If you can point a modest-resolution (1+mp) camera at the patient's face, you can easily train a neural network or SVM to categorize "faces with eye(s) closed" and "faces with eye(s) open". As I recall, we did this in graduate school in the mid-90s so nowadays it should be simple - of course, assuming you don't need sub-millisecond timing, etc-- if you just want to know "eye closed at time T" where T is not super-exact, e.g. maybe down to the 1/10th second or something, I would just use a camera - noninvasive and you could do the processing on a laptop either using real code or a package like MATLAB.
Opening and closing an eye is a mechanical process so we can use the change in mechanical properties or movements as sensing signal to detect it whether an eye is opened or closed. But may i know this author (asked this question) why an what is the importance of such sensor and where we can use it ?
If you can point a modest-resolution (1+mp) camera at the patient's face, you can easily train a neural network or SVM to categorize "faces with eye(s) closed" and "faces with eye(s) open". As I recall, we did this in graduate school in the mid-90s so nowadays it should be simple - of course, assuming you don't need sub-millisecond timing, etc-- if you just want to know "eye closed at time T" where T is not super-exact, e.g. maybe down to the 1/10th second or something, I would just use a camera - noninvasive and you could do the processing on a laptop either using real code or a package like MATLAB.
Eye tracking camera's often can be used to extract information on blinks. If the head movements are within the eye tracker field, the missing data points are caused by blinks. Eye trackers sometimes can also calculate (variations in) pupil size. The specifications of an eye tracker should tell whether the above is supported. If not, you can also manually extract information about missing data points from the raw data.
Try NeuroSky MindWave. It's trivial, pseudo-eeg device, that certainly can't be used up to its purpose, but blinking detection is what it's best at. (it's because of the fact it uses only one electrode and it's located right over eyelids muscles). Pretty nice price and feasible SDK could make it exactly what you need.
You can use a camera.........detect eyes in real time with Haar-like classifiers (available in OPENCV Library). If Haar-like classifiers finds eyes then eyes are open but if Haar-like classifiers does not detect eyes, it means eyes are closed.