View Full Version : No tech skills but an idea for a community project

06-30-2014, 04:54 AM

I'm thinking (with the help of a developer and maybe the Nevron community) to try to build something to help an epileptic and autistic child. So I wondered if any of you could answer some questions which my developer has asked about Nevron?

Basically my friends have a 7 year old son who is diagnosed with autism and epilepsy and unable to communicate with him as he is unable to talk or write in any way.
Over the past year they have almost lost the boy 3 times, as when he has a seizure he can stop breathing, on average they call 999 around once a week.

Because of this my friends are taking shifts to be with him in the evening to watch over him as he sleeps incase he lapses into a life threatening seizure, as you can imagine this is exhausting for them.

So I wondered if it would be at all possible to set up a Kinect or Asus Xtion camera to watch over the child when he sleeps and to eventually get to a stage to sound an alarm if he has a seizure?

My developer believes he can help and we wondered if anyone else who is more competent with the software could help advise or maybe more?
Is there a Developer version of the software where we can explore any of the following?
At the moment we wonder if there was any way of extracting the data from Nevron which can be analysed from limbs and joints?
If so can the data be brought into and understood by anything like Action Script or any other programming language?

Any help or advice would be gratefully received

Many thanks in advance

06-30-2014, 04:58 PM
This seems to be an ambitious approach and may reap further rewards, but a less intensive alternative would be to get a wireless/wired IP camera that has motion detection. We have three Tenvis IPRobot3 2014 cameras for our Bengal and Maine Coon rescues (http://bengalmania.org/video/kittenclosure-ip-camera-streams/), and strays that happen by, and they record stills and video to an FTP server (a USB flash drive on our R7000 wireless router) when anything triggers the hot areas. The sensitivity and target areas (4) are adjustable and there is an IOS Alarm Message Push that can be enabled for when the motion alarm goes off. They have Infrared LEDs for night vision, a microphone and speaker so you can monitor/communicate and are Pan/Tilt/Zoom units, so you can look around; with 8 vector presets to skew to specific targets. Then get IP Cam Viewer Pro for your IOS and/or Android devices. It can also record motion detected events and set off an alarm on the device. Both the cameras and the app can send email.

The added bonus is that the app allows remote viewing over the Internet when away. Requires knowledge of port-forwarding to get through the firewall and maintain security through username/password to the camera.

$70.00 you're in business.

07-02-2014, 02:22 AM
Hi Spherical
Thank you for taking the time to reply.

Is what you're suggesting a motion detection system, which fires off when it identifies any kind of movement or a specific type?
Could you help me understand if what your suggesting is able to identify specific movements that the child will make, as I'm guessing he'll toss and turn as he sleeps. So what we'd like in an ideal world is to be able to sound an alarm when he has a seizure as opposed to general movement, which is why we thought of using something that can read limb movement.

Kind regards

07-02-2014, 01:50 PM
Yeah... no... sorry to say, but, terrible idea.

First off, having an "inbetween" like nevron would be utterly pointless, youd want to build a custom app which ran directly off the kinect SDK. Second... getting a computer to understand the difference between a seizure, or just general wriggling motions simply by looking at body motion... VERY difficult. Third... kinect gives pretty "shaky" motion from its raw output anyway... so even if you overcame point 2, a mannequin *could* read like it was shaking.

What you'd want to look into would be perhaps something along this line... http://www.extremetech.com/computing/159309-mit-researchers-measure-your-pulse-detect-heart-abnormalities-with-smartphone-camera ...using vision to detect heart rate changes which could potentially be interpreted correctly... but more ideally, youd need a system based around something like this... http://emotiv.com/eeg/conduct_eeg.php modified into some kind of cap that could be worn during sleep, and using the research sdk to build a system that could interpret the result.

07-03-2014, 04:59 AM
What RebelHill said. As I voiced, what you are proposing is ambitious at the very least; especially for a community project.

The cameras I suggest do not detect specific types of movement, no. Not a lot that is easily affordable will. You need to understand that it is not an easy thing to discern between this is "normal" and this is "bad". In any system there will always be false positives. How much you want to pay for a given level of fewer false positives is up to you. I'd take more false positives and have a system that can be working tomorrow for $70 than invest a magnitude more than that and not have it work all that much better or wait for a system that does someday.

Bear in mind, the camera motion detection threshold is adjustable. There are four adjustable area boxes that can be specified, each having their own threshold. So, you would probably be able, over time, to tune the settings to decide whether it is rolling over or uncontrolled gyrating. They're not perfect, by any means—they don't cost that much. Still, they would lighten the monitoring load by a significant amount and allow everyone to get more sleep. The remote monitoring capability alone is worth the price.

07-07-2014, 01:50 PM
..ok thanks for your input, I'll look to other methods to help out

07-07-2014, 02:04 PM
Might suggest a deep brain stimulus implant if they haven't already looked into that. Have no idea what the feasibility is of that or cost.

07-07-2014, 03:10 PM
Instead of Kinect and the like, I'd explore detecting through an attachment (like a band) or device on the kid. Those would be way more reliable...assuming of course the kid would keep it. E.g., see the Android app https://play.google.com/store/apps/details?id=com.rd.epdetect&hl=en.

07-08-2014, 12:29 AM
Instead of trying to do it visually, what about some pressure sensors under the bed, or stress/strain sensors between bed and wall to detect movement of the bed as a result of a seizure?


07-18-2014, 12:02 PM
We are currently looking at activity wrist bracelets which may help.
His seizures vary from violent which cause his arms and legs to convulse with about 10-15cm in distance to his legs which can move quite about the same amount but from his hips.
Harder still, his gentle seizures look like 'tired eye blinks' when he appears to be nodding off, but is in fact in a spasm.

Thank you for all of your ideas and thoughts I shall check back from time to time to see if anything else has sprung to your minds


07-19-2014, 12:01 AM
Harder still, his gentle seizures look like 'tired eye blinks' when he appears to be nodding off, but is in fact in a spasm.

Get a couple of the cameras I pointed you to and set them so that they cover the head/shoulders area from two vectors and turn on Motion Detection. Then set another camera, so that it sees the whole body and do the same; this time setting each of the four target areas so that they focus on the body sections and not the whole frame.

08-26-2014, 05:49 PM
Checkout this website and their MYO armband it reads the elictrical impulses in the arm and can send those impulses to a computer for analysis and the SDK is open for developement. If you have some programers on your team who knows what you might be able to come up with. And if your team succeeded it would be a Godsend to many thousands of other parents and familys with children afflicted with this problem.


08-29-2014, 08:20 PM
Folks, with all due respect, most have noted that Nevron wouldn't be of particular value in this scenario. Continuing this discussion in a Nevron forum seems counter-productive. Wouldn't it make more sense to move the discussion to a more appropriate forum?