# Smart bracelet for habit breaking

#### 001FJ

##### New Member
Hello!

I need help for a cool project idea!

I am a mechanical engineer based in Toronto and I have an idea for a “smart bracelet”. You see, I have trichotillomania (hair pulling disorder), and I want to make a smart bracelet that buzzes or vibrates or flashes when I lift my hand to pull my hair out. I want to make a smart bracelet similar to the Keen bracelet by HabitAware.

-The idea is to make a DIY smart bracelet that’s open to the public and won’t cost more than $50 to make yourself. But let’s not worry about this now because that’s the long term goal. -The bracelet will help bring awareness when people try to pull their hair out or bite their nails. The idea is that the buzz / vibration / flashing will condition them to be aware of their habit to help them stop. -I am a mechanical engineer and I have a 3D printer so I can do the mechanical aspect (housing, strap, etc) Again, that’s for the next stage, for now I need help with the electronics part. -I am handy with soldering and assembling things but I know very little about electronics / circuit boards / programming / sensor, etc. I can do it if someone guides me step by step. So please be patient with me! -For now I need to buy hardware: sensor(s), buzzer, LED light, battery, etc. And I need someone to guide me into making a schematic. Can someone help out with starting? -For now I prefer if there is not microprocessor / programming. I just want a simple circuit to test the idea. Any help or ideas are much appreciated! #### tvtech ##### Well-Known Member Most Helpful Member I honestly don't know if this is possible without using a Microprocessor because the thing needs to be trained to react..... #### DrG ##### Active Member This is not a trivial project. The "smart" aspects are an integral part of that product, but at some point, detection of the behavior does take place. I would ask you to consider some of what goes into a simpler fitness bracelet and a home-built bracelet. After a cursory look at the advert, it seems likely that there is a good deal of proprioceptive detection going on...and it is centered around the arm that is wearing the bracelet. Another words, you measure the position of the bracelet (as it is being worn)...maybe not just the position, but also the path to get to the position. Think about that...you are tracking hand/arm movements as well as possible and maybe including clenching or tensing of muscles...all dimensions you can think of. Next is the training. You have to identify the pattern involved and be able to discriminate that pattern from all others so as to specificlly target the behavior. Once you have generalized a "class" of patterns that fit the target behavior, you have to constantly track all behaviors and evaluate whether or not they belong in the class of the target behavior and then operate the vibrator (which is probably the only trivial part). Add to that the BLE aspects and now you get an idea of why they are priced as they are...and I don't know how well they work. Would they be able to detect the difference between nail biting and covering your mouth with your hand when you cough? To look into and understand accelerometers is necessary, but hardly sufficient and, frankly, I need someone to guide me into making a schematic. That is more than a tall order. For now I prefer if there is not microprocessor / programming. I just want a simple circuit to test the idea. I just do not see that happening. I hate to discourage you, but I think you need to get much deeper into the technical definition of the problem. To put it another way, you need to develop a switch that operates a buzzer, except this is not a switch that you press with your finger, it is one that is triggered by detection of the target behavior. Starting there, what sensors do you need to understand and incorporate into your "simple circuit"? Perhaps you could start with detecting when your arm is raised or lowered. I would suggest that you look at the two links that I provided and start thinking about all that is involved in those. Maybe, you could find the patent (if one was filed) for that device to get some background on what they are doing. Maybe research prosthetic devices and aids that are used in rehabilitation that might be detecting positions of limbs, even incorporating EMG...I don't know. It s certainly interesting and challenging. edited to fix link Last edited: #### Pommie ##### Well-Known Member Most Helpful Member Would a mercury switch and a vibrator motor work? Mike. #### narkeleptk ##### Active Member Pommie, wouldn't that vibrate when ever you move your arm around? #### DrG ##### Active Member Reading a little more, they may be using a gesture sensor (along with accelerometers) - that is actually a pretty cool way of doing it, I think. #### Pommie ##### Well-Known Member Most Helpful Member Pommie, wouldn't that vibrate when ever you move your arm around? So, when should it vibrate? I figured if your lower arm was vertical then you're either picking your nose/teeth, pulling hair, adjusting your glasses etc. How do you design a system that differentiates? Mike. #### tvtech ##### Well-Known Member Most Helpful Member So, when should it vibrate? I figured if your lower arm was vertical then you're either picking your nose/teeth, pulling hair, adjusting your glasses etc. How do you design a system that differentiates? Mike. With a system that can learn as in a Microprocessor based one. The thing fits