A stigmergic receptive field (SRF) is suggested as an improved variant of machine learning (ML). The authors successfully prove their proposed variant. They analyze heartbeat rate and accelerometer data coming from a smart watch to determine, via fuzzy analysis, coherence with sleep quality as reported by patients. So on one hand they have simple physiological data, and on the other subjective reports from patients; both are analyzed regarding how coherent they are with each other.

The advantage of SRF over traditional ML is that with less data but more dynamic interactions with new incoming data, it is possible to obtain some good agreement with reported sleep quality for many (albeit not all) patients.

I highly recommend this paper for sleep researchers looking for simpler, perhaps more enriching alternatives, and for computer scientists looking to avoid the burden of computationally demanding ML.