System and method for improving clinical decisions by aggregating, validating and analysing genetic and phenotypic data

A method for predicting and outputting a clinical outcome for a first subject, based on a first set of genetic, phenotypic and/or clinical data from the first subject, a second set of genetic, phenotypic and/or clinical data from a group of second subjects for whom a first clinical outcome is known, and a set of statistical models and training methods from published reports of experts, the method comprising:

• integrating, on a computer, the second set of data from the group of second subjects into a standardized data model according to a first set of standardized data classes that have unambiguous definition, are related to one another based on computed statistical relationships and/or expert relationships, and encompass at least a portion of all available relevant genetic, phenotypic and clinical data, each standardized data class being represented by a corresponding random variable that describes a corresponding feature;
• structuring, on a computer, the first set of data from the first subject according to the first set of standardized data classes, the structured first set of data including first-subject feature values for the features corresponding to the standardized data classes;
• automatically selecting, on a computer, from the set of statistical models, a first statistical model for predicting the clinical outcome of the first subject in response to a first intervention based on the first set of standardized data classes and the first and second sets of data by selecting features corresponding to the standardized data classes for the first statistical model to improve a predictive value of the selected features for predicting the clinical outcome, the first statistical model operating to relate the random variables corresponding to the selected features to the clinical outcome;
• automatically selecting, on a computer, from the group of second subjects, a patient subgroup with characteristics similar to the first subject by comparing corresponding second-subject feature values with the first-subject feature values for the selected features;
• training, on a computer, the first statistical model based on the second set of data from the patient subgroup together with the first clinical outcome of the subgroup of patients;
• applying, on a computer, the trained first statistical model to the first set of data of the first subject to predict the clinical outcome for the first subject in response to the first intervention;
• and
• outputting the predicted clinical outcome on a fixed medium.