on your wrist. No levers and cogs and etc. #### DrG ##### Active Member So, when should it vibrate? I figured if your lower arm was vertical then you're either picking your nose/teeth, pulling hair, adjusting your glasses etc. How do you design a system that differentiates? Mike. That is the crux of the biscuit. So, a gesture sensor + an accelerometer + ? + training. Something like that, but how well it discriminates is unclear. #### 001FJ ##### New Member Thank you for all the replies! You all have good ideas. If I have to incorporate a microprocessor then so be it. After reading further online it seems that the smart bracelets out there can't differentiate between a bad habit or just steering the car. So here is an idea: the bracelet has two parts. The main part is the bracelet, and the other part acts as an activator. The activator can be installed on the collar as a pin, or attached to the glasses. Whenever the bracelet gets within range of the activator then the bracelet warns (buzz / vibrates / flashing light). So it kind of works like a metal detector concept. Pro's of this concept: 1) The hand is free to go wherever it may except close to the head (e.g. no hair pulling) or mouth (e.g. no nail biting). You control where the limit of the hand is by where you place the Activator. 2) There is no learning necessary for the bracelet (easier to make b/c no programming or much easier programming than learning; and might be cheaper to make too). What do you think of this idea? Any ideas? Last edited: #### DrG ##### Active Member Thank you for all the replies! You all have good ideas. If I have to incorporate a microprocessor then so be it. After reading further online it seems that the smart bracelets out there can't differentiate between a bad habit or just steering the car. So here is an idea: the bracelet has two parts. The main part is the bracelet, and the other part acts as an activator. The activator can be installed on the collar as a pin, or attached to the glasses. Whenever the bracelet gets within range of the activator then the bracelet warns (buzz / vibrates / flashing light). So it kind of works like a metal detector concept. What do you think of this? Now, you are getting to the technical part of the problem definition. Look up hall effect sensor...but also realize how difficult it is to differentiate pulling your hair from touching your hair. Also, look into a gesture sensor because that could well be the first stage. Like I said, this is not trivial. #### 001FJ ##### New Member How about the activator be an RFID chip (no battery necessary), and the bracelet be the RFID sensor? #### tvtech ##### Well-Known Member Most Helpful Member Little sensors on each fingernail, plus a sensing band around your head. All talking to the wrist control HQ. Using high frequency signal, short distance technology. No antenna needed. OK. That's solved. #### 001FJ ##### New Member Little sensors on each fingernail, plus a sensing band around your head. All talking to the wrist control HQ. Using high frequency signal, short distance technology. No antenna needed. OK. That's solved. The idea was for the bracelet to be the sensor, and have a collar pin as an RFID chip. I don't think people want to wear a band on their head when they are at work, haha! #### DrG ##### Active Member Just a few thoughts... Simple proximity detection could be tried, but work that through....there are different ways to implement proximaty detection, but thinking it through, can you discriminate 2 inches from 1.5 inches or 6 inches? That is an important consideration. Additionally, proximity of the bracelet on your wrist to the sensor on your collar is not the same as detecting when you are pulling your hair. Would it be close enough to be of any value - would there be a large number of obnoxious false alarms? I remember an early behavior modification technique (please note that I am not, have never been, and will never be a clinician) where the technology was wearing a rubber band on your wrist. When you had an intrusive thought, you were supposed to snap the rubber band and the the brief, sharp pain was intended to harmlessly reset your brain, so to speak. Consider a matrix of IR detectors in the band and the complementary processing. A type of "gesture sensor", if you will. Mounted in a stationary location, you can detect a hand wave. With the proper sensors and processor and programming, you can wave your hand and it can discriminate a lower-left to upper-right wave, or a middle wave left-to-right or right-to-left. Those are available to everybody and relatively cheaply. I think, but I don't know, that they are using something like that, except the gesture sensor is on the wrist and the "wave" is relative to the stationary head. That and an accelerometer. That is what I think might be going on at the sensor end. Last week (while playing with an LCD and avoiding finishing an actuator project), I went and purchased a little camera for a Jetson Nano. This camera is off-the-shelf and ~$25 - I am sure you can get them cheaper. Running one of their demos (canned but you build it and can modify and so on), I took this picture less than an hour ago:

That blurry thing in the lower left is the camera. I am holding a remote control up to it and you see what the camera records on the screen. The software accurately identifies what it is (see what is written on the screen), AND, I can move the remote around in several positions and it still accurately identifies it as a remote.

