Q: I have been using the new automated Before Flexion functionality in LL2017 where you can use your baseline straight evaluation as your baseline before flexion in the comparison report. I have noticed on a few horses that the autogenerated 8 stride baseline looks quite different than the original straight line trial of 25 strides? Why is that?
A: This is likely due to stride to stride variability within the trial. One segment of the data is different than the next segment of data. This could be explained, for instance, by the lameness trotting away in one direction being a little bit different than when it trots back. In a regular straight line trial of trotting the horse down and back or down and back twice (getting ~ 25 strides), the mean will be of all those strides. But, if you just take the first segment of the horse trotting away from you, the mean of those strides might look different. When the program creates the autogenerated baseline, it takes the FIRST 8 strides of that straight line trial. These strides are likely when the horse was just trotting away. By taking the FIRST 8 strides of your baseline straight trial, we are creating the closest conditions to which to compare your “after flexion” test, which presumably are the first 6-14 strides of the horse just trotting away from you.
This is one of the reasons why, prior to LL0217, the recommendation was to collect a separate SHORT baseline – and not to compare a flexion test of 6-14 strides to your baseline straight of 25+ strides. Comparing like strides under like conditions is most appropriate.
Below is an example of the referenced question. Fig. 1 is the baseline straight evaluation of approximately 25 strides. This data was collected in two segments – the horse trotting away and trotting back.
Fig. 2 is the autogenerated 8 stride Before Flexion (left) alongside the After Flexion trial (right). The autogenerated Before Flexion consists of the FIRST 8 strides of the baseline straight trial of ~25 strides. As you can see, the 25-stride straight line forelimb plot in Fig. 1 and the autogenerated 8 stride Before Flexion forelimb plot in Fig. 2 (left side) do look different. The data is more variable in the baseline straight of 25 strides. There are asymmetric strides on the LF and asymmetric strides on the RF. Because of this, the mean Diff Min Head is lower, and the standard deviation of the Diff Min is higher (because the asymmetry is flipping sides, Diff Min is negative on some strides and positive on others). This also drives down the Vector Sum.
However, if we take only the first 8 strides of that trial, represented in the autogenerated Before Flexion baseline (Fig. 2 left side), you see that all strides were asymmetric on the LF. The mean Diff Min Head is much less variable, and the resulting Vector Sum is higher.
The After Flexion consisted of 13 strides. Not identical numbers of strides, but close. This was also the horse just trotting away. You can see that matching the conditions of the strides included in the analysis (horse just trotting away) is a much more appropriate comparison than comparing the flexion against all 25 strides -- where, in this case, the lameness was different depending on which way the horse was traveling.
While using the autogenerated baseline Before Flexion saves time by eliminating the need for a separate trial, keep in mind you will get the truest comparison by collecting a separate before-flexion trial of 6-14 strides right before your flexions. This is recommended particularly if the horse has been lunged for a while after your straight line evaluation, or you have performed other activities that may have changed the lameness between running your initial straight line trials (remember it is important to get 2 baselines!) and starting your flexions. In these cases, it is best to collect a separate “Before Flexion” trial of 6 to 14 strides versus using the autogenerated Before Flexion.