Medical Transcription and in general mining information from (largely handwritten) clinical notes is probably the primary bottleneck in any attempt at automatically management - collection, storage and retrieval - of medical records. Any such system must have diverse capabilities such as (i) text mining capabilities, (ii) optical character recognition of handwritten text and/or speech recognition using the following contextual information, (iii) bring in the vast medical knowledge-base  to serve as the context for the mining task, and (iv) use the patient and doctor profiles effectively to personalize, for it to be useful. Many other extensions are possible such as integrating speech recognition to directly transcribe clinical notes from  dictations. These are active research areas with dedicated sessions for these in most conferences focused on machine learning and/or health care. To start with we are targeting a two-year milestone to deliver a preliminary proof of concept and publications to add credibility to the algorithms used for the prototype.