These are chat archives for nightscout/intend-to-bolus

11th
Mar 2015
diabeticgonewild
@diabeticgonewild
Mar 11 2015 01:08
@danamlewis and @scottleibrand , I am commenting your code as a favor for somebody and I need to know what treatment.glucose and treatment.glucoseType variables are. Is treatment.glucose a SMBG as in a finger prick?
Ben West
@bewest
Mar 11 2015 01:10
I would guess treatment.glucose is a rescue carb
glucose taken for treatment
diabeticgonewild
@diabeticgonewild
Mar 11 2015 01:11

Cool. Thank you But idk, what treatment.glucoseType is

Also, why is there a "+1" on this line? I haven't been able to figure that out yet.
var x = minAgo/5+1;

diabeticgonewild
@diabeticgonewild
Mar 11 2015 01:44
I don't think it is rescue carbs, because that variable is only used in one place, and is actually not used in any computations, treatment.glucose, that is. It must not be that important.
Scott Leibrand
@scottleibrand
Mar 11 2015 02:01
Monica: are you talking about in static/treatments.html ?
if so, that was some early draft code that is no longer maintained. I think we later standardized on treatment.carbs instead.
Jason Calabrese
@jasoncalabrese
Mar 11 2015 02:02
That's what you get from a care portal entry
Everything is a treatment, meal, snack, or rescue
Scott Leibrand
@scottleibrand
Mar 11 2015 02:03
oh, is treatment.glucose the current BG?
Jason Calabrese
@jasoncalabrese
Mar 11 2015 02:03
Yeah, for recording the meter value
Scott Leibrand
@scottleibrand
Mar 11 2015 02:04
ok, now I remember, yeah.
Jason Calabrese
@jasoncalabrese
Mar 11 2015 02:04
Not sure where you're commenting, but ideally it would be a PR targeting dev
diabeticgonewild
@diabeticgonewild
Mar 11 2015 02:04
That's what I thought. Thank you @scottleibrand and @jasoncalabrese !
I am not going to use that code. It's just a favor.
Scott Leibrand
@scottleibrand
Mar 11 2015 02:04
and glucoseType looks to be either "meter" or "sensor"
diabeticgonewild
@diabeticgonewild
Mar 11 2015 02:04
What do you mean specifically, @jasoncalabrese ?
Cool.
Scott Leibrand
@scottleibrand
Mar 11 2015 02:05
I think @jasoncalabrese is saying that if you're documenting the code, you could do it in the code itself, and then submit a PR to add the comments to dev
so in future it's easier for people to figure out what everything means
Jason Calabrese
@jasoncalabrese
Mar 11 2015 02:05
I know most people here are only using iob-cob, but I've been keeping that current with the mainline
diabeticgonewild
@diabeticgonewild
Mar 11 2015 02:05
I am going to sound dumb, but what's a PR?
Scott Leibrand
@scottleibrand
Mar 11 2015 02:06
pull request
diabeticgonewild
@diabeticgonewild
Mar 11 2015 02:06
of pull request
nevermind
Scott Leibrand
@scottleibrand
Mar 11 2015 02:06
that's how you submit code for inclusion in github
or, more specifically, propose to pull it from your fork to the mainline
diabeticgonewild
@diabeticgonewild
Mar 11 2015 02:06
yeah, I remember that. I don't know everything going on, as I am not familiar with JavaScript, Angular.js, or JSON stuff, but I am just doing a favor for somebody.
Like, I am bound to make errors.
Jason Calabrese
@jasoncalabrese
Mar 11 2015 02:07
My plan is the merge all the features of iob-cob to the mainline, but as multiple options that can be enabled and adjusted
diabeticgonewild
@diabeticgonewild
Mar 11 2015 02:07
Also, I am adding the Medtronic 530 IOB function.
That sounds awesome, Jason
Jason Calabrese
@jasoncalabrese
Mar 11 2015 02:07
Check out what I did with iob on Dev
Scott Leibrand
@scottleibrand
Mar 11 2015 02:07
if you are writing up something that might be useful to others, then submitting it as a PR serves two functions. In addition to helping others in future, if provides an easy way to have the rest of the devs check your work and make sure you understood everything correctly
Jason Calabrese
@jasoncalabrese
Mar 11 2015 02:07
Need to merge that to iob-cob
diabeticgonewild
@diabeticgonewild
Mar 11 2015 02:07
will do. checking it out right now.
