Predictive validity

A second objective is to produce behavioural changes in animals that are analogous to depression so that the model can be used to discover its neurobiological cause(s). This is a far more demanding problem and its success rests on satisfying at least three criteria (see Willner 1984) face validity (i.e. the behaviour looks like depression), construct validity (i.e. the causes and consequences of the behavioural change are the same as in depression) and predictive validity (i.e. the behaviour is reliably prevented only by drugs which have antidepressant effects in humans). 

Morisky, D.E., Green, L.W. 8c Levine, D. M. (1986). Concurrent and predictive validity of a self-reported measure of medication adherence. Med. Care, 24, 67-74. 

Molnar, L. et al., A neural network-based classification scheme for cytotoxicity predictions Validation on 30,000 compounds, Med. Chem. Letts., 16,1037,2006. 

There are three generally accepted criteria for validating animal models for human psychiatric disorders face validity, construct validity, and predictive validity. Face validity refers to the outward appearance of the model, i.e. does the animal s behavior adequately reflect the human behavior being modeled In this dimension, anxiety models have a clear advantage over other psychiatric models it is usually quite apparent if an animal is frightened, whereas it is a much more difficult to assess whether an animal is displaying psychotic-like or depressive-like behavior, for example. 

Predictive validity is the ability of a model to predict the effect that pharmacological or other manipulations will have on the condition being modeled. This criterion can present a real difficulty, in that drug development is often dictated by animal models. For example, if a given model only detects a subset of effective compounds (i.e. those belonging to a specific chemical class), then useful candidates will be discarded long before clinical trials, and the flaw in the model s predictive validity will not be discovered. Thus, the possibility that a model will yield false negatives cannot be ruled out. 

Medicinal Chemistry Letters, 16, 1037 (2005). A Neural Network Based Classification Scheme for Cytotoxicity Predictions Validation on 30,000 Compounds. 

REACH is an extraordinarily ambitious program. There are discussions underway regarding proposals to limit the numbers of chemicals to be subjected to these requirements. The potential for toxicological testing on a massive scale raises questions about the availability of facilities to carry out such tests, and runs counter to the objective of reducing the numbers of animals used for such purposes. The need to accomplish REACH objectives without the overuse of laboratory animals has promoted discussion and research regarding the use of alternative methods to collect the necessary data tools such as in vitro tests and quantitative structure-activity relationships (QSARs) are being promoted, and this has led to substantial research efforts to test their predictive validity. Time will tell where all of this activity leads us. 

Fig. 6.4 a Characteristic times (r) at 330 K. VI filled triangle, PI in PI/PVE empty triangle, PVE in PI/PVE empty circle, WE filled circle, b (r)Q vs. Q. Solid lines Rouse prediction valid for low Q dashed and dashed dotted lines are guides for the eye. (Reprinted with permission from [47]. Copyright 2000 The American Physical Society)... 

Figure 3.10 Hallmark signature of significant sampling bias as revealed in chemometric multivariate calibrations (shown here as a prediction validation). Crab sampling results in an unacceptably high, irreducible RMSEP. While traditionally ascribed to measurement errors, it is overwhelmingly due to ISE.

Table 4.25 summarizes the prediction validation tools discussed in this section. The first column in the table lists the name of each tool and the second column describes results from both well-behaved and problematic data. 

Three prediction diagnostic tools are discussed below and a summary is found at the end of the section in Table 5-12. As discussed in the introduction to this section, one disadvantage of models with few variables is that the prediction validation diagnostics are limited. 

The ideal animal model for any human chnical condition must fulfill three criteria (McKinney and Bunney 1969) (1) pharmacological treatments known to be effective in patients should induce comparable effects in the animal model (predictive validity) (2) the responses or symptoms observed in patients should be the same in the animal model (face validity) (3) the imderlying rationale should be the same in both humans and animal models (construct validity). In other words, the ideal animal model for anxiety has to respond to treatment with anxiolytics such as benzodiazepines with reduced anxiety it has to display defense behavior when confronted with a threatening stimulus the mechanisms underlying anxiety as well as the psychological causes must be identical. 

Hijzen TH, Houtzager SW, Joordens RJ, Olivier B, Slangen JL (1995) Predictive validity of the potentiated startle response as a behavioral model for anxiolytic drugs. Psychopharmacology (Berl) 118 150-154... 

Hollis, C. (2000) Adult outcomes of child- and adolescent-onset schizophrenia diagnostic stability and predictive validity. Am J Psychiatry 157 1652-1659. 

Voelker, S.L., Lachar, D., and Golawski, L.L. (1983). The personality inventory for children and response to methylphenidate ptelim-inary evidence for predictive validity. J Pediatr Psychol 8 161— 169. 

The combination of sophisticated challenge paradigms with biomarker detection provides new and technically advanced quantitative opportunities for PoC trials. However, objective and balanced evaluation of PoC and PoM paradigms sometimes may be hindered by publication bias trials with negative or inconclusive outcome tend to be published less frequently than studies with positive outcome, a fact that can mislead readers with regard to the predictive validity of certain models. 

Gaither CA. 1998b. Predictive validity ofwork/career-related attitudes and intentions on pharmacists turnover behavior. /Pharm MarketManag 12 3. 

Crompton, J. and Love, L.L. (1995) The predictive validity of alternate approaches to evaluating quality of a festival. Journal of Travel Research 34 (1), 11—25. 

Whilst it is relatively easy to carry out, the correlation between the test results obtained on a candidate and the actual performance on the job, known as predictive validity, is variable and depends on what characteristics are being looked at in the test. The correlation coefficient for cognitive tests has been assessed as 0.35 and for personality tests as low as 0.15 [A-4], However, when such tests are well done and analysed by a suitably qualified person they are very useful in checking or supporting the conclusions of the interviewing team, particularly where there is a divergence of opinion, but they should never be used as the sole basis for selecting a candidate. 

---