Your stipulation of "in a remote area" is the limiting function. I would guess that if a cellular signal is unavailable, there is even less of a chance that there's a reliable "community-operated" LoRaWAN signal from some random radio base station available. But maybe you are lucky.
However, if YOU can put up a LoRaWAN gateway that you control (manage, supply power, and internet connectivity), it can be stationed a few miles away, and you're almost done. Presumably you can also put up some agent that will be the endpoint for your LoRaWAN traffic and can actually submit the data to the (MongoDB) database. (This agent that gets the data from the LoRaWAN sensor could be co-located with your gateway and internet connection and can take care of morphing the LORA traffic to a TCP/IP message, with whatever security and endpoint you desire OR you could put it "in the cloud" with a LORA "carrier" and still have to deploy a cloud server to get it into your database)
Ultimately though, if you can't reliably get what you want by paying a carrier, you're probably not going to get it "for free" from the community. Data to/from remote areas is difficult/expensive for a reason (and your own LoRaWAN base station might be the cheapest solution if you can pull it off logistically.)
Personal experience includes using a LoRaWAN hat for a Raspberry Pi (which a sensor was connected to) and a Tektelic gateway. The LoRaWAN software stack for the Pi HAT was not very mature. I think you're on a more paved path with the STMicro solution. The Tektelic gateway was fine until it was provisioned for a private LoRa cloud. Conversely, it worked great when I had it on the office LAN and connecting to AWS directly.
