• A better vitamin D tracker
    • Currently dminder is a bit slow to use, requiring 4 or 5 taps to do basic things such as track taking a 5000 IU vitamin D pill. It should be 2 (open app, tap 5000 IU)
    • dminder’s UI isn’t as useful as it could be. It devotes a lot of the front page UI to just what is the current sun phase what is your current estimated vitamin d level. I want quick entry / quick sun session button that is one tap, along with my most recent history in the front page along with smaller sun info.
    • dminder’s sun exposure model is you just sun bathing and has a literal flip your body reminder, I’d like a tracker that works with someone going for a walk.
    • dminder does not have good OS integration, it’s reminder does not let you say ‘yes I’ve taken my 5000 IU suppliment today’ directly on the notification, you have to go into the app to do it. No integration with apple health with outdoor activity such as runs, bike rides or walks, no quick short cuts from a long press, no widgets.
    • dminder’s UI is a bit ugly
    • Good practice swift UI & new iOS API practice app
  • Quick water & caffine tracker
    • this probably exists, but it’s a good sample app
  • No editing image taker for healthcare
    • Today’s smartphone cameras with computational photography apply too much AI skin smoothing and you cannot turn it off. This gets in the way of medical diagnosis.
    • This app would take RAW photos and also have a workflow to submit these pictures to doctors.
    • First iOS and then maybe android phones.
    • https://news.ycombinator.com/item?id=25436815
  • Using ML & a camera to tell you what your body fat % is
    • Hypothetically, by someone taking a mirror selfie, and a large training set of images, you should be able to tell what the body fat percentage of someone is with a picture.
    • This is an expensive dataset although, because you need expensive $10k equipment to be able to tell what someone’s bodyfat is, and labelling such a dataset is far more costly than asking people to tell you which squares contain a bicycle for free as a captcha mechanism or hiring random people to label which objects are what.
    • In a way, you need to bootstrap this by selling consumer 3d scanners like shapescale is to start getting a rich data set
    • There are a few apps that do it, but like <bodybarista.com> and <fitimage.io>, but I’m not sure if they are using ML or the lidar in the iphone front camera to make measurements, and bodybarista says it’s a beta that has 0.6 the accuracy of skinfold calipers.
    • There was another failed startup that also tried to do this about 6 years ago, but I can’t find them.