I have several installation that are measuring the AQI using a BME680 sensor with the BSEC2 library. We use the Arduino R4 Wifi board as a convenient platform, and gather the data at a central server. We are observing swings in the AQI reading from a nominal value into very high readings that correlate with changes in temperature. We are in north America, where heating is provided by a single furnace that circulates air through ductwork around the house. It appears that when the furnace is running, the AQI spikes, as it seems to correlate with change in temperature, and as little as 0.5C can make a significant difference. Is there some post processing that we need to do on the data to normalize the reading, or is this sensor picking up something we should know about? We have similar data in several locations, I have attached a sample from mine. The columns are a device identifier, timestamp, temperature, humidity, pressure and AQI.