A new study from Australia has purported to show that physicians faced with difficult cases would benefit from Googling the symptoms and the disease or treatments. Not all the medical profession is jumping into the great search engine of the Internet. I was intrigued by the present and future possibilities and the alternatives for physicians now who use the Web for professional information.
Medical News Today reported in “Google Good Source For Doctors To Diagnose Hard Cases” by Christian Nordqvist that researchers from Brisbane, Australia (as reported in the British Medical Journal) found using Google to search for information in difficult medical cases helpful 58% of the time.
The basis of the study was that physicians needed ever higher amounts of knowledge to accurately diagnose illnesses—especially those less ordinary. The two they used (of 26 “hard-to-diagnose” cases), found in the New England Journal of Medicine, were Cushing's syndrome and CJD (Creutzfeldt-Jakob disease).
The report showed the physicians used three to five search words in Google for each medical case. Theoretically, they did not know what the correct diagnosis should be beforehand. The three highest ranked diagnoses based on symptomatology (when compared with the New England Journal) were taken. “They say that 58% of diagnoses carried out using Google searches were correct.”
The British Medical Journal reports the results as the Google searches came up with a correct diagnosis in 15 cases, which translates to 58% of the time -- which shows a "95% confidence interval 38% to 77%."
They conclude, “As Internet access becomes more readily available in outpatient clinics and hospital wards, the web is rapidly becoming an important clinical tool for doctors. The use of web-based searching may help doctors to diagnose difficult cases.”
What I found the most interesting in these articles about a mass data search for important diagnostic tools in medical treatments was not the somewhat simplistic study but the response in the comments (just like Blogcritics, medical people can now discuss medical journal articles as they appear) section of the BMJ. The first was Dr. Joseph
Britto, who is the CEO and Clinical Director of Isabel Healthcare Inc. in Reston, Virginia, which provides an Internet-based DDSS (diagnosis decision support system) for professional healthcare providers. His system, he feels, is far superior and more professional for medicos. I must quote him, since the programming math is far beyond me. His “Isabel” system:
- uses natural language processing algorithms that searches by context and meaning a database of medical textbooks and journals - to 'understand' rather than just ‘find.’ Isabel suggests diagnoses rather than documents and these diagnoses are filtered using the patient’s age, gender, pregnancy state and geographical-region prevalence heuristics. Further, as a quality metric, we analyze and make available on Isabel results of Isabel’s diagnostic performance on current NEJM CPC cases. A study submitted for publication looked at Isabel’s performance on 50 NEJM CPC cases from 2005 using whole text data entry [entire case presentation cut and pasted verbatim] and entry of extracted clinical features. Isabel came up with the final diagnosis in 74% and 96% respectively.
Joy Kennedy, a reference librarian from Virginia, commented that she was initially shocked at the idea of using Google for such technical research, but then decided “By using Google to search the web, they were essentially doing a full text search of a giant database. Admittedly this database contains good, bad and indifferent material but the concept is not an unreasonable one. There is great value in searching the full text of journal articles, tables of content of books and other, more reputable tools...”
Article comments