|
|
|
|
|
|
by jononor
708 days ago
|
|
|
I have found two datasets of cats wearing accelerometer, none of these contain seizures. However they would be valuable in understanding the normal activity, and the data coming from that. Especially the natural variability, since one needs to discriminate the seizures from that.
"Domestic cat accelerometer data calibrated with behaviours" (Dunford, 2024):
https://datadryad.org/stash/dataset/doi:10.5061/dryad.q2bvq8... "The Use of Triaxial Accelerometers and Machine Learning Algorithms for Behavioural Identification in Domestic Cats (Felis catus), A Validation Study" (Smit, 2023):
https://www.mdpi.com/1424-8220/23/16/7165
https://figshare.com/articles/dataset/R_dataframes_of_annote... And I found one paper on real-world detection of seizures in dogs: "Evaluation of a collar‐mounted accelerometer for detecting seizure activity in dogs" (Muñana, 2020).
They conclusion was: "Generalized seizures in dogs can be detected with a collar-mounted accelerometer, but the overall sensitivity is low."
https://onlinelibrary.wiley.com/doi/full/10.1111/jvim.15760
Their methodology for the model development seems generally sound. However, it does not seem like they have spent a lot of time on it, or are ML/DSP specialists. So there might be considerable room for improvement. If one could get a hold of this dataset, it might be possible to work on improving the detection method. With the goal that this would