My telemetry is fantastic, but designing the architecture to make best use of it is another matter! 

Jean-Gérard Napoleoni, Andreas Theodorou, Jonathan Pitcher

If the first line has a familiar ring to it, then this article is aimed directly at you. It describes what you should know before you design, build and equip your facility to conduct telemetry experiments The importance of good architecture is greater when the experiments involve a large experimental area and/or large populations. In-depth case-by-case design is needed to exploit existing data and network management technologies.

trends

Setups based on telemetry systems have become widely established  in safety pharmacology  and toxicology studies. There are obvious advantages associated with data acquisition from conscious, freely moving animals. Furthermore, in recent years, there has been a shift towards the use of non-invasive telemetry as an alternative to, or in combination with, invasive telemetry (use of implants). Laboratories are increasingly recognizing the distinctive advantages offered by non-invasive systems:

• no adverse physiological impact, costs or down-time related to surgery;
• no recurring costs;
• ability to measure multi-lead ECG, activity, skin temperature, respiration.

There is also increasing demand for large-scale facilities for running safety pharmacology or toxicology studies. Standard studies routinely involve from 4 to 40 subjects. Large animal facilities with a capacity of 200 to 400 animals, designed for running several small, medium and large studies simultaneously. It is becoming common to have as many as 100 to 150 subjects in 4 to 8 studies being run simultaneously.

The highest degree of flexibility, and thus of productivity, is achieved if non-invasive telemetry is possible from any pen. Each subject can then be allocated to a particular pen according to other requirements, without the additional concern of whether telemetry is available in that pen.

Telemetry provides continuous data for hours and days. This is wonderful from the scientific point of view since the quality of information that can be derived from such long recordings is greatly enhanced, as compared with, for example, one-minute ecg recording snapshots which, until only recently, were the rule in toxicology.

As a result, the volume of data generated by single studies has increased tremendously. Even more so now that synchronized video recording has become available. A video stream places strong demands on memory, even after compression based on the most recent algorithms. Nevertheless there is a strong argument in favour of video recordings that offsets the extra time and effort to have it installed: in effect, what is better than an actual image to assess animal behaviour when an outstanding cardiovascular or respiration event is detected in the analyzed data?

issues

Taking everything into account, one day of continuous recording from one subject will typically generate from 100 to 150Mb of physiological data, up to 1.5Gb if video is added. For a large study of 40 subjects, each of them undergoing 6 to 8 recording sessions of 24 hours, the total amount of data will range from 30Gb to 400Gb.

As can be seen, the large size of the setup and accompanying data further complicates what is already a formidable task faced by study directors and technical personnel. Some of their responsibilities are:

• To keep track of details from all subjects within each study and ensure that IDs of animals being recorded perfectly match study design;
• To monitor 100 to 150 subjects in to 4 to 8 simultaneous studies;
• To make sure that all hardware components function correctly, and that radio signal quality is maintained;
• To verify that signal quality is satisfactory, bearing in mind that, with an average of 6 traces per animal, there are 500 to 1500 traces in total;
• To minimize data loss by being able to respond rapidly to any problems, whether this entails reallocating hardware, radio frequencies, or subjects to pens.

Vendors of telemetry systems will need to respond to several challenges, described in the next section.

desiderata

It should be clear that designing large-scale setups does not mean simply scaling up what works for smaller ones. We attempt to describe below what constitutes the ideal architecture and tools to manage large safety pharmacology and toxicology studies using telemetry.

The system should be one that can be efficiently managed by a limited number of people, while providing high-quality data produced in compliance with quality assurance and regulatory rules. In more concrete terms, the data acquisition and management system should have the following features:

• centralized to minimize repetitive tasks, such as recording start and stop, real-time data control and visualization;
• yet accessible from a flexible number of entry points;
• secure to prevent any loss of data as well as loss of data traceability, from a regulatory point of view;
• powerful enough to cope with large volumes of data and high amount of analysis.

