Extending models via gradient boosting: An application to Mendelian models