"... In PAGE 11: ... In order to show the advantage of using LVEF together with HRT instead of LVEF alone we used the percentage of patients in the high risk group for comparison. Table3 gives the corresponding cut-points for LVEF together with the number of deaths in the high risk group. Table 3: Comparison of the prognostic scores based on LVEF alone or LVEF in combination with HRT... ..."

"... In PAGE 18: ...2. Example 3: Prognostic risk factors for Czech auto workers The data in Table8 come from a prospective epidemiological study of 1841 workers in a Czechoslovakian car factory, as part of an investigation of potential risk factors for coronary thrombosis (see Edwards and Havranek, 1985). Here we build on the analyses reported in Dobra and Fienberg (2003) and Fienberg (2004), interpreting them in the context of searching for association rules.... In PAGE 18: ... Here we build on the analyses reported in Dobra and Fienberg (2003) and Fienberg (2004), interpreting them in the context of searching for association rules. In Table8 , A indicates whether or not the worker smokes, B corresponds to strenuous... ..."

"... In PAGE 9: ...ime requirement, invasiveness, cost, etc.). However, because these measurements are currently the object of extensive research, which may make them more useful in the near future they have been discussed in section 3.6 and listed in Table4 together with an assessment of their clinical value and limitations. The issue is further discussed in Section 3.... In PAGE 16: ... In elderly hypertensives, cognitive tests may help to detect initial brain deterioration. Table4 summarizes availability, prognostic value and... In PAGE 18: ... Though a wider clinical use of pulse wave velocity and augmentation index measurements may add further precision to the assessment of arterial damage, the avail- ability of these technique is largely limited to research centres. As shown in Table4 several other methods for detecting vascular organ damage cannot be supported for clinical use for a variety of reasons. An increase in the wall to lumen ratio of small arteries can be measured in subcutaneous tissues obtained through gluteal biopsies.... ..."

"... In PAGE 17: ...Table3 remained in which feed-forward neural nets were used for classi cation, some of them considered statistical methods as alternatives. The topics of these articles were diagnosis of cancer (23 articles), automatic tumor grading (6 articles) and prognosis (10 articles), furthermore 3 reviews and 2 commentaries.... In PAGE 17: ... It became apparent to us that some mistakes occurred over and over again. These mistakes are indicated in Table3 and are described in detail in the sequel. 1) Mistakes in estimation of misclassi cation probabilities a) Biased estimation The dramatic underestimation of the apparent error rate seems to be well known, in none of the articles the apparent error rate was used as an estimate of the misclassi - cation probability.... ..."

"... In PAGE 12: ...7% of the time, and lymph node status only 0.5% of the time ( Table5 ). The Kaplan-Meier (16) plot of the RSA data derived by cross validation is shown as Figure 2.... ..."

"... In PAGE 12: ...7% of the time, and lymph node status only 0.5% of the time ( Table5 ). The Kaplan-Meier (16) plot of the RSA data derived by cross validation is shown as Figure 2.... ..."

"... In PAGE 5: ...001 and P H11005 0.038, respectively; Table5 ). The differences between the marker model versus TNM stage (P H11005 0.... ..."