228
T. Hanser et al.
Fig. 11.11 Estimating the confidence in an individual QSAR prediction can be achieved stepwise
with a gradual consolidation of the confidence level. The legitimate application of the model, the
reliability of the prediction and finally the assertiveness of the outcome are assessed in a cascading
process following a strict chronology
Fig. 11.12 Combining the decision domain framework along with information supporting the decision (interpretation of the prediction and access to the underlying evidence) provides a comprehensive decision framework and helps the end user take a well-educated decision
the underlying evidence help to extend the assessment of the prediction and grow
the confidence in subsequent decisions. This transparency of the prediction together
with the decision domain form a complete framework that helps the end user make
a well-informed decision (Fig. 11.12).
The different elements of this framework are captured by the TARDIS principle (Transparency, Applicability, Reliability, Decidability, Interpretability, Support).
T. Hanser et al.
Fig. 11.11 Estimating the confidence in an individual QSAR prediction can be achieved stepwise
with a gradual consolidation of the confidence level. The legitimate application of the model, the
reliability of the prediction and finally the assertiveness of the outcome are assessed in a cascading
process following a strict chronology
Fig. 11.12 Combining the decision domain framework along with information supporting the decision (interpretation of the prediction and access to the underlying evidence) provides a comprehensive decision framework and helps the end user take a well-educated decision
the underlying evidence help to extend the assessment of the prediction and grow
the confidence in subsequent decisions. This transparency of the prediction together
with the decision domain form a complete framework that helps the end user make
a well-informed decision (Fig. 11.12).
The different elements of this framework are captured by the TARDIS principle (Transparency, Applicability, Reliability, Decidability, Interpretability, Support).