This is the experience of a emergency room physician who has treated over 80000 patients, and many a times was put in a position of "guessing over and over," in an information vacuum. In situations where people "die right in front of you," he said he often felt he was "one data element away" from stopping a patient from dying.

When electronic medical record (EMR), is adopted, life will be different. But the problem is we are nowhere near the total level of adoption and the barriers are challenging. Privacy, interoperability, liability issues, etc., are all main stays of resistance. A survey on Indian hospitals say that less than 12 percent had even minimal EMR adoption, however, more than 50 percent of hospitals had certain functions on the list, such as recording patient demographics or viewing results of lab tests. ?€?They just haven?€?t put the pieces together?€? to create a comprehensive EMR.

The overriding reason to seek better EMR and record processing is to acquire the ability to use clinical information to help us move beyond the present practice. But a stumbling block is that, many information resources including sophisticated image analysis tools are independent. As with any system, standards such as Health Level 7 (HL7) have evolved and on the other hand, XML supports an ideal basis for information integration. So, its very natural to use XML to describe EMR standardized to HL7.

As the use of EMR becoming more widespread, it is natural to look beyond the normal use of an EMR, as an effective means of information discovery. This requires us to tackle the problem of facilitating ontology-aware information discovery on a corpus of XML-based EMR documents. Reasoning over the ontology allows more focused searching than a simplistic key-word approach. The query ?€?Show me all documents about dysuria?€? could return documents with terms such as difficulty in passing urine, micturition disability, etc..

A large corpus of work has gone into developing medical expert system since. This experience-based reasoning paradigm, exploits the knowledge acquired from previously experienced knowledge to solve new problems. Given the above approach, the EMR provides a critical mass of real-life cases via a set of situation and an associated solution derived by a medical expert. This can in turn be the knowledge base for the case-based reasoning system.

When we look beyond, what we want is an organization of information that's closer to humans perspective than a machine's. We are walking into the realms of Artificial Intelligence. We in fact want the computers to describe and analyze a given information along three dimensions: spatial, thematic and temporal. The thematic dimension describes what occurred, the spatial dimension describes where it occurred and the temporal dimension describes when it occurred. This leads to better use of EMR.

All the issues on EMR exchange such as security, privacy, etc. can be sorted out and a flexible mechanism can be evolved. This will web-enable the EMR in its true sense. This infrastructure can help us to integrate agent technology to preform complex tasks. This will extend programs to perform tasks efficiently with less human intervention.

This paper is not intended as a comprehensive technical tome. Rather, I hope that I have convinced that several strands of research can be brought together in exciting and interesting ways. The pieces are coming together, and thus, the semantic EMR with agents is no longer a science fiction future. It is a practical application on which to focus our current efforts.