Sunday, April 06, 2008

Complex Event Processing in Healthcare

It is interesting how new ideas and technologies rise to meet the needs of one industry and then diffuse into other industries. Classic examples are the trickledown from aerospace into the motor industry, and the application of diagnostic imaging technology into airport security.

I've recently spotted the early stages of a technology which has risen in the investment banking sector now starting to find its feet in healthcare. The technology is known as complex event processing (CEP).

As recorded on Wikipedia, a complex event is "what one infers from simple events" and gives the example of a lady in a white dress, a man in a tuxedo and lots of rice flying through the air being a wedding. In banking, CEP is used to take on board very simple events (e.g. sell prices) and infer from them something richer (e.g. market trends) from which a decision can be made (e.g. go short on a specific portfolio stocks). A search on Google of "complex event processing healthcare" returns a relatively modest 193,000 results. I can see this exploding and look forward to checking in with Google Trends in due course to see the references rolling in.

One area of health care where I think this will hold great value is in post-operative recovery monitoring of vital signs. The Association of Anaesthetists of Great Britain and Ireland recommends that during anaeasthesia and recovery a variety of monitors should supplement clinical observation. The challenge is that once outside of theatre the available clinical resources are limited and can only observe on a sampled basis. Also, the analysis of the monitor readings is trying to spot early onset of a complex range of problems such as hypoxemia, hypoventilation, hypotension, hypertension, hyperthermia, hypothermia and dysrhythmias.

In the event of deteriation, rapid response is a critical success factor in intervention. Different jurisdictions recommend different frequencies of monitoring (research for this article found a range from 5 mins to 30 mins for most vital signs) so you can't help but feel this is driven by the limiting factor of resources rather than clinical need. In fact, muscle relaxants (one possible intervention) are often chosen on the basis of an onset speed which is measured in minutes, not tens of minutes. CEP could be applied to spot problems on a real time basis, moving the limiting factor onto being the monitors' sample rates rather than the care pathway. I'd be very interested in talking to anyone who's working in applications in this area.

Like many other technologies in healthcare, no doubt we will see the rise of this technology met by the acronym being misinterpreted as clinical event processing. Look out for that!

No comments: