ESTIMATION OF THE PROBABILITY OF ERRONEOUS INTERPRETATION OF THE RESULTS OF PCR ANALYSIS. PART 1. FAILURES OF TECHNICAL DEVICES

Based on recognized probabilistic models of the distribution of time periods to the first failure of technical devices, an estimate of the probability of obtaining an erroneous result of genetic analysis during a real-time polymerase chain reaction is given. Models of the probability distribution density of the time periods to the first failure from a set of Weibull distributions with different parameters are used. Most of the time estimates are based on the Weibull distribution with parameter 1, which coincides with the most recognized exponential law of the distribution of time periods to failure. Cases of short uptimes (from 100 to 1200 hrs) of the device are investigated. Using the concept of a "weak link" as the most likely failing unit, or an element in this unit, it is confirmed that in the conditions under consideration, in more than 65% of cases, the failure of the device is caused by the failure of this element. The often used hypothesis of failure of an only element, for example, for constructing a table of fault functions, is insufficiently justified even for short uptimes of the device. The estimates obtained can be a guideline for predicting uptime in conditions when replacing elements (unit) is practically impossible. In the case of using a probabilistic model — the Weibull distribution with parameters 1.5 and 2, which implies an increase in the probability of failure over long periods — the estimated values of the probability of failure for continuous operation time periods up to 700 hrs are reduced by about half compared to the exponential law.