Article Preview
TopEditorial
Human ability to reason and comprehend the world requires the coherent activity of billions of neurons. At the same time, the biological existence is rooted in seamless interactions between thousands of genes and metabolites within the cells. These are complex systems which capture the fact that it is challenging to derive their collective behavior from the knowledge of the system's components. Given the critical role, complex systems play in daily life, their understanding, mathematical description, prediction, and eventually, control is one of the significant intellectual and scientific challenges of the 21st century. The advent of Network Science is a clear demonstration that science can live up to this challenge (Barabási, 2016).
Network Science is rooted in the discovery that despite the apparent diversity of complex systems, the structure and the evolution of the networks is driven by a standard set of fundamental laws and principles. Therefore, notwithstanding the striking differences in form, size, nature, age, and scope of real networks, most networks are driven by common organizing principles.
Management also tends to rely on the informal network, capturing who communicates with whom. Furthermore, accurate mapping of organizational networks can help to identify individuals who play an essential role in bringing different departments and products together and expose the lack of interactions between critical units. There is also increasing evidence that employee's productivity is determined by the position in the informal organizational network (Wu et al., 2008). Therefore, numerous companies offer tools and methodologies to map out the actual structure of an organization. These companies offer a host of services, from identifying opinion leaders to reducing employee churn, optimizing knowledge and product diffusion and designing teams with the diversity, size and expertise to be the most effective for specific tasks.
In the same direction, modern healthcare has explicitly accepted and internalized the idea of a network by adopting:
- •
The transparency of quality.
- •
The infrastructure of cross-continuum connectedness.
- •
The renaissance of accessibility.
- •
The integration of ecosystem partners.
From a contrasting perspective, Statistical Process Control (SPC) is an optimization philosophy and a set of procedures for continuous improvement which was developed in the 1920s by the physicist Walter Shewhart (Thor et al., 2007). The SPC approach has been founded in the theory of variation. It uses the Shewhart charts, which are more commonly called 'control charts', and incorporates the concepts of:
- •
The analytic study, process thinking, and prevention.
- •
The stratification, stability and capability.
- •
The experimentation, data collection, and measurement.
However, under the circumstances of network science, it is noticeably obsolete.
SPC charts in the presence of autocorrelation, and data pattern, do not detect and classify the fault correctly. As a result, many researchers propose substitute methods for process monitoring. Neural networks are massively parallel computing mechanisms emulating the human brain, which have a satisfactory performance when they are used for a wide variety of applications. In recent years, they have been applied in statistical process control (SPC).
Psarakis (2011) discusses issues concerning the combination of Statistical Process Control (SPC) and neural networks as well as the use of neural networks in:
Moreover, Fu (2013) acknowledges that network surveillance algorithms are increasingly important due to their ability to monitor a wide variety of data. Traffic metrics count data that display a non-stationary pattern in their mean structure. In this context, he develops three tracking statistics for anomaly detection. Two of the statistics are variants of a Bartlett-type sequential probability ratio approach: