I found the oncotherm patent of Szasz at
https://patents.google.com/patent/US9320911/en . They discuss a complex modulation involving some positive feedback from a sensor. I dont understand it all, but you could read it with the clue that the sensor unit is an SWR sensor. They mention 1/f pink noise. I am thinking the initial signal might be either clean sine or clean rectified sine ... and the shape of the mod signal is then further distorted or refined as a fourier sum (with some reference to the SWR feedback sensor input). It may be that the patent text does not clearly spell out the intention of the signal modification .... but it might be as simple as adding just enough odd harmonics (or pink number harmonics, whatever) to get optimal SWR, and it stops adding extra harmonics once this extra action makes SWR worse. If my guess is correct, there is no special clinical or physiological aspect to this signal shaping, but minimising carrier reflection might be clinically optimal anyway (or not). If the mod shape can impact on target impedance, then maybe that is like an electrical engineer's ecstatic experience of sacred geometry.
According to
https://www.rife.de/oncotherm---rife...erthermia.html they changed the faceplate label from MOD to RIFE, perhaps because the Rife story is so popular.... but probably the invention is originally unrelated to Rife. There is a typo in the rife.de text. 1/3 should read 1/f.
Why would Szasz play with modulation in the first place ? Maybe because it allows higher peak RF power while limiting heating (related to average power). In older texts he used the term fractal modulation, which I beleive means like flower petals, but it may have some more obscure mathematical meaning.