Time pressured clinicians overwhelmed with the exponential increase in bio-medical knowledge are desperately seeking clinical diagnosis decision support systems (CDDSS). Google, whilst extremely helpful in shifting the 'search paradigm', is of limited usefulness as a CDDSS. Google's accuracy of 58%, reported by Tang and Ng in the British Medical Journal, 10th November 2006, is less than that achieved by older generation rules based CDDSS and will not engender widespread adoption.
Isabel (www.isabelhealthcare.com) is a web-based, point-of-care CDDSS designed used by healthcare professionals and has been extensively validated in clinical studies in terms of ease of use, accuracy and impact as a diagnosis reminder system. Isabel has been shown in published studies to be accurate in over 90 % of cases and to cause frontline physicians to consider an important diagnosis they should have considered in 1 in 8 cases.
Isabel uses natural language processing algorithms (www.autonomy.com) 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. An independent study submitted for publication looked at Isabel's performance on the same set of cases 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.
The aim of CDDSS is not to replace but to quickly and easily give the 'learned intermediary' (clinician) a differential diagnosis to consider. Sophisticated and validated CDDSS are now able to rapidly assist diagnosticians and make the cognitive process of diagnosis more accurate and consistently reliable.
Slashdot Mirror
Time pressured clinicians overwhelmed with the exponential increase in bio-medical knowledge are desperately seeking clinical diagnosis decision support systems (CDDSS). Google, whilst extremely helpful in shifting the 'search paradigm', is of limited usefulness as a CDDSS. Google's accuracy of 58%, reported by Tang and Ng in the British Medical Journal, 10th November 2006, is less than that achieved by older generation rules based CDDSS and will not engender widespread adoption. Isabel (www.isabelhealthcare.com) is a web-based, point-of-care CDDSS designed used by healthcare professionals and has been extensively validated in clinical studies in terms of ease of use, accuracy and impact as a diagnosis reminder system. Isabel has been shown in published studies to be accurate in over 90 % of cases and to cause frontline physicians to consider an important diagnosis they should have considered in 1 in 8 cases. Isabel uses natural language processing algorithms (www.autonomy.com) 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. An independent study submitted for publication looked at Isabel's performance on the same set of cases 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. The aim of CDDSS is not to replace but to quickly and easily give the 'learned intermediary' (clinician) a differential diagnosis to consider. Sophisticated and validated CDDSS are now able to rapidly assist diagnosticians and make the cognitive process of diagnosis more accurate and consistently reliable.