Ann: Positive Results for Australian Prospective Paediatric Study, page-123

Not sure about fluff ... but good question anyway. Imo.

Reverse engineer the results to calculate a few values.

For the 60 children who had pneumonia if you relied on the app 8 would be undiagnosed. For the 508 who didn’t have pneumonia the app said 76 of them did.

The positive predictive value (probability you have the disease when the test says you do) for pneumonia is around 40%. For lung ultrasound PPVs are around 61% and for x-ray 71%.

Is the amount of additional information coming out of the app worth the trouble over and above what you get from symptoms alone? A question to ponder.

The primary outcome for the US study is pneumonia. The FDA benchmark is 75% PPA and NPA. This means that the lower bound of the 95% CI of the estimate needs to be better than 75%.

The Australian study has produced PPA 75% and NPA 82% - lower bound of 95%CI. There is no margin cushioned in the Australian study results to the US clinical trial.

The US study has a larger sample so CIs are likely to shrink. But a clinical trial will be more methodologically rigorous than a study. Hope for the best … but asthma might be more the way to go.