New Method Assigns Probabilities Across Multiple Psychiatric Diagnoses
DDx-PRS jointly estimates posterior probabilities of each diagnostic category (for example, schizophrenia 50 percent, bipolar disorder 25 percent, major depressive disorder 15 percent, control 10 percent) by modeling variance-covariance structure across disorders. Researchers applied it to Psychiatric Genomics Consortium data with training sets of 41,917 to 173,140 cases (total sample 1,048,683) and a held-out test set of 11,460 with equal numbers per diagnostic category. The method was well calibrated and well powered. True diagnosis probabilities in the top deciles of predicted diagnosis probabilities were considerably larger than prior baseline probabilities.
The method may help clinicians differentiate between genetically overlapping psychiatric conditions when diagnosis is uncertain.
Written by the Genomes desk from the primary source linked above and checked against it. Research use only; not medical advice. Corrections: [email protected].