Go Down

Topic: Recommendations Needed (Read 2325 times) previous topic - next topic

Cedricium

Hey guys, I need some advice for a project I'm thinking of doing.

My project would be a device that can diagnose whether a person is sick by taking several sensor readings. Using that data, the device then puts the data through some algorithms to determine the sickness.

I am completley new to the Arduino, but I think it will work for my project. The questions I have are which board should I use, and what other hardware (sheilds, sensors, etc.) should I use?

Thank you for your recommendations :)

strykeroz

Hi,

What symptoms are you looking for with this project?

Geoff
"There is no problem so bad you can't make it worse"
- retired astronaut Chris Hadfield

wildbill

How many sensors do you plan to read? That'll tell you which arduino is most appropriate

el_supremo

Which algorithms are you going to use?

Pete

focalist

Are you serious?

LOL.... yeah, no problem, we'll just design the whole thing for you, a device which would sell for hundreds of thousands of dollars and sell tens of thousands of units.  Especially since you have no sensor ideas, algorithm ideas, or anyting else, other than a vague "So, I think a medical diagnostic platform might be neat.."

Could you maybe provide a tiny bit more information?
When the testing is complete there will be... cake.

el_supremo

My next question would be - are you going to share the prize with everyone who helps you make this thing?
And why didn't you post this in Project Guidance?

Pete

focalist

In all seriousness, I think you really don't understand the scope of what you are proposing.. In addition, anything that does anything even vaguely diagnostic is subject to stringent testing and carry no small amount of liability.  What if we design it, it doesn't detect what you think it does, and a patient goes untreated.. or even dies?

One simple example is EKG.  For a time, I worked as an EKG analyst, reading what are known as Holter monitors, which are 24 hour recorded EKGs.  The data was prescreened by a computer running specialized proprietary hardware as well as some heavy duty software.  It took this computer (fast at the time) about 20 minutes to prescreen a data set.. Which yielded a roughly 80% FALSE POSITIVE for what would be potentially serious cardiac arrhythmias.  Even today's best devices, such as the Defibrillators now found in schools and such, have a huge error margin.. But basically the logic is that in that particular case, overreaction is the right reaction.  Implanted devices like implanted defibrillators have a significant margin of error.. And have a given level of risk of killing you by INDUCING arrhythmia.

I applaud the general thought.. But I suspect you are in far over your head here.
When the testing is complete there will be... cake.

dxw00d

Who knows, he could be the person that invents the medical tricorder.

el_supremo

Quote
Who knows, he could be the person that invents the medical tricorder.

That's why I asked about the sharing of the prize. I can't remember where I saw it, but fairly recently a rather substantial prize was offered for anyone who could implement a device which could diagnose (some specified?) illnesses non-invasively like the Star Trek tricorder.

Pete

robtillaart


First a warning, medical diagnosis is a very dificult task and a serious business. Don't expect to condense the 6+ year training of a doctor in 32KB.


Quote
My project would be a device that can diagnose whether a person is sick by taking several sensor readings. Using that data, the device then puts the data through some algorithms to determine the sickness.


This is typical a very interesting project with far too little information to give a meaningful answer (my opinion) So to get the requirements a bit right a few questions:
What is the goal of your project? a sellable device or a proof of concept?
Can you tell more about the sensors you have in mind?
Can you tell more about the algorithms you have in mind?
What are your ideas of the measurements? invasive, skin or remote?
What is your idea about the frequency of the measurements?
What is your idea about the duration of the process/analysis?
Can you define what sicknesses your device should be able to recognize? Or just yes / no?


Quote
I am completley new to the Arduino, but I think it will work for my project.

How can you draw that conclusion when you are completely new? Or it implies you have a reasonable elaborated design in mind ...?
Can you tell us more about your software and hardware skills ? What level? (biometric / medical skills)
What have you tried and investigated so far?
Do you have a team? What is their knowledge level? (also biometric / medical)


So far first round of questions, now some answers (to keep yin yang in balance ;) with some URL's to get started.


I would measure the following:
- temperature: e.g. DS18B20 is an often used sensor, although water resistant versions exist I don't recommend them for invasive use
  good is between 35C and 38C
  bad is below or above
  http://en.wikipedia.org/wiki/File:Body_Temp_Variation.png

- heartbeat rate:   no sensor comes in mind
  60-100 is OK
  http://www.heart.com/heart-rate-chart.html

- breathing rate:  air pressure sensor? or fast humidity sensor?
  deepness of breath should be taken into account, ==> total volume per minute.
  http://www.normalbreathing.com/index-nb.php

- breath analysis:   gas sensor(s) e.g. humidity outgoing air versus inhaling air -
  often used DHT11 / DHT22 are to slow for this.
  http://wiki.answers.com/Q/What_gases_do_you_breath_out

- bloodpressure: systolic and diabolic levels - no idea how to measure
  http://www.vaughns-1-pagers.com/medicine/blood-pressure.14.pdf

- conductivity of skin: => sweatiness - sick people often sweat more
  http://en.wikipedia.org/wiki/Skin_conductance
 
furthermore to interpret some of theses measurements you need to know if the patient is (fe)male, his or her age, known medical conditions e.g. allergies! eating habits, time of the day, current stress levels (e.g. due to work), ... For a lot of things there is no direct sensor only a list of Q&A.


After doing this quick scan I think you definitely can make a bio-measuring device with an Arduino but that it is not trivial to make a diagnose.
I certainly would not give another diagnosis than "all parameters are within normal limits" and "go see a doctor".

For esthetics I recommend a graphical screen to show all values at once. Finally most of the parameters mentioned above need to be monitored over a certain time (maybe even a day) so logging to an SD card should be considered.

Hopes this helps you to get started! (anyone can ;)
Succes!

< all disclaimers apply>
Rob Tillaart

Nederlandse sectie - http://arduino.cc/forum/index.php/board,77.0.html -
(Please do not PM for private consultancy)

Go Up