Data Science: Similarity Least Squares (SLSTM) + physics + Statistical Design of Experiment (DOE) A New Paradigm for Analysis and Management of Complexity |
Examples: Consider a tale of two patients, a and b, at week 4 of treatment, a male and a female: 0)patient Profiles: p sex age Genotype Fibrosis Act PriorEndPt LogVLoad LogALT LogVDropW4 a M 37 3 2 2 NA 5.088 1.845 3.088 b F 60 4 3 3 NR 6.326 2.376 0.585 1)Convert all info to scaled values via scaling tables: sex age GT Fib Act PriorEndPt LogVLoad LogALT vLogDropW4 -0.07 37 -0.98 -0.52 0.07 -0.48 5.088 1.845 3.088 -0.49 60 0.44 0.25 0.29 0.49 6.326 2.376 0.585 2)Z-score (Zing) all info: sex age GT Fib Act PriorEndPt LogVLoad LogALT vLogDropW4 0.74 -1.08 -1.08 -0.44 0.36 -0.47 -1.908 -0.542 0.313 -1.34 1.34 1.04 0.68 0.66 1.32 -0.110 1.192 -1.459 3)Apply SLSTM model: a V= -2.73 predicts SVR for the male b V= 1.69 predicts NR for the female |
Copyright of James M Minor, July 4, 2004.
|