just googled more :) - https://en.wikipedia.org/wiki/Pulse_oximetry - an adapted for waterpool conditions earlobe oximeter is probably what is needed here (with probably a transmitter/amplifier with battery in the wristband). But currently it is in the $100+ territory.

It's not that simple unfortunately. One concern I have is that the oscillatory motion of the drowning response is well within the range of cardiac rates (1-3Hz from the video). Pulse oximetry data is quite noisy and is especially affected by motion affecting the sensing apparatus. Pulse Oximeters employ algorithms to attempt to reject noise (including noise from motion). Unfortunately these algorithms tend to depend on the cyclical nature of the heartbeat signal (using FFT or auto-correlation to latch onto the heartbeat signal, for example). The presence of an oscillatory confounding signal (noise induced from the motion of the drowning subject) would significantly reduce the reliability of these algorithms.

A device for alerting for a drowning subject would be classified as a medical device (in Europe most likely class IIb as it is measuring a vital function in a life support scenario) as the malfunction of the device could fail to prevent the saving of a life. The burden of certification of such a device is understandably high, and one would have to be able to demonstrate that the device can accurately and reliably measure in its intended use and show that any risk that it doesn't has been adequately reduced (I'm simplifying greatly).

Unfortunately it's basically impossible to test, after all how would one get ethical approval for exposing subjects to conditions of near-drowning, so one would be dependent on placing prototype devices on swimmers and catching episodes of near-drowning by chance. It would take a considerable length of time to gather sufficient data to confirm the reliability of the device under such conditions, obviously...

(and no, I would not consider it appropriate to test the device under general swimming conditions as these would not adequately predict performance under drowning conditions).

In terms of cost, an earlobe pulse oximeter doesn't have to be more expensive than a fingertip one (aside from some miniaturisation costs perhaps) but part of the costs of these devices is due to patent issues: http://www.law360.com/articles/106524/philips-accuses-pulse-...

That might be too expensive to hand out to everyone at the pool, but if it were workable for this application it would definitely be in range for something to sell directly to parents.

It wouldn't be able to interface with a device the lifeguard has in this case, but it could surely start making noise and/or flashing in an emergency.