Search

We provide a novel method for validating any purported set of the four most prominent indicators of diagnostic testing (Sensitivity, Specificity, Positive Predictive Value, and Negative Predictive Value), by observing that these indicators constitute three rather than four independent quantities. This observation has virtually been unheard of in the open medical literature. We defined two functions, which serve as consistency criteria, since each of them checks consistency for any set of four numerical values claimed to be the four basic diagnostic indicators. Most of the data we came across in various Saudi medical journals met our criteria for consistency, but in a few cases, there were obvious unexplained blunders. We relate our present findings to the more general issue of detection and ramifications of flawed, fabricated or wrong data. We observe that the research field handling the detection of flawed data is still in its infancy, and hope that this field will reach maturity very soon.

This chapter deals with the reliability of a multi-state delivery network (MSDN) with multiple suppliers, transfer stations and markets (depicted as vertices), connected by branches of multi-state capacities, delivering a certain commodity or service between their end vertices. We utilize a symbolic logic expression of the network success to satisfy the market demand within budget and production capacity limitations even when subject to deterioration. This system success is a two-valued function expressed in terms of multi-valued component successes, and it has been obtained in the literature in minimal form as the disjunction of prime implicants or minimal paths of the pertinent network. The main contribution of this chapter is to provide a systematic procedure for converting this minimal expression into a probability-ready expression (PRE). We successfully extrapolate the PRE concept from the two-valued logical domain to the multi-valued logical domain. This concept is of paramount importance since it allows a direct transformation of a random logical expression, on a one-to-one one, to its statistical expectation form, simply by replacing all logic variables by their statistical expectations, and also substituting arithmetic multiplication and addition for their logical counterparts (ANDing and ORing). The running example used for demonstrating PRE derivation in this chapter is admittedly long and tedious. The statistical expectation of the expression is its probability of being equal to 1, and is simply called network reliability. The proposed method is illustrated with a detailed symbolic example of a real-case study, and it produces a more precise version of the same numerical value that was obtained earlier by an alternative means. This chapter is a part of an ongoing activity to develop pedagogical material for various candidate techniques for assessing multi-state reliability.