My thanks to Lead Engineer, Alexander Schonfeld for bringing this one to my attention. CureHunter is a Portland, Oregon 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”. Their mission is nicely wrapped up in a single quote: “Quantified [...]
CureHunter – precision medical data mining
My thanks to Lead Engineer, Alexander Schonfeld for bringing this one to my attention. CureHunter is a Portland, Oregon 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”. Their mission is nicely wrapped up in a single quote:
“Quantified evidence for drug efficacy made accessible during the patient visit”
Specifically, their goal is to develop an artificial intelligence-based computer that can autonomously search all the known biomedical research journals, collate the published drug efficacy evidence for specific diseases and present it in a format that is available in real-time (10-20 seconds) for patients and physicians to review. With integration into existing physician record management systems, physicians can use (free of charge) the drug research interface and obtain up-to-date summarized clinical effectiveness information on a wide range of drugs and diseases while the patient is sitting in the room!
For patients, CureHunter tries to answer the question:
What does the scientific community think the best treatment options for disease Y are?
Patients simply need to enter the disease they wish to know more about in the search box on the front page. They utilize the Mesh-based ontological terms to help narrow their search down to the specific disease they are searching for and CureHunter returns to them: a) Key Drugs and Agents for the treatment of that disease, b) Other Related Diseases, and c) Key Therapies for that disease. For convenience, the Mesh-based ontology is shown to help them understand where their disease fits in the biomedical categorization system (in case they wish to search for similar diseases). It is at this point that patients are able to purchase an extensive pdf summary (the sample psoriasis report is over 150 pages!) of this information for $24. This summary, which has mined efficacy, related drug, related disease and related therapies information from countless drug trials published by the US National Library of Medicine Medline Archive, provides a (somewhat overwhelming) snapshot of which drugs are considered (by the AI engine) to be the most appropriate treatment for the selected disease. For the information hungry consumer, this is excellent content to have at hand during discussions with treating medical professionals.
An additional feature on the site which will also be of interest to patients and general consumers of health information is the Visual Medical Dictionary, which provides not only a definition of the disease you enter, but a visual representation of the drug and disease relationships that exist between the search term and other diseases/drugs in the literature. Hover your mouse over the relationship map and you will get definitions of these related terms. A really really neat way of visualizing the range of drug agents used to treat specific disorders.
CureHunter has not just been built with general consumers in mind though. Physicians and researchers are, through subscription or registration, able to access specific interfaces that allow real-time analysis and breakdown of the AI search results. Researchers can even export data from specific searches to common analysis tools for further manipulation.
The potential of CureHunter lies in the AI engine and the speed with which it can collate and summarize information across multiple drugs, trials and related diseases. It is not outlandish to think that in the hands of an experienced drug researcher, this artificial intelligence engine could lead to unique discoveries regarding the efficacy and application of drug therapies. When you think how difficult it would be for physicians to realistically and comprehensively review the literature on all drugs they prescribe, this kind of engine has significant potential. When I wrote back to Alexander about the site, I was very excited at the prospect of this kind of data mining being applied to the hundreds of different psychological therapies.
The data mining engine has been in development for a few years now, and after two prototyping iterations was released to the public as a beta in July 2007. The CureHunter corporation are still at early startup stage and are still seeking venture funding and/or a partner. I wish them all the best as they look to take their product further. I think they are creating an amazing tool for physicians and I hope the concept spills over to allied health professionals.