Yeah, there would haven't to be some sort of correction from "temperature measured at the wrist" to "estimated core temperature" using some fancy models or ML. I have a feeling there hasn't been a lot of research done by Garmin into doing this.