Deep learning with multi modalities phenotypes and biomarkers

There is a growing interest in the biomedical world in utilizing multi-modal multi-featured machine learning applications to create models that can predict disease development. Identifying vulnerability to the development of health problems entails important prevention options including treatments and lifestyle changes. Working with multi modalities data requires additional steps and preparation, making sure that the combined modalities don't skew the results. In the current example, we used a dataset that includes a combination of demographics, clinical diagnosis, genetics, and biomarker features. We used supervised deep learning with Python/Keras to create a model for identifying individuals with vulnerability to develop major depression.