OK, I will made sure to do that.
Jason Calabrese
@jasoncalabrese
Mar 11 2015 02:08
A single iob.js that gets used everywhere
Client and server
I'm going to do the same for other parts
diabeticgonewild
@diabeticgonewild
Mar 11 2015 02:09
very nice!
Jason Calabrese
@jasoncalabrese
Mar 11 2015 02:09
The AR2 is currently duplicated in the server and client also
diabeticgonewild
@diabeticgonewild
Mar 11 2015 02:09
:D
Jason Calabrese
@jasoncalabrese
Mar 11 2015 02:10
Then we can return the loss with /pebble
Lots of options
diabeticgonewild
@diabeticgonewild
Mar 11 2015 02:10
that is so awesome! :D Thank you!
Jason Calabrese
@jasoncalabrese
Mar 11 2015 02:10
Some of that could become a standalone module
Scott Leibrand
@scottleibrand
Mar 11 2015 02:10
we need a better name than "loss"
Jason Calabrese
@jasoncalabrese
Mar 11 2015 02:11
Cost
diabeticgonewild
@diabeticgonewild
Mar 11 2015 02:11
agreed
Scott Leibrand
@scottleibrand
Mar 11 2015 02:11
what is the definition of "loss" in this case? Is it the log of how far from target you are?
Jason Calabrese
@jasoncalabrese
Mar 11 2015 02:12
It's from Lane, there's a ppt that explains it great
Scott Leibrand
@scottleibrand
Mar 11 2015 02:12
I've seen his explanation of AR
Jason Calabrese
@jasoncalabrese
Mar 11 2015 02:12
Why it's called loss, etc
Yeah
For the different iob functions, it would be cool to select them, pick the one that matches your pump, etc
Maybe that is part of the profile
diabeticgonewild
@diabeticgonewild
Mar 11 2015 02:17
I have the for the Medtronic 530G, Medtronic 522, Tandem t:slim, and the Animas 2020
Like the curves fitted, as % IOB remaining. Obviously you would divide the result by 100 and multiply that by the initial bolus dose, in the simplest sense
Jason Calabrese
@jasoncalabrese
Mar 11 2015 02:18
Animas 2020 is what they call iob1 now? With the ping
diabeticgonewild
@diabeticgonewild
Mar 11 2015 02:18
No, the only Animas pump I have is the 2020 pump. The first pump with a color screen, LOL
Jason Calabrese
@jasoncalabrese
Mar 11 2015 02:20
Ping had iob1 and iob2
Think iob1 is a straight line and 2 is the normal curve
diabeticgonewild
@diabeticgonewild
Mar 11 2015 02:21
Oh, I didn't know that. Mine definitely is not a straight line for the 2020. As for the 4 pump % remaining IOB functions, check out the attachments for the posts from me on Feb 23. https://groups.google.com/forum/#!topic/openaps-dev/w_YNAmaguoc
I can explain it more, if you want.
The final post I made on that group about the equation that relates IOB to % remaining IOB is correct, and it could be potentially valuable, for understanding things, but who knows.
  • on the thread
Jason Calabrese
@jasoncalabrese
Mar 11 2015 02:30
So far approximate iob had been ok for me, so not matching the pump hasn't been a problem, but it would be nice if it could match
diabeticgonewild
@diabeticgonewild
Mar 11 2015 02:32

Yeah, this definitely matches, the pump. I confirmed my work.

Basically I did this, recording the IOB value at 5 minute intervals. The initial bolus dose was 10 units on all of the pumps. http://forums.childrenwithdiabetes.com/showthread.php?15364-Extrapolating-curvelinear-IOB-charts-from-Animas-and-Paradigm

I did this for 5 hours. Of course, for different DIAs you just use the same points and change the "time intervals" by stretching or contracting.
61 data points
Jason Calabrese
@jasoncalabrese
Mar 11 2015 02:33
I added some very basic tests, but would be nice to really expand them https://github.com/nightscout/cgm-remote-monitor/blob/dev/tests/iob.test.js
Scott Leibrand
@scottleibrand
Mar 11 2015 02:34
if you're not familiar with how tests work, they basically provide a set of input data, and expected output data, and make sure the code still returns the expected output
diabeticgonewild
@diabeticgonewild
Mar 11 2015 02:34
yeah, so it's like flags.
