Patient　data,　more　exactly,　electronic　medical　records　(EMR),　usually　contain　a　lot　of　free　texts.　Those　unstructured　medical　data　cannot　be　easily　understood　by　computers.　In　addition,　EMR　data　have　a　strong　privacy,　which　hinders　the　sharing　and　use　of　medical　data　and　makes　it　impossible　to　conduct　more　in-depth　medical　research.　This　paper　presents　a　method　of　the　realization　of　semantic　EMR　by　making　semantic　annotations　on　free　texts　in　medical　records.　We　will　show　how　to　use　Natural　Language　Processing　(NLP)　tools　to　create　semantic　annotation　with　wellknown　biomedical　terminologies/ontologies　such　as　the　Unified　Medical　Language　System　(UMLS).　Moreover,　we　will　describe　how　to　make　the　semantic　annotations　on　a　set　of　virtual　patient　data　for　depression,　which　are　generated　by　using　the　Advanced　Patient　Data　Generator　(APDG),　a　knowledge-based　patient　data　generator.　In　short,　our　goal　is　to　use　semantic　technology　to　improve　the　sharing　and　utilization　of　medical　data　and　the　interoperability　among　systems.　©　2018　Association　for　Computing　Machinery.