CLASSIF1 Classification

• < AML     < fig.2, > fig.4

fig.3: Predictive Disease Classification Masks

- The classification results of fig.2 were obtained with the most discriminatory data patterns (disease classification masks) selected by the CLASSIF1 data sieving algorithm from the database parameters of fig.1.
- Patient age as well as %CD4 and %CD45 positive AML blasts were common for the 5-year and 2-year survival masks.
- Predictive and prognostic data patterns show only partial similarity (CD4, CD11b, CD13, HLA-DR)
- Only a relatively small part (10.6%, 16.7%) of all parameters carried the predictive information.
- No cytogenetic parameters are contained in the predictive data patterns. The reason for this is probably that only about half of the patients (L2) express cytogenetic abnormalities while all AML blasts express CD antigens on their cell surface.

• < AML     < fig.2, > fig.4