Scott Leibrand
@scottleibrand
Mar 11 2015 02:34
that way it's easy to tell if a code change unexpectedly changes the program's output, often somewhere you wouldn't expect
Jason Calabrese
@jasoncalabrese
Mar 11 2015 02:36
So you did all of that manually, Every 5 minutes?
diabeticgonewild
@diabeticgonewild
Mar 11 2015 02:36
Yeah, I try to be really careful with my work. I am almost always reluctant to trust it, unless I really, really, know my stuff backwards and forwards. But those tests look very useful. I really need to learn JS.
Yeah, I am a very patient person.
I sat for 5 hours, checking 4 pumps every 5 minutes
Not that exciting, but that's alright. I would definitely consider it worthwhile.
diabeticgonewild
@diabeticgonewild
Mar 11 2015 02:42

Like, this is what I am working on right now. The last 3 lines are of interest. It's the Medtronic 530G IOB function.

https://gist.github.com/diabeticgonewild/4616c456e01ed2634b2b

there are errors in my comments, too, at least by my books
I have a lot of work to do
Jason Calabrese
@jasoncalabrese
Mar 11 2015 02:44
That should fit in nicely
diabeticgonewild
@diabeticgonewild
Mar 11 2015 02:45
I would agree.
Jason Calabrese
@jasoncalabrese
Mar 11 2015 02:46
@scottleibrand what do you think about adding iobFunction or something to the profile? Then we just name the different functions and can let the user pick
Jim Sifferle
@jimsiff
Mar 11 2015 02:48
So that way you could choose an IOB curve that most closely matches your pump?
diabeticgonewild
@diabeticgonewild
Mar 11 2015 02:51
yeah, but we would need ones like the Animas Ping, Animas Vibe, Omnipod, Accu-Chek Aviva Combo, etc., too.
Just to make it complete, but I can be pretty compulsive.
Jason Calabrese
@jasoncalabrese
Mar 11 2015 02:53
Really cool
diabeticgonewild
@diabeticgonewild
Mar 11 2015 02:54
Thanks. I would agree. LD
:D
I used the MATLAB curve fitting toolbox, which is very robust. It's better than Excel because it has much more features and checks for adequate convergence on its own.
I get MATLAB for free as a student. I go back to school (online) in less than a week! I took the past 8 weeks off due to the blood clot in my lung...
MATLAB is my "native" programming language.
Jason Calabrese
@jasoncalabrese
Mar 11 2015 03:19
Lots of good reasons to track iob outside a pump too, like in the ping iob is lost if you change that battery
diabeticgonewild
@diabeticgonewild
Mar 11 2015 03:22
It's good to have for parents, to protect their children at school or whatever.
Plus, directly on the pebble, with the CGM value, you can't go wrong! :D
Scott Leibrand
@scottleibrand
Mar 11 2015 03:26
I'm fine with having multiple IOB curves, but I don't like the idea of just trying to match the pumps.
diabeticgonewild
@diabeticgonewild
Mar 11 2015 03:26
I did an Apidra function...
Scott Leibrand
@scottleibrand
Mar 11 2015 03:27
IOB curves are a feature of physiology, not a feature of the pumps, and we should be optimizing to match the user's physiology, not their pump settings.
IOW, the IOB curve should be optimized to
diabeticgonewild
@diabeticgonewild
Mar 11 2015 03:27
Well that is dependent on the dosage given an insulin absorption, and that is modeled typically by DiffEQs
Scott Leibrand
@scottleibrand
Mar 11 2015 03:28
make the predictions come out as closely as possible to reality
Hovorka uses a controller to find the "best fit" between all of those models, for the current "artificial pancreas"
Scott Leibrand
@scottleibrand
Mar 11 2015 03:28
I don't care how we model IOB, as long as the goal is to minimize predictive MARD or similar.
The goal should NOT be to match the pumps.
diabeticgonewild
@diabeticgonewild
Mar 11 2015 03:29
Well, I am doing things my way, with Hovorka's work. As for the Apidra curve, let me show you the file.
Scott Leibrand
@scottleibrand
Mar 11 2015 03:30
Except as a starting point to take advantage of the work they've all done. But we should be choosing the "best" pump IOB algorithm that best matches a patient and their insulin. Not the one that matches the pump they happen to have.
diabeticgonewild
@diabeticgonewild
Mar 11 2015 03:30
I would agree.
