Systems Biology and Ratio-Based, Real-Time Disease Surveillance

SummaryMost infectious disease surveillance methods are not well fit for early detection.To address such limitation, here we evaluated a ratio- and Systems Biology-basedmethod that does not require prior knowledge on the identity of an infectiveagent. Using a reference group of birds experimentally infected with West Nilevirus (WNV) and a problem group of unknown health status (except that theywere WNV-negative and displayed inflammation), both groups were followedover 22 days and tested with a system that analyses blood leucocyte ratios. To testthe ability of the method to discriminate small data sets, both the reference group(n = 5) and the problem group (n = 4) were small. The questions of interest wereas follows: (i) whether individuals presenting inflammation (disease-positive orD+) can be distinguished from non-inflamed (disease-negative or D ) birds, (ii)whether two or more D+ stages can be detected and (iii) whether sample sizeinfluences detection. Within the problem group, the ratio-based method distin-guished the following: (i) three (one D and two D+) data classes; (ii) two (earlyand late) inflammatory stages; (iii) fast versus regular or slow responders; and(iv) individuals that recovered from those that remained inflamed. Because ratiosdiffered in larger magnitudes (up to 48 times larger) than percentages, it is sug-gested that data patterns are likely to be recognized when disease surveillancemethods are designed to measure inflammation and utilize ratios.IntroductionNew infectious disease surveillance methods are needed(McClintock et al., 2010). The fact that most surveillancemethods are not designed for early detection (Bravataet al., 2004) justifies the development of new methods. Tothat end, new systems could be designed to facilitateprompt responsiveness – a concept associated with both aminimal data set (the smallest data set that can supportdiagnosis and decisions) and a ‘critical response time’ (theminimal time required to deploy interventions, Rivas et al.,2003; Kloeze et al., 2012).However, there are additional reasons why newinfectious disease surveillance systems are needed, includinga particular challenge: to choose an appropriate method,the identity of the infectious agent should be known –which cannot be known unless a specific test is used. Thecircularity of this problem leads to the following question:Is it possible to develop generic disease surveillance meth-ods, even when the infecting cause of disease is unknown?One way to answer this question is considering the factthat the immune system displays a ‘mirror image’ of theinfection – an ‘image’ that, although imperfect and not spe-cific, can tell, rapidly, whether leucocyte profiles departfrom the healthy profile. In the absence of an alternativeexplanation, when an inflammatory or immune response isdetected, such information may indicate an infectious pro-cess. Assessments of the immune response are used whenmicrobiological test results display a poor sensitivity, forexample, leucocyte counts and differential percentages arecurrently used to monitor pulmonary infections in children(Brennan et al., 2008). Similarly, early inflammatoryresponses are also used to evaluate vaccines (Mastelic et al.,2013).