205

metabolites (Schweitzer et al. 2019) whose strong or weak expression indicates a disease

or its (in)favourable course. Classification models (prediction models) are being devel­

oped for this purpose.

 

­

­

14.1  

Here we classify between positive (“sick”, alternatively: yes, 1, correct) and negative

(“healthy”, alternatively: no, 0, wrong) as follows:

• True Positive (TP; true positive cases): test and reference positive (test and refer­

ence “sick”)

• False Positive (FP; false positive cases): test positive, reference negative (test “sick”,

reference “healthy”)

• False Negative (FN; false negative cases): test negative, reference positive (test

“healthy”, reference “sick”)

• True Negative (TN; true negative cases): test and reference negative (test and reference

“healthy”)

In order to be able to assess how meaningful (accurate) a classification model (predictive

test) is, i.e. whether the classification made is correct or incorrect, there are various statisti­

cal quality criteria/measures (performance metrics). These are:

• Sensitivity (true positive rate, sensitivity; positives detected as positive)” =

St

ar

tF

raction normal upper T normal upper P Over normal upper T normal upper P plus normal upper F normal upper N EndFraction

• False positive rate (“false alarm”, positives that are actually negative) e

q

ua

ls

StartFraction normal upper F normal upper P Over normal upper T normal upper N plus normal upper F normal upper P EndFraction

= 1

specificity

• Specificity (negatives detected as actual negatives) =

St

ar

tF

raction normal upper T normal upper N Over normal upper F normal upper P plus normal upper T normal upper N EndFraction

• Positive predictive value (PPV, precision; probability of actually being posi­

tive) =

St

ar

tF

raction normal upper T normal upper P Over normal upper T normal upper P plus normal upper F normal upper P EndFraction

Table 14.1  Overview confusion matrix for a classification model

Reference

 +

−

Test (prediction)

+

TP

FP

−

FN

TN

14.3  Current Applications of Artificial Intelligence in Bioinformatics