There are several technical considerations that will guide the choice of architecture and tools (hardware and software, respectively). Some of them are as follows:

• Flexibility of pen usage:
  the ideal system should make it easy to use any telemetry receiver in any pen, without having to go through complex and time-consuming system reconfiguration;

• System and network monitoring:
  the ideal system should provide the tools to quickly check overall system status and to provide fast and clear alerts whenever a piece of equipment becomes faulty;

• System local/remote usage:
  even though the ideal system is centralized, users will need to be able to access and operate it from a number of local or distant work stations and PCs - users will need to perform tasks while in the animal facility itself, but also from their desks in a possibly distant office building.

• System access:
  as your system will be fitted with a number of local or distant access points, you will also need to organize it as you would a subcomponent of the company network, by defining access rights, policies, etc.

• System dynamic load:
  you will need to estimate the maximum flow of real-time data that will be transferred from telemetry receivers to acquisition servers; this will determine the capacity of network links and backbone.

• Data storage capacity:
  you will need to verify that your data storage servers and the backup systems can handle packets of data larger than the total data volume of your largest studies;

• Link to your LIMS:
  you probably already have most of your in-house data going into a LIMS. Obviously the telemetry system, which is going to generate considerable amounts of data, should be able to directly and securely transfer its data to the LIMS.

• There is more to the setup than the installation itself. It is worthwhile checking whether a particular vendor's package provides the following features, which could prevent a lot of headaches, time and money down the line.  

• Carefully plan your installations:
  it is disproportionately expensive to add new instruments or cables once installation is completed. Thus, if your animal facility is about to be built or revamped, try to lay network cables ahead and reserve space for network switches and server rooms.

• Think big:
  there is stiff competition among vendors to bring new devices to market, for measuring an ever greater number of parameters; it is safe to predict that data volume will continue to grow.
  Your setup should be prepared for such changes. For example, your colleagues might say that it is enough to have a few video cameras for monitoring a small proportion of animals; it cannot be excluded, however, that video recording in all pens becomes the norm or even a legal requirement.

• Pre-production system:
  all systems require maintenance, updates and upgrades. Before implementing such changes, you need to ensure that they will not create unforeseen problems in your production system. You may thus consider having a small-scale pre-production system aimed at testing those changes before implementation in the production system.

• Local and distant maintenance:
  your system provider will most probably also offer to maintain the system. You should bear in mind that allowing distant maintenance is a win-win situation: if correctly implemented, you will have a less expensive yet more effective service.  Solutions exist to overcome the security issues associated with remote access.

• Finally, you should bear in mind that systems like the ones discussed in this article are new to this sector of the industry. Do not expect every vendor to be able to provide the solution you are looking for. Most will accept to demonstrate the system onsite or to let you test it. You should take advantage of such offers.

Last but not least...

It is all very well to design a state-of-the-art telemetry system that can safely and effectively record huge amounts of valuable raw data. The real goal is of course to discover the key events and features within the data.

Your project will only be a success if it is fitted with the tools that will make study organization and set-up smooth and easy, and then allow for fast and reliable analysis:   

• Study management tool: it seems essential that a single management tool be available to organize the data into a structured database in which any information related to the telemetry study will be stored. This management tool should allow the study directors and technical team to:
  • pre-define the studies: study details, information on sessions, and phases, animal details and identifier;
  • (on specific recording days) provide the list of subjects tagged for recording;
  • (prior to recording) provide an interface with tools used to authenticate subjects (RFID, other) and ensure that subjects about to be measured match those that were planned for that day.

• Strong analysis capabilities: once recordings have ended, the somewhat daunting task of data analysis ensues. It is relatively easy to detect a clear trend in heart rate; arrhythmias, which are very important to detect, are a different matter. Ascertaining that there are no arrhythmias buried in a 24-hour recording entails painstaking scrutiny of ecg traces on screen, taking 2 or 3 hours. Valuable time can be gained by using a set of tools to fully or at least partially automate data analysis.

You should exert all of your influence to make sure that the project team identifies analysis tools that can cope with the increased amount of data that the new system is about to generate! Remember that you need to be seen as a vital contributor to a successful experiment, not the person responsible for generating data that no-one can use!