While waiting to get the stage data working on my new Zeo EEG headband, I noticed a few interesting hikes in the BPM right before some electrical activity like in the minute below that could be a starting point for the heartbeat algorithm:
Download entry as Lucid Scribe Data (LSD) or Comma Separated Values (CSV).
LSDBase 
I built a wrist-worn Arduino with a finger tip heart rate detector (like a pulse oximeter without the oximeter) to monitor my heart rate while sleeping. I also used an accelerometer for body motion detection. My algorithm then looked for higher heart rate with no body motion to detect REM sleep. It worked well for detecting my REM sleep, but my stimuli (which ranged from small vibrational motor at finger tip or wrist to vibrating the mattress near my feet) only awakened me. The stimuli rarely manifested in my dreams before awakening me, and so I have never achieved lucidity with my device. I have been using the Zeo Personal Sleep Coach in my lucid dream experiments to verify the accuracy of my own Arduino device, and I really like it, but it needs a new headband sensor. Does anyone know where I can find headband sensor replacements for the Zeo?
Elite. I have had better luck with audio cues that get progressively louder… the trick is to keep background music playing so as not to trigger the novelty detector neurons. And if you do wake up, there is a good chance to get the hallucinations going again by just lying still and focusing on any dream fragments you can remember.
You could try making your own sensor: http://www.instructables.com/id/DIY-Replacement-Zeo-Sleep-Monitor-Headband-Sensor/. I might also have one floating around somewhere.