That is an interesting quesion whose answer may differ among fields. But i think there would be some egenera lrequirements for any new findings to be considered as science. Some degree of novelity is needed, but in medicine however, sometimes we do tens of trials on the same drug in the same disease with small changes everytime. This is required due to the diffeences among humans and that we can not apply a treatment based on a single study.
Another more important criterial is authenticity and representation. The data must be authentic and should represent the population from which they were obtained. Otherwise, their findings may be irrelevant. Fianlly, i think everyone would always consider what he does as scientific, but it is up to the scientific community and time on the long term to prove that.
A new article in a medical journal could be considered a new science if it is a proven innovation. I am not sure that "new science" is the best term to describe a breakthrough medical tests' results, because the term indicates a completely new trends in many scientific areas. "Innovation" seems to be a better term, a "breakthrough discovery".
I would not write (even in a newspaper article) that a research paper reports "a new science". Except if it had very specific, breakthrough characteristics:
1. The field must be capable of receiving a new name, and not just be an old method in new clothing, or a extension of an old method. It may be a unique combination of two or more previous methods.
2. The method must be unique and take an unanticipated turn.
3. The field must be capable of generating useful results that could not be derived any other way.
For example, it may be to investigate the viability of Martian "soil" for plant life.
1. Martian agronomy
2a. Sends an automated "soil" digger to Mars which feeds a sample to a remote automated lab for analysis
2b. Recreates the same mix on Earth to test viability
2c. Investigates microbes needed to seed the "soil" to enable plants to be grown.