Scott Leibrand
@scottleibrand
Mar 11 2015 03:30
FWIW, I'm not arguing with you, Monica. I'm responding to Jason.
diabeticgonewild
@diabeticgonewild
Mar 11 2015 03:30
That's one of the reasons why I used more than 1 pump, besides, killing multiple birds with one stone.
Scott Leibrand
@scottleibrand
Mar 11 2015 03:30
yeah
diabeticgonewild
@diabeticgonewild
Mar 11 2015 03:30
Sorry if I sounded like I was disagreeing.
Scott Leibrand
@scottleibrand
Mar 11 2015 03:31
wasn't sure. Hard to tell when we're both typing things out at the same time. :-)
diabeticgonewild
@diabeticgonewild
Mar 11 2015 03:32
yeah, things like this can get intense anyways. Diabetes can really be a touchy subject.
I try to be careful, LOL!
Yeah, so for doing % remaining IOB for Apidra, Novolog, and Humalog, you would have to use the glucose infusion rate graph (which represents potential energy , like the voltage of a battery, which is comparable to the % IOB remaining). You would not use the plasma insulin levels (which represents (kinetic energy and the activity of that doesn't matter until the insulin binds on the cell and lets glucose in, representing a loss in potential energy. Think of plasma insulin levels as current in a circuit, which is just the flow of electrons).
I hope that makes sense.
Scott Leibrand
@scottleibrand
Mar 11 2015 03:42
I think so. But I think of it much more simply as "how much more is your BG going to drop from the insulin you've already injected?" (Divided by ISF of course)
and because we're using IOB and ISF to predict BG, we can see if the predictions come out right or not, and therefore figure out if the ISF or IOB curve is correct.
diabeticgonewild
@diabeticgonewild
Mar 11 2015 03:43
Which is potential energy. I just wanted to clarify with you. I made the mistake of using plasma insulin levels when interpolating the first time around.
Scott Leibrand
@scottleibrand
Mar 11 2015 03:44
Ya
diabeticgonewild
@diabeticgonewild
Mar 11 2015 03:44
Correct, and you can add the error from the predicted - actual CGM value to the current prediction, too.
Like as for emailing that Apidra stuff, I have to find the article with the graph I used. Then I will email you. I didn't take my work seriously, as I was just "playing around" and that's why I don't know where it is.
diabeticgonewild
@diabeticgonewild
Mar 11 2015 04:43
OK, I found the article. So I can email you now, but my work is not represented as a % IOB remaining curve, so let me update it so you get to compare it directly.
Jason Calabrese
@jasoncalabrese
Mar 11 2015 04:52
I agree @scottleibrand, it would be best to pick the iob function that best matches the actual response. Just by adding a few of these any letting the user pick will make it easy to try other things
we really need to way to plot the iob on the top of the chart
Scott Leibrand
@scottleibrand
Mar 11 2015 05:14
yeah, that might be interesting...
Dana Lewis
@danamlewis
Mar 11 2015 06:27
Fwiw, I was suggesting that a while back, but I think that got confused as an issue of care portal data points. Would love to have net IOB visualized at top of graph.
diabeticgonewild
@diabeticgonewild
Mar 11 2015 06:28
I have an idea. Why don't the basal profiles be pre-programmed into NS. And then the change in net IOB be represented on the same axis beside it?
Dana Lewis
@danamlewis
Mar 11 2015 06:30
It's hard when basal rates are changed, requires user to go back in and update. (Much easier if everyone had readability from their pump.) not a not-do-it, just something to keep in mind as a usability consideration
diabeticgonewild
@diabeticgonewild
Mar 11 2015 06:30
yeah, true.
But you could do total IOB, if there was a feature to enter in typical basal rates, and then put in temp basal rates into Care Portal, if you know what I mean.
Dana Lewis
@danamlewis
Mar 11 2015 06:32
Yea. We have an average baked into DIYPS, and temps are adjusted and calculated based on that. Good enough for me, but not sure everyone will feel that way until they see it in action
However, I've got minor variations in basals all day. Not sure for folks w wider variations how that would translate
diabeticgonewild
@diabeticgonewild
Mar 11 2015 06:33
Yeah, well I have that patience. IDK if most people do. I bet you and Scott are pretty patient people to develop the DIYPS
Patience to enter it into Care Portal
Dana Lewis
@danamlewis
Mar 11 2015 06:36
Hahaha yes. Building and doing at same time convinced me of worth of all of this. Positive reinforcement with good outcomes ftw :) but not sure if I only had care portal that I'd be as ok w data entry and everything else. Specifically built DIYPS with easy buttons for key actions, less clicks and less time spent
diabeticgonewild
@diabeticgonewild
Mar 11 2015 06:36
haha :D
You're awesome!