It is an example of deep learning, AI, better known as "classifying things)...cue nsaspook to post the pic of blueberry muffins versus chihuahua puppies

What I am getting at is this: a camera on the wrist band. The targets are pictures, as, or shortly before, hair pulling or thumb sucking or whatever the behavior. Once trained, the software is constantly evaluating the camera output and comparing it to the targets and alerting when a set likelihood of the target is reached. Mind you, it might not discriminate holding your hand to your mouth when coughing from thumb-sucking but it could come mighty close.

Yeah, it is a lot of processing in a small amount of space and with a large power demand....but I am of the mind that proof of concept is a good place to start.

#### 001FJ

##### New Member
Just a few thoughts...

Simple proximity detection could be tried, but work that through....there are different ways to implement proximaty detection, but thinking it through, can you discriminate 2 inches from 1.5 inches or 6 inches? That is an important consideration. Additionally, proximity of the bracelet on your wrist to the sensor on your collar is not the same as detecting when you are pulling your hair. Would it be close enough to be of any value - would there be a large number of obnoxious false alarms?

I remember an early behavior modification technique (please note that I am not, have never been, and will never be a clinician) where the technology was wearing a rubber band on your wrist. When you had an intrusive thought, you were supposed to snap the rubber band and the the brief, sharp pain was intended to harmlessly reset your brain, so to speak.

Consider a matrix of IR detectors in the band and the complementary processing. A type of "gesture sensor", if you will. Mounted in a stationary location, you can detect a hand wave. With the proper sensors and processor and programming, you can wave your hand and it can discriminate a lower-left to upper-right wave, or a middle wave left-to-right or right-to-left. Those are available to everybody and relatively cheaply. I think, but I don't know, that they are using something like that, except the gesture sensor is on the wrist and the "wave" is relative to the stationary head. That and an accelerometer. That is what I think might be going on at the sensor end.

Last week (while playing with an LCD and avoiding finishing an actuator project), I went and purchased a little camera for a Jetson Nano. This camera is off-the-shelf and ~\$25 - I am sure you can get them cheaper. Running one of their demos (canned but you build it and can modify and so on), I took this picture less than an hour ago:

View attachment 119883

That blurry thing in the lower left is the camera. I am holding a remote control up to it and you see what the camera records on the screen. The software accurately identifies what it is (see what is written on the screen), AND, I can move the remote around in several positions and it still accurately identifies it as a remote.

It is an example of deep learning, AI, better known as "classifying things)...cue nsaspook to post the pic of blueberry muffins versus chihuahua puppies

What I am getting at is this: a camera on the wrist band. The targets are pictures, as, or shortly before, hair pulling or thumb sucking or whatever the behavior. Once trained, the software is constantly evaluating the camera output and comparing it to the targets and alerting when a set likelihood of the target is reached. Mind you, it might not discriminate holding your hand to your mouth when coughing from thumb-sucking but it could come mighty close.

Yeah, it is a lot of processing in a small amount of space and with a large power demand....but I am of the mind that proof of concept is a good place to start.
I like that! I have some friends who are decent at electronics so I will ask them, because I have no idea where to start!

The problem with cameras is that they need light to work. Driving at night is a very tempting time to pull one's hair. My wife is American and when we go see her family it's a 12hrs drive, so you can imagine how much hair is pulled out in 12hrs of night driving!

#### DrG

##### Active Member
I like that! I have some friends who are decent at electronics so I will ask them, because I have no idea where to start!

The problem with cameras is that they need light to work. Driving at night is a very tempting time to pull one's hair. My wife is American and when we go see her family it's a 12hrs drive, so you can imagine how much hair is pulled out in 12hrs of night driving!
IR camera

#### gophert

##### Well-Known Member
That is the crux of the biscuit.
Keep using phrases like that and you'll get the "'Super' Senior Man" moniker.

#### DrG

##### Active Member
Keep using phrases like that and you'll get the "'Super' Senior Man" moniker.
That phrase is from the song, Apostrophe, by F. Zappa. In Grad school, I once used the question, "What is the crux of the biscuit?" in an Freshman Intro Psych class that I had to teach. It was multiple-choice and the outcome was a greater-than-chance frequency of the "correct" answer. Of course, every one got the answer scored correct - for some reason the Freshman brain liked the idea and I would use it with some regularity even though it had almost no effect.

Now, you have to move on from the "senior man" obsession before that is all you can think about and someone finds you wandering the streets, barely clothed and muttering "senior man", "senior man" (reference Seinfeld, gym teacher, Costanza...can'tstandya)...I don't want to be seeing the Youtube phone video on Fox News!