Incomplete data was also suspected in some of the test values. For example, in case 125 the only X-ray finding mentioned was pleural effusion. This was a case of chronic heart failure, so there was probably evidence of pulmonary congestion in the lung fields as well. Since pleural effusion is a finding indicating more chronic disease, and in a sense more severe disease, only it was mentioned. However, there are other causes for pleural effusion, so with X-ray lung field indications false (assumption of completeness in a category), the program produced a hypothesis with pulmonary congestion false and a more unusual explanation for the pleural effusion.
The lack of times on test values, especially echocardiogram results, was a problem for the program. Often the echocardiogram results were years old, but they still provided useful information since many of the conditions only get worse. To account for this, initially the program considered any unspecified echocardiogram values as unknown rather than false. This proved to be a problem because the program would often propose a disease such as aortic stenosis as a cause, based primarily on its prevalence. Since aortic stenosis is a disease that progresses over years, even an old echocardiogram without aortic stenosis is good evidence against it. On the other hand, a disease such as mitral regurgitation can happen acutely. As a result, we developed a table of probabilities that the various disease states would appear as findings if the test were current, days, months, or years old. In the absence of information about the age of the test, months is assumed. Still, the normal ejection fraction (indicating normal systolic cardiac function) on an old echocardiogram in case 212 was enough to misdirect the program to account for the congestive findings with renal rather than cardiac causes.