Dan Ford, IBM Almaden Research Center, Provides an Update on the Spatiotemporal Epidemiological Modeler (STEM) Work of the Eclipse OHF Project

I haven’t seen a lot of activity on the dev list lately and I  know that that doesn’t reflect the high level of activity in the OHF project.  To jump start our communication channel a bit I’ve written a short summary of the status of the STEM project (http://www.eclipse.org/ohf/components/stem/) which is developing an open platform for disease modeling based on Eclipse.

Some highlights:

We recently (last week) hired a technical writer for the project who will be helping us with documentation, wiki, etc.

We received multi-year funding from the USAF to incorporate features for disease model validation.  The implementation of those features is going well.

We recently extended the core modeling framework to add an additional component called an "Experiment" which is a container for the specification of multiple simulations all based on a single root Scenario.  This is a powerful addition to the STEM representational framework as it allows a base model to be modified in specific ways to explore the potential effects of different public policy decisions.  It also enables the process of disease model validation by making it easier to explore the variations in disease models and compare them to known data sets.

The development team is migrating from Eclipse3.4M5 to Eclipse3.4M6.

The STEM build process has been automated.  Though there is still some work here to complete the job.  Right now the process builds STEM for Windows, MAC and Linux, but it does not yet deploy the builds to the download site.  Consequently, the last available build was in December.  This needs to change.

We’re working with the legal department to clear our geographic datasets.   

The MapView graphics can now display the edges connecting geographic regions.  This was literally an "eye opener" as it made several mistakes in our common border edge data sets blindingly obvious.  There were errors in the USA county border connects and others in Africa and China.  This wasn’t too surprising as the connections were generated automatically by examining region Lat/Long border data, but even 99.99% correct isn’t good enough.

We’re rewriting the EMF generated editors.  The EMF editors work well enough, but they really need to be customized to the particular framework component being edited to make them more user friendly.

There has been some work on profiling the performance of STEM and this has resulted in some fine tuning of the internal MapView graphics.

We currently have 92 open bug reports with 16 unassigned.