Statistical models for genetics data are often surprisingly challenging, and often require advanced and new statistical methods. Using probability machines on whole genome data is a recent invention, with the original research on probability machines appearing in Methods of Information in Medicine (September 2011). Our methods point to refined and personalized probability predictions using a wide range of biomarkers, medical information and whole genome data. The detection of childhood-onset schizophrenia using 800,000 snps using probability machines has error rates of 15% or less, and the list of predictive snps can be filtered down to a list of less than a few hundred. Other studies of psychiatric conditions (ADHD, bipolar) are also now underway using probability machines and personalized medicine, subject-specific predictions.