Interesting question. I will speculate without actual data for entertainment purposes. I would assume that a candle would produce more or less constant output

Interesting question. I will speculate without actual data for entertainment purposes. I would assume that a candle would produce more or less constant output in heat and CO2 and you likely will light it before having sex and not in the middle of it. I would also assume that two exercising humans will produce more output than a candle. So it would produce some bump in the baseline but you still would have a measurable spike in it. You could also distinguish the candle from the light bulb when you have CO2 reading. Of course it would be also better if you could execute controlled tests i.e. you could control the heating and possibly the ventilation. As the temperature sensor in the room is usually in the form of thermostat then the first one would be viable. A fireplace would deplete oxygen from the room but (unless one without exhaust) will not increase the room CO2 concentration. But it will likely introduce an additional airflow. So this would be more complicated case. But you likely could trick the users in the configuration phase to give you additional information to make strange cases more explainable. I.e. you could ask such harmless question like do you have a fireplace.