1/n) Brilliant talk by @f2harrell yesterday "Musings on Statistical Models vs. Machine Learning in Health Research": https://www.fharrell.com/talk/mlhealth/ What about combining machine learning with multiscale #mathbio models...
2/n) #mathbio #systemsbiology combined 3 multi-scale models with machine learning for pharmacology prediction problem: . https://pubmed.ncbi.nlm.nih.gov/26950176/ . Yet simple "pre-school" maths was sufficient: https://pubmed.ncbi.nlm.nih.gov/27617708/ Why? Sometimes mechanisms can be linear!
What I've learnt is that complex methods in the wrong hands can cause harm! Its rife in life/medical science where complex approaches have been made "easy to use" such that every person and his dog is doing this stuff - it looks like a dogs dinner!