I am trying to predict the probability of a disease by using audio and images, the audio and the images do not come from the same source. I am thinking of combining the outputs (maybe average them) of two models one utilizing image data to calculate the probability the other using audio data. I am feeling unsure of this approach as I will not be able to verify the final combined output. Will the outcomes be reliable, or should I use a different approach?
You can try to fuse the image and audio data. You can fuse in different ways:
1) Early fusion - fusion at feature level
2) Late fusion - fusion at output level (similar to your idea)
Here is some papers for your reference:
Look, Listen and Learn https://arxiv.org/pdf/1705.08168.pdf
Learn to Combine Modalities in Multimodal Deep Learning https://arxiv.org/pdf/1805.11730.pdf