Bayesian Sensor Configuration to aggregate motion sensors

I’ve been trying to create a couple bayesian sensors that use motion sensors and just can’t seem to figure it out. I’ve read through the previous posts and tried a bunch of things but haven’t gotten very far. Here is my configuration:

   - platform: bayesian
     name: "Upstairs Bathroom Occupied"
     prior: 0.5
     probability_threshold: 0.75
     observations:
       - platform: 'state'
         entity_id: 'binary_sensor.upstairs_bathroom_shower_motion_sensor'
         prob_given_true: 0.9
         prob_given_false: 0.1
         to_state: 'on'
       - platform: 'state'
         entity_id: 'binary_sensor.upstairs_bathroom_motion_sensor'
         prob_given_true: 0.9
         prob_given_false: 0.1
         to_state: 'on'

It never triggers, even though I was thinking it should change to true if either motion sensor activates. I did try changing prior to 0.75 and then it does trigger, but it’s always triggered (which makes sense I guess?)

The motion sensors themselves work and they are classed as motion sensors. Do I have to use Detected instead of on? dev-state shows on and off, so I figured that was correct.

When I pull up the sensor from the states page, Observations is always blank.

Thanks for your help

And like magic, suddenly they are working. Maybe it just takes awhile to collect enough data?