures (relating to quality of life and and usage experience versus existing solutions). More adequate models of not only the physiological but also human (lifestyle, psychological, interface) factors in self management using diabetes technology will also enable us to quantitatively and qualitatively assess usability. 70 percent of patients currently do not download their data, and the majority of patients on insulin pumps have hbA1Cs of 8.0% or higher. By creating a tool that enables users to see their data integrated with contextual information and obtain actionable insight into their disease, we believe there is clear opportunity for measurable improvement.
Decision-making Based on your target audience, how does your concept enable better decision-making? What types of decisions does your concept inform? Across the spectrum of type 1 and/or type 2 diabetes (lifestyle and environmental factors to diagnosis, treatment, maintenance, and beyond), when does your concept make the greatest impact and how? (250 words max)	The target audience for this project can be broken down to three categories of beneficiaries. Those for whom our project will produce tangible benefits include: first, type 1 diabetics, primarily those using insulin pumps; second, their physicians, with whom patients need to collaborate to achieve care optimization; and third, researchers and device companies involved in in silico and outpatient trials who can benefit from greater understanding of human factors in type 1 management via technology. Our team aims to bridge the gaps most often occurring in software projects aimed at addressing issues in diabetes, such as a. lack of lead user (i.e., patient and clinicial) knowledge, b. inability to design and refine for the human user, c. outdated concepts of the user as having unlimited time and/or rationality, and d. crossfunctional yet narrow views of the problem at hand. We are able to bridge and identify, translate, and coordinate knowledge and expertise between relevant stakeholders including patients, doctors, researchers, device makers, charities, and independent experts. Our knowledge management protocols are key to this process.
DATA SCIENCE How did you arrive at the creation of your concept from a design and development perspective? What analysis methods (e.g., baseline knowledge models, evidence-based practice, predictive analysis) and data sets (e.g., your own, government, specific industry) did you utilize or generate to manifest your concept? Are you combining data sets to create new methods of analysis? (250 words max)	We arrived at our concept through a. deep dives into personal experiences and datasets; b. large scale micronarrative research involving intensive mining of social media conversations as well as in-depth interviews with diverse stakeholders including friends, family, physicians, researchers, trial participants, and funding organizations; and c. quantitative data available from sources such as T1d Exchange, key technology presentations and meetings, and literature. Additionally, we have consulted and learned from prototypes of data tracking and analysis applications, information visualization generated through statistical coding, information design, and user and task analysis. For our concept, we are using patient data sets of our network of prototype testers collected through their devices as well as via self-tracking tools we are including as part of the platform. Ultimately, we hope to proceed, in a stepwise fashion, from integrating device data and contextual lifestyle and health data in the first generation, to integrating data from other applications of the users’ choice, from bike computers, to fit bits, to nutrition-tracking apps, and finally, to prototype a system for portable information upload and mobile access to realtime data and self-experimentation.