CureHunter is an interesting entry in the medical search engine space. Actually, what makes it interesting is that it does not categorize itself as a search engine, but as a data mining tool. Now these are very deep waters for those of us who are not computer scientists. And rather than risk embarrassing myself with a bungled attempt at an elucidation of the differences, I will simply quote from the discussion of these matters on the CureHunter Web site and then give my own impressions.

How is CureHunter different than Google or other Medical Information sites on the web?

CureHunter is unlike Google and ordinary search engines used in general commerce and health information retrieval in several important ways–including its basic theory of operation. CureHunter is built on the model of scientific instruments. The machine must control samples and sample preparation. It must produce testable results consistently and be subject to test by other methods of analysis outside itself to provide 3rd party validation of its conclusions and returns. It must also have extremely large and carefully controlled technical medical dictionaries that are generally not found in common search engines that fail to recognize many words in the scientific research and thus deliver incorrect results.

Does CureHunter use “relevance ranking” to find good information?

CureHunter does not use the concept of “relevance ranking” at all. All its data extractions (search results) must be precisely relevant to start with. A very advanced natural language processing module has the task of reading the literature the same way a human scientist would. There are no “top 10 or 20″ supposedly important articles followed by millions of “hits found” of totally irrelevant or weakly related information.

How do search engines differ from medical data mining systems?

Search engines point to information in distributed articles of all kinds and tell you to then click one article at a time, go read the article yourself and see if the article has any information you believe useful. They also ask you to write 1 to many hundreds of different queries with different spellings of words and names for various related ideas. Often the great majority of information returned has no meaningful value to you at all: “42 million hits found.” CureHunter automatically extracts the key data (mines it), reads it, analyzes it, and draws scientific conclusions for presentation to you and your doctor.”

Now, that is all quite interesting. I tend to like any search-related or tech company that offers helpful backgrounders such as that. It is good marketing to explain what exactly your product is and why we as consumers would want to use it. It is amazing that so few companies employ this simple technique of winning over potential customers.

But there are some marketing problems with CureHunter, too. There is no “About Us” page, which is unfortunate given that the primary potential customers for CureHunter’s products are physicians, drug companies, medical libraries and affluent consumers. Most of those groups are going to want some idea of the credentials of the management team. We need to see the credentials in terms of academic degrees and published papers and a list of customers. We need to see at least a few names of management and list of clients. See for example, Deep Web Technologies; Private Access; and Vivisimo.

All confidence-enhancing, credibility-establishing information about the quite impressive people running those companies. CureHunter simply says:

“CureHunter is an Oregon, USA, corporation founded in 2003, by a team of scientists with extensive experience in medical data mining, artificial intelligence software development, computational linguistics and computational biology: A “perfect storm team” for the task of training a machine to find cures for human disease—by reading the scientific literature.”

Soooo, could we have some idea who is on this team of scientists and what they have accomplished in the field (e.g., positions held elsewhere, scholarly papers on artificial intelligence). Also, it is crucial that we be able to see who is actually using the products in question. See here for instance, or here.

Even the rather junky PharmaSURVEYOR features more info on it founders and management than CureHunter does:

I mention the need for better marketing for CureHunter because I actually do wish it well. It does have useful features.

For example, I tried my usual search term, “amyotrophic lateral sclerosis,” and was immediately given the opportunity to set up an RSS feed, “Subscribe to Selected New Research on Amyotrophic Lateral Sclerosis.” I clicked on that to see what some of the articles were and was able to learn of the newly published article, “Palliative Treatment of Dysphonia and Dysarthria” in Otolaryngologic Clinics of North America. The other articles were equally substantive. I wish that CureHunter featured the marvelous email alerting capacity for saved searches that Mednar does.

CureHunter, like GoPubMed and Mednar, features among its results the names of leading experts in the subjects searched and the results for each author listed are user friendly, presented attractively and useful. I wished that CureHunter offered GoPubMed’s capacity to offer users the ability to contact, via GoPubMed’s mediation, such authors.

One especially useful feature of CureHunter is its custom RSS Feeds for pharma.

You can search the extensive list by company name and from there to a specific drug. I tried Amerisource Health Services as a test and noticed that there was a feed listed for the drug, baclofen. I checked out the feed and found the abstract for the article, “Intrathecal Baclofen for Spasticity-Related Pain in Amyotrophic Lateral Sclerosis: Efficacy and Factors Associated With Pain Relief,” in the journal Muscle & Nerve. I was rather surprised that the feed seemed to consist of basically the standard literature a la PubMed. I was expecting and rather hoping for feeds directly from pharmaceutical companies in the form of press releases or white papers of their doings at an earlier stage of development of drugs testing than is reflected in what appears in the literature after safety and efficacy have been pretty well established. Thus, the custom RSS Feeds for pharma are similar to what one would get PubMed itself and not as rich in terms of Deep Web sources as Mednar alerts.

Another problem with CureHunter’s custom RSS Feeds for pharma is that users can’t search by the name of a drug, as far as I could tell. For instance, I only happened upon the feed for baclofen because I happened upon it as one of the drugs listed when I randomly chose Amerisource Health Services as my test.

Thus, I would have to know who manufactures Rilutek (which I determined by popping over to Google instead on being able to stay in CureHunter) and then clicking through the various branches of Sanofi-Aentis to find feeds for that drug from the Canadian and French office of the company but not the listings for the German and US offices. This is a rather time-consuming method for finding data in a supposedly highly efficient search environment.

Also, it was not clear to me that the rather pricey report on drug treatments (that for ALS is $24.00 & delivery via PDF download updates for one year, for instance) are worth the money. Consumers can get pretty much the same information via MedlinePlus and medical librarians and clinicians have specialized databases and pharmacists to consult.

Still, CureHunter does offer useful information on drugs and biologics and those in need of specialized information on such matters might want to check it out. There is a sample report on the treatment of psoriasis on the CureHunter Web site.