Biomedical Engineering Reference
In-Depth Information
every time the process is executed. Once the important parameters are identified
and captured, the data from these parameters are represented in the form of various
charts and visualizations that can help in detection of trends and deviations
occurring in the process. This type of regular monitoring helps in decision making,
fault detection and root cause investigations whenever unusual process activity is
encountered [
11
].
2.2 Tools and Approaches
The approaches for process monitoring can be mainly divided into two broad
levels. Under each of these levels there are various tools, charts and graphs that
help to assess the process performance from various angles and provide useful
insights regarding areas of improvement. These two levels are:
• Level 1: basic visual monitoring.
• Level 2: statistics-based monitoring.
2.2.1 Basic Visual Monitoring
Basic visual monitoring mainly consists of basic displays of data that are easily
understood by a wider audience without having advanced knowledge of statistics.
The main focus of tools/displays at this level is communication of process trends to
easily identify unusual events and anomalies that occurred during the process
execution. This makes the process stakeholders (operators, manufacturing super-
visors) aware of and efficient at handling and troubleshooting of day-to-day
operational issues. Another benefit gained from the usage of these basic displays of
data is that they provide standard views to represent parameter trends, parameter-
to-parameter relationships and comparison of batches. This standardisation pro-
vides a single language to discuss data and process issues and thus enables healthy
communication among all stakeholders.
Some of the charts/displays that are useful at this level for monitoring of
pharmaceutical data are as follows:
• Parameter profiles.
• Bubble charts (scatter plots).
• Parallel coordinates.
• Treemaps.
Parameter profiles display the behaviour of a particular parameter over a period
of time. As pharmaceutical data consist of many parameters that follow a par-
ticular trajectory for a process, to achieve consistent output it is important to
monitor these parameters to ensure that they always follow the same trajectory for
a given process. There are two approaches to show this kind of data:
Search WWH ::
Custom Search