First impressions of the ICON EHS Database Analysis Tool

What do you do this holiday season when the turkey’s lost its appeal, you’ve seen every movie worth watching ten times over, and conversational déjà-vu sets in?  If you are really desperate, you could play “nano-trivia”—and thanks to the fine folks at the International Council On Nanotechnology (ICON) you now have the perfect widget to help craft those cunning questions that will have your nearest and dearest wracking their brains.

Questions like “between 2000 and 2006, what percentage of scientific papers addressing the toxicity of carbon-based nanomaterials considered exposure via mucous membranes (or the skin)?”

OK, so maybe playing “toxic particle trivial pursuits” is the last resort of the desperate, and likely to result in an ever-decreasing circle of close friends.  But for all that, the new ICON Environmental Health and Safety Database Analysis Tool has its merits… Most importantly, it provides a fascinating insight into how new knowledge on nanomaterial safety is progressing—or not, as the case may be.

Backtracking a little, the EHS Database Analysis Tool (lets just call it “the widget”) is an add-on to the ICON nanoEHS Virtual Journal.  I’m a long-time fan of the Virtual Journal, which is probably the foremost repository of information on scientific papers addressing the potential health and environmental impacts of engineered nanomaterials.  Established and maintained by ICON, it links to close-to every paper published that has some relevance to understanding and addressing the possible impacts of nanomaterials, and is an essential resource for anyone doing work in this area.

But those clever people down at Rice University didn’t just stop at cataloguing the constant stream of publications coming out of researchers around the world.  They went one step further and added some useful information—such as what material was studied in the published research, how it was studied, which aspects of hazard or risk were addressed, who the publication was aimed at, and so on.

And that opened up the way for “the widget.”

What the widget does is enable sophisticated searches on the database, and then displays the information graphically (as well as giving direct access to the source-paper records).

Imagine for a moment you are interested in the relative numbers of papers that have been published to date on different routes for carbon-based particles to get into the body—ingestion, inhalation, or through the skin or mucous membranes.  Plug the desired information into a reasonably easy to use matrix on the widget’s web page, select a “Simple Distribution Analysis” plot for the years 1961 through to the end of 2008, and press “Generate Report.”

Hey presto, the widget creates a neat little pie chart clearly showing the requested information.  (For the interested, across these three exposure routes and for the years and material in question, 86% of papers address inhalation, 11% dermal/mucous membrane penetration, and 3% ingestion).


This analysis gives you a sense of how research has balanced out over different areas over a number of years.  But what if you want to know how things are changing—whether more is being published now on carbon nanoparticles for instance than was being published five years ago?  You should not be surprised to hear that the widget can handle this also.

On the widget’s web page, choose “Time Progressive Distribution Analysis” and hit go, and you get the number of papers published per year in each category, displayed as a bar chart.  You can also narrow or widen the number of years covered by each bar—the example below shows the number of publications every two years. (“Series 1″ represents dermal/mucosal exposure – a niggling aspect of the widget display I tackle further down the page).


As well as providing hours of fun for the socially-challenges (regrettably, I suspect I fall in this group), the widget is a great gateway into the rich publications database ICON are amassing.  One trend it shows readily is an apparent exponential increase in the number of nanomaterial environment, health and safety papers being published.  The following plot is the time series from 1981 to 2008 for papers covering any exposure pathway, summed up in two year blocks.


More interestingly, this trend can be separated out by exposure route—ingestion, inhalation, dermal, multiple routes, or the catch-all category “other/unspecified.”  And when the data are plotted out (see below), things get interesting—the exponential rise in publications is only seen for the “other/unspecified” category.  Publication rates for papers dealing with inhalation or dermal uptake seem somewhat static.


Further exploration reveals that the rise in papers within the “other/unspecified” category is predominantly associated with in vitro studies.

Even after just an hour or so playing with the widget, it is clear that it is a powerful tool for assessing trends, and beginning to identify deficiencies in the global research agenda.  Beyond producing useful web graphics, each analysis can be downloaded as a PDF or Excel spreadsheet—a useful feature.  And as has been mentioned, the papers each analysis is based on can be listed and examined in more depth.

But the widget isn’t perfect.  Like any data analysis tool, how the numbers are interpreted is everything (remember the old computer adage—garbage in, garbage out)—and it’s easy to generate numbers that can be misleading.  Just one example: plot the numbers of papers on inhalation exposure between 1983 and 2008, averaging over every five years, and you will be shocked to see that there seems to be a dramatic decrease in publications over the past five years.  Until you realize that the last bar on the plot represents just one year (January 2008 – December 2008), and not five.  Artefacts like this can be misleading if a weather-eye isn’t trained on what is actually being shown.

Another example is the pie chart shown earlier in this post.  It’s tempting to read the percentages shown on the plot as the fractions of papers published on either inhalation, ingestion or dermal exposure.  Whereas they are only the fractions of papers within this particular grouping—add other possibilities into the mix (injection, multiple exposure pathways, the catch-all “other/unspecified” category), and the percentages change.

And there are some niggling things about how the widget does its stuff, like a tendency for the plot legend to revert to “series 1” or “series 2” etc. when the full descriptor is too long (as on the chart above).  Or a pet peeve of mine—referring to a bar chart as a “histogram!”

But despite these potential pitfalls and minor irritations, this is a powerful tool that is extremely useful for mining the nanoEHS Virtual Journal’s extensive data.

And, of course, the widget is the perfect resource for that last-ditch game of “toxic particles trivial pursuit.”

Go on, give it a whirl—I dare you!

* Of course, I must point out that not all nanoparticles are necessarily more toxic than their larger counterparts, and that some nanoparticles seem to be pretty benign compared to others.  But the title “Particles of indeterminate toxicity and trivial pursuits” just didn’t seem to scan quite so well…