Dana Lewis
@danamlewis
Mar 11 2015 06:36
Care portal is great for a lot of things, but self entry of data consistently is hard regardless of tool
Ben West
@bewest
Mar 11 2015 06:37
yeah
I built a php web app in 2004 when I was diagnosed, and was alarmed at the wall of inputs I had built for myself
quickly realized I'd never actually use such a thing
Dana Lewis
@danamlewis
Mar 11 2015 06:38
Heh. I used to tell HCSM and mhealth conferences "guess how many diabetes apps I us
Ben West
@bewest
Mar 11 2015 06:38
by 2009 it was clear, I needed to work on protocol to get data in/out
Dana Lewis
@danamlewis
Mar 11 2015 06:38
*use.
answer before DIYPS was always zero.
Exactly.
diabeticgonewild
@diabeticgonewild
Mar 11 2015 06:38
same!
Ben West
@bewest
Mar 11 2015 06:39
I don't think app designers are always aware of what they are asking of people
"what kind of control does this present on user's time?"
Dana Lewis
@danamlewis
Mar 11 2015 06:40
Well, sonny vu of misfit was so proud of getting up and saying he never asked what patients wanted or needed. Ugh.
Was thinking about this when evaluating meters. I want a meter that gives me a a number, is accurate, and tiny.
I need zero additional features.
Ben West
@bewest
Mar 11 2015 06:41
devices are like pets... they require constant maintenance
"touch me I need power"
Dana Lewis
@danamlewis
Mar 11 2015 06:41
Heh, good analogy
Ben West
@bewest
Mar 11 2015 06:41
"touch me it's time for your blood sacrifice"
Dana Lewis
@danamlewis
Mar 11 2015 06:41
Does that mean my APS is like my pet rock in my pocket? :p
Ben West
@bewest
Mar 11 2015 06:42
what happens when your pet pancreas misbehaves?
hehe
sorry my pancreas ate my homework
Dana Lewis
@danamlewis
Mar 11 2015 06:42
:)
Ben West
@bewest
Mar 11 2015 06:42
common excuse for not sharing my starburst though
Scott Leibrand
@scottleibrand
Mar 11 2015 06:48
Usually it's "my pancreas wouldn't sleep, so I was up all night". More like an infant than a pet. ;-)
diabeticgonewild
@diabeticgonewild
Mar 11 2015 08:06
so what is activityContrib?
I am trying to plug everything in, and I guess idk what it is. I guess I need to relate units
I mean, just looking at it, it looks like the (potential) influence the IOB can have on BG, just relating units and stuff, but it's hard to tell what's going on.
Jason Calabrese
@jasoncalabrese
Mar 11 2015 08:11
yeah BG impact of the active insulin
I really need to goto bed
diabeticgonewild
@diabeticgonewild
Mar 11 2015 08:13
Well, good night. You should really get some sleep. Your health and well being comes first, though I have trouble believing that...
*trouble believing that myself
thanks.
Scott Leibrand
@scottleibrand
Mar 11 2015 08:15
Easiest way to figure out what a variable is is to look at how it's set. If it's still not clear, paste the line where we set it and someone here should know.
Jason Calabrese
@jasoncalabrese
Mar 11 2015 08:15
been waiting trying to get Andrew to a stable bg, at least I got to test alarms, error codes, and raw data...
diabeticgonewild
@diabeticgonewild
Mar 11 2015 08:15
couldn't activityContrib just be = iobContrib*sens;
true
activityContrib=senstreatment.insulin(2/dia/60-(minAgo-peak)2/dia/60/(60dia-peak));
I mean I know it's in mg/dL/min, I think
Scott Leibrand
@scottleibrand
Mar 11 2015 08:16
Heh, thanks gitter. :-)
Jason Calabrese
@jasoncalabrese
Mar 11 2015 08:16
need to use backticks
we should add the mmol conversion there now
Scott Leibrand
@scottleibrand
Mar 11 2015 08:17
So ya, that should be the BG impact from insulin activity.
diabeticgonewild
@diabeticgonewild
Mar 11 2015 08:18
so couldn't you just do activityContrib = iobContrib*sens; "
?
