signal quality assessement of the pulse sensor

hi , i would like to implement a neural network on arduino , for the quality signal assessement of a Pulse sensor
but first i want to know how to assess the quality of the pulse sensor.
Thank you

Pulse of WHAT??

You need to tell us a lot more to get help...

i would like to implement a neural network on arduino

It's OK, you don't have to ask permission.

Based on his question I am guessing he want's to know if some "Pulse" sensor has a good enough quality for use with a neural network monitoring human heart rate ("Pulse") . [he used Caps for Pulse when most of us would have written "pulse" so I am inferring he thinks that "Pulse" means human circulation pulse whereas "pulse" is just something ordinary that we would talk about.

I want to assess the quality of a sensor ( ) it's an oxymetre with a neural network but first i want to know how i can find out what are the characteristics of a bad quality signal since i have to convert the analog signal to a digital signal .

As I said, heart pulse... We can't help you until you obtain and post a datasheet. The website doesn't have one readily available so it is obviously not a professionally developed product with extensive testing and datasheet like you would find from say for example Linear . My guess is it's piece of garbage. Find a sensor that comes with a real datasheet, not some fly by night hack job.\



thank you for the response , now can you tell me how to assess the quality of a periodic signal on arduino .

You have to state your criteria for assessment first. You cannot simply say " I want to assess the quality." You can say it but it is meaningless as far as any engineering or scientifice analysis is concerned. You have to say " I want to assess the quality for an accuracy within such and such range. If you know what you are asking then you must know you need an A to D . Look at the linked datasheet (by a real company) . State the criteria. Define your terms. “Quality” is not a quantitative word so I don’t know how you plan to assess “qualiity” if you don’t know what you are looling for. When you can state that , we can tell you how to look for it.
Do you plan to look at the data and say “Oh, ok, that looks ok…” ? Do you see my point ? “Quality” doesn’t tell us much. Do you know what you are looking for ? Are you a doctor or and engineer or scientist ? How do you know when the signal is good enough ?
You haven’t even stated what resolution of measurement you are looking for (8 bit,16 bit etc…)

Ok, the sensor is very sensitive to artifacts in daily life, like body movement. and before using the data from the sensor , i have to know if it's a good quality data or not that's all. ps : i have a bad level in english , sorry of i couldn't make my salf clear.

Ok, the sensor is very sensitive to artifacts in daily life, like body movement. i have to know if it's a good quality data or not that's all.

I don't have any problem with your English . I can understand you perfectly well I understand that even after reading my last post you fail to understand that you can ask the same question over and over until Lassie comes home and the answer will be the same when she does. What are your criteria for " good quality data " ?

Do you have a problem understanding that ?

the sensor is very sensitive to artifacts in daily life, like body movement.

Also, in case it hasn't occurred to you, when you go to the doctor and the nurse sticks the pulse oximeter on your finger and tells you not to move, the reason is that movement degrades the data so if you were planning to collect data with someone moving then you really can't expect any quality data , can you ?

what if the patient can't keep him self from moving once in a while , i mean you can't stay immobile all the time specialty if the recording will take a long time , the pulse sensor is connected to bedside monitor , i don't know what kind of information will help me know if the sensor is well connected to the finger , i thought that maybe there is a way to detect high frequencies in the signal , and put a threshold to say of it's a good signal or not. i don't know how to do it , and i hope you understand now . and sorry again. Thank you.

You can forget about detecting high frequencies in the signal because the people that design oximeters put a low pass filter in it. Patients who require 24-hour heart monitoring do not wear the finger sensor. They wear EKG pads on the chest with an RF pocket transmitter which includes a panic button. That's been the standard for the past 30 years.

Ok , can you help me find an other way to now that the pulse sensor isn't properly touching the finger. Thank you for your patience.

Sure. If you don’t read at least 40 pulses per minute, the patient is either dead or the sensor is not touching the finger. That’s pretty clearcut isn’t it ?

what if i'm assessing the quality of a 2ms recording .?

So what ? What does that have to do with anything?

hahah, thank you it was hard form me to understand what you were talking about , now i know: i'll calculate the time from peak to peak and when it shows a long time , less then a bradycardia (when you calculate the bpm) you can understand that there is a poor adhesion of the sensor to the skin :). This is probably not as common as tachycardia for anyone in a hospital. The only healthy people with a resting pulse under 40 bpm are long distance runners. (probably less than 10 in the whole world with a resting pulse that low) and they would never be using the product anyway so you can eliminate that scenario.

professional cyclist Miguel Indurain had a resting heart rate of 28 BPM).[2] Martin Brady holds the world record for the slowest heartbeat in a healthy human, with a heart rate measured in 2005 of just 27 bpm.[3]

As the above link states, less than 40 bpm is absolute bradycardia A waking heart rate below 40 BPM is considered absolute bradycardia. I think you can use 40 bpm as your alarm threshold.

Thank you for your time , and help !

Your welcome .