Pubmed Faces is a fun application that illustrates the power of this visualization technique.
The following screenshot shows Chernoff Faces for some of my papers (though the citation counts seem way off compared to Google Scholar counts).
As explained on http://www.postgenomic.com/faces/index.php. [Accessed: 2008-06-19. (Archived by WebCite® at http://www.webcitation.org/5Yh6x5X1O)]
The age of a face correlates with the publication date of the paper. Younger faces are more recent papers.
A smile means that the paper has been cited more times than expected (based on its age). Larger smiles mean more citations.
A frown means that the paper has been cited far less than you might expect.
The raised eyebrows correlate with the impact factor (sort of - actually the Eigenfactor) of the journal in which the paper was published.
While I do not necessarily agree with all the metrics they used (as I said, the citation counts look suspect, and I find the use of the Eigenfactor questionable in this context - what should have been used is the relative impact factor ranking [rank percentile] of a journal within a given discipline), and while in this particular application I am personally more comfortable with looking at numbers than trying to read faces (perhaps its because I am left brain-hemisphere dominated male and generally bad in reading people!) - the concept of using Chernoff faces for visualization is intriguing.
My suggestion would be to experiment with this in other contexts - especially in the context of personal health records, or more specifically what I call Personal Health Records 2.0, i.e. PHRs which include social networks for patients like PatientsLikeMe. Chernoff faces could be used to represent avatars of real people, summarizing patient features (in addition to more specific nuggets). Face features like the smile or the raised eyebrows could be influenced by the general health status or specific health problems.