Scott Leibrand
@scottleibrand
Mar 11 2015 08:18
Not quite the same as ionContrib * sens
iobContrib is the integral of future activity.
Activitycontrib is the current activity.
diabeticgonewild
@diabeticgonewild
Mar 11 2015 08:19
like so it would be mg/dL/min, so you could just do iobContrib*sens/60 ?
Scott Leibrand
@scottleibrand
Mar 11 2015 08:19
or in your analogy, potential vs kinetic energy
diabeticgonewild
@diabeticgonewild
Mar 11 2015 08:20
that's what I thought
so I need to take the derivative of my equation and go from there.
Scott Leibrand
@scottleibrand
Mar 11 2015 08:20
the units might work that way, but they're measuring different things.
I'd have to sit down at a computer to work out all the details. On my phone in bed right now. :-)
diabeticgonewild
@diabeticgonewild
Mar 11 2015 08:21
Oh, then go to bed, and good night!
Thank you.
Scott Leibrand
@scottleibrand
Mar 11 2015 08:22
Heh, haven't managed to get to sleep yet... Guess I'm not sleep deprived. ;-)
diabeticgonewild
@diabeticgonewild
Mar 11 2015 08:22
LOL sleep? what's that?
Scott Leibrand
@scottleibrand
Mar 11 2015 08:22
I get a lot of sleep sometimes. 9h a night for the last few. :-)
so my body is all "sleep? Who needs that?" ;-)
diabeticgonewild
@diabeticgonewild
Mar 11 2015 08:23
man, like this class that I have to take over this term (due to non-academic reasons--from medical withdrawals at my old school)...I was so behind that I studied 60 hours straight non-stop
and then went into the exam.
Scott Leibrand
@scottleibrand
Mar 11 2015 08:23
i pulled an all-nighter. Once. Never did that again. :-)
diabeticgonewild
@diabeticgonewild
Mar 11 2015 08:24
Got a 91. Then got on an airplane to the Society of Women Engineers national conference in Chicago that afternoon, interviewed with Intel, got an internship offer. But you know I had to decline it.
Oh, I have done multiple 2 all nighters in a row....haha
Scott Leibrand
@scottleibrand
Mar 11 2015 08:24
if I'm well rested I can function on just a few hours, but not zero.
Anyway, time to try again. Ttyl8r
diabeticgonewild
@diabeticgonewild
Mar 11 2015 08:25
ttyl
jstrishak
@jstrishak
Mar 11 2015 18:14
Hello All!
This is Jon Strishak. Scott I sent you an email this morning introducing myself.
I've been T1 for over 20 years...currently on Minimed Pump and Dexcom CGM.
In San Francisco....
Dana mentioned you're in SF once in a while. Can you please give me a heads up when you're going to be here next? It would be great to meetup so me and Jessica (T1 for over 20 years, ex-Dexcom, ex-Abbot Diabetes) and learn about your work.
TuDiabetes.org is also in Berkeley. Are you aware of them and their website?
Dana Lewis
@danamlewis
Mar 11 2015 18:53
Yup, @jstrishak, was talking to Manny (president emeritus of DHF) this morning actually.
Scott Leibrand
@scottleibrand
Mar 11 2015 19:17
@jstrishak Glad to have you! I'll be down there @ Twitter HQ next 3/23-3/26. We could meet up for dinner Monday, Tuesday, or Wednesday if you'd like.
Ben West
@bewest
Mar 11 2015 21:31
I have a bunch of posts on tudiabetes :-)
I'm in SF too
Scott Leibrand
@scottleibrand
Mar 11 2015 21:46
yeah, ben and I usually meet up at least once each time I'm down in SF.
Ben West
@bewest
Mar 11 2015 21:47
:-)
jstrishak
@jstrishak
Mar 11 2015 23:38
Thanks @scottleibrand Let's plan to meetup on 3/24 or 3/25. I've got lots of friends at Twitter (engineers, sales, etc.). Also, my ex-girlfriend runs a theater nearby called the Cuttingball Theater and all Twitter folks paid out here entire audience to attended one of her plays a few months back. Please shoot me an email or text one week prior so we can set that up.
Nice to meet you @bewest !