Thread #108614015
File: crying-sad.gif (63 KB)
63 KB GIF
>be me
>see cool contest with a chance to win the new pebble smartwatch
>write an entire open-source pharmacokinetic simulator for caffeine with raw-dog fixed point math
>the code is fire: lookup tables, 8term Taylor series, binary search using exponent's derivative
>actually model gut-to-blood transition, not just basic decay
>literally everything has a config entry
>run it 100% on-watch, zero idle battery drain, ideal pebble app
>pour my soul into the roadmap for future sensor fusion and closed-loop predictors
>62nd place out of 180
>most top places are generic watchfaces with a simple animation
>tfw you realize
You didn’t fail because the project was bad.
You failed because you posted it to the wrong audience.
Normies don’t reward:
effort
architecture
actual code literacy
It rewards:
spam
flashy colors
animated listings
37 RepliesView Thread
>>
>>
>>
>>
>>
>>
>>
>>
>>108614122
Pebble app store:
https://apps.repebble.com/halflife-smart-caffeine-tracker_abd6da93b862 465d836185e6
Github:
https://github.com/FPuzeras/HalfLife
>>
File: Aipepelaugh.gif (2.4 MB)
2.4 MB GIF
>>108614157
>HalfLife isn't just a counter; it’s a predictive tool
>>
>>
>>
>>
>>
>>108614314
>>108614288
All code was human written - README wasn't.
Can't be bothered to write the slop myself
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
>>
File: 1554067036317.jpg (27.1 KB)
27.1 KB JPG
you did all this for a chance at a 100 dollar chinkshit smartwatch that no doubt is selling all your data to thirdparties? What the fuck is wrong with you?
>>
>>
>>
>>108614796
Well that's where the HRV closed-loop model comes in. HRV is a really good indicator of caffeine, after the initial training period of 72h - 4weeks it gets really good accuracy according to studies, granted they used much better sensors...
>>
>>
File: dance-dancing-2488877032.gif (282.9 KB)
282.9 KB GIF
>>108614015
you did what you could, and no dirty fucking normie can take that away from you, anon.
>>
>>
>>108615611
Yes, that is the killer feature, you can tune it (manually on current devices, new ones will support the advanced heartrate data and i'll update the model).
But the existing apps pretty much just use population average metabolic rates (if even).
And getting the ability to customize on a low-power device like this was a pain. But fun to do (yes i know my brain is broken). And it works without compromise.
>>
>>
>>
>>
>>
File: 1726323927983.jpg (503.2 KB)
503.2 KB JPG
This shows exactly why i never bother creating an app myself. I have the technical skills but never come with an useful idea.
For example, you used mathematical modelling, low level and so on but whats the purpose here?
So i drink a cup of coffee and wait for my body to process it before i take another? but guess what, god already put a mechanism for that in our body. If i drink a cup, then another and feel weird... i wont be taking the 3rd. I dont need equations, a spyware, biodata and all that bullshit because i already have feelings.
The same for heart beat monitor. You can train cardio using heart beat frequencies, you can be autistic and use a spyware to know exactly you are holding 120bpm while running, also you can just use your feelings to estimate you effort/bpm.
These techs are pretty useless and
>HalfLife isn't just a counter; it’s a predictive tool
also remove this ai shit, it is corny as fuck