Using the clinical features of these cases you can evaluate your own diagnostic skills, compare and calibrate your diagnostic performance to that of the physicians at MGH. If you are registered with Isabel, you can use the diagnosis reminder system and expand the list of likely suspects. Click through to the synoptic and/or detailed diagnosis specific and up to date knowledge from Isabel's database of textbooks and journals.

• Click on each case below and you will find clinical features of the case extracted by our team.
• Alternatively, if you have full-text access to NEJM, you can use the link in the document, read the case presentation and extract the clinical features yourself.
• Before you get the differential diagnosis of each case, why not construct a complete differential diagnosis yourself and compare and calibrate your performance to that of the experts at Massachusetts General Hospital [MGH].
• How often did you consider the final diagnosis of the case?
• You will also be able to evaluate Isabel's differential diagnosis results - in screen dumps.
• Compare your performance to that of Isabel's.
• If you are registered with Isabel, you can also cut & paste the extracted clinical features into Isabel and let the diagnosis reminder system jog your memory with a list of likely suspects !

• Click on each case below and you will find the clinical features extracted by our team.
• See how Isabel performs on these diagnostically challenging cases.
• The analysis of Isabel's performance includes :

• Does Isabel suggest the final diagnosis of the case?
• How many of the MGH differential diagnoses does Isabel suggest?
• How well does Isabel perform when clinical features are extracted [negation avoided & numerical values interpreted]?
• How well does Isabel perform when you 'cut & paste' the entire presentation into Isabel's query box?

• Compare Isabel's performance to that of the collective wisdom at MGH.

In July 2005 Mark Graber et al published an article Diagnostic Error in Internal Medicine which found that cognitive factors contributed to diagnostic error in 74% of cases studied. The most common cognitive problems involved faulty synthesis. Premature closure, i.e. the failure to continue considering reasonable alternatives after an initial diagnosis was reached, was the single most common cause.

In November 2005 Graber and colleague, found that when key clinical features from 50 challenging CPC cases, reported in the New England Journal of Medicine, were entered into Isabel , the system provided the final diagnosis in 48 cases (96%). This degree of accuracy by a diagnosis decision support system is unprecedented.

Using 'cut & paste' whole text data entry of the entire presentation, the correct diagnoses was suggested in 37 of the 50 cases (74%), a sensitivity that exceeds that of earlier CDSS systems, all of which require entry of discrete key findings