Apparently machine learning is lousy if the data going into it is skewed by being either too heavily male or female:
IMO, i would look at bias of any kind :
-A Researcher that specialises in Brain metabolism may be biased to hypothesise that the brain function is important to ME
- A Researcher that specialises in Viruses may be biased to hypothesise that the viruses are important to ME
-An ME Organisation may be biased in providing specific researchers with funding because the commitee of experts of the organisation may be biased in their own ways.
-A computer scientist who specializes in Machine Learning may be biased to hypothesise that Machine Learning can find the solution to ME (...)
But, if you ask me -and perhaps being biased myself- algorithms - if used properly- are more likely to find a solution in an unbiased way than any other human (if the input data are not Garbage).