I really think that with heart failure, we need the biomarker equivalent of somatic pre-cancerous DNA changes in cancer. But we also need (to steal from the psychiatrists), better endophenotyping, because many different aberrant states lead to very similar clinical syndromes of "heart failure".

A group, perhaps yours or in concert with yours, could both define the disease subsets based on biomarkers and then diagnose/prognose/inform treatment. Surely the same is true for other diseases, this is just one high-impact example.