... elderly^1.1

Throughout this report the elderly or the aged will refer to people of age greater than or equal to the age of 65.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... (CSSE)^1.2

Faculty of Information Technology

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... (ECSE)^1.3

Faculty of Engineering

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... fingerprint^2.1

Ambulation monitoring systems such as this could conceivably be employed to extract and recognise the finer details of an individual's walking patterns.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... males^2.2

Although it should be noted that females have a longer life expectancy, with male:female ratio in these studies of 1:1.4.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

...%)^2.3

Following transport-related injuries (29%) and suicide (32%).

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... network^3.1

Also known as a Bayesian network, probabilistic network, causal network, causal graph or knowledge map.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... system^3.2

Also known as the system's `state space'.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

...B^3.3

This relationship may be a simple correlation, or a stronger causal relationship (B causes A). The effect of different causal relationships is investigated by the support-2 DBN model in section 8.4.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... ^3.4

Also known as Bayes' law or Bayes' theorem

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... software^4.1

Although the `record-and-playback' data-logging technique using Compumedics' playback software was investigated in Jean McInerney's project (Section 1.4.1).

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... future^4.2

Additionally, it could be used with any of the datalogger's communications methods, as it does not require the data to be recorded in real time.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... `polling'^4.3

Waiting for data to be written to a file by SiestaView so that the DBN software could read it for processing.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... system^4.4

This version of the software can be run in both Linux (the primary development environment), and Windows on the MS-DOS command-line.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

...Compumedics^5.1

Compumedics Pty. Ltd. is a Melbourne-based engineering company, that designs and produces hardware and software for Medical research and analysis. Compumedics is located at 1 Marine Parade, Abbotsford, VIC 3067, Australia. Further information is available on the Internet: http://www.compumedics.com.au.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... LAN^5.2

Local-area network.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... monitoring^5.3

Although they were used extensively by Jean McInerney (Section 1.4.1) and other projects incorporating the Siesta.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... peak-to-peak^6.1

The datalogger is capable of measuring up to 500 millivolts peak-peak, but as custom sensors were being used - which may be subjected to higher body weights and more pressure than they were tested with - it was decided to exercise caution and aim for a smaller output range.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... sole^6.2

Rubber-soled shoes were used during testing.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... slices^7.1

The number of slices added to the network as a result of new evidence is usually one, but the ability to expand multiple slices per evidence entry was added to test handling of missing data (see Section 9.5).

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... states^8.1

`Impossible' conditioning states were combinations of parent states which could not be realised in practice. For example, the combination Support<sub>N</sub>=left, Action<sub>N</sub>=ldown was considered impossible, because if the support state is `left', the left foot is already on the ground and the `ldown' action doesn't make sense.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... strikes^8.2

This introduced the problem of not detecting foot-strikes that didn't involve the heel, which may occur during running or in `toe-walkers'. As this network was designed to detect normal heel-strike walking, it didn't bother handling this situation, but the configuration could easily be updated to trigger foot-down evidence on heel-down or toe-down events.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... support^8.3

The sequence of supports in a gait cycle can be considered `second order Markovian', as the current support usually depends of the previous two support states

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... `stumble'^8.4

The `stumble' state was used to represent belief in the occurrence of the stumbling or stepping strategy balance response, discussed in Section 2.3.2.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... individuals^8.5

A `small' interval would be expected to occur in running or jogging (along with double support states in the gait cycle being replaced by no support), and is used by later networks to detect such modes (Section 8.8).

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... uncertain^8.6

All states tend towards the same value, so no clear preferred state can be established.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... second^8.7

The timeout filter also takes input from the left-toe signal (LTRAW). Nothing is actually done with this data; as LTRAW outputs once every sample, it is merely used to provide a timing signal so the timeout filter can update its internal counters.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... recording^9.1

Running was not recorded due to the physical restrictions imposed by the equipment and cables attached to the subject during recording.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... software^10.1

Which was not implemented due to complications arising from way Netica stores its networks.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... events^10.2

An idea resulting from controversies in the 2000 Olympic games walking races.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... description^12.1

With the exception of inside feature extraction function descriptions, comments may be placed anywhere.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

... 64Hz^12.2

Future software implementations could easily incorporate methods to automatically determine the input channel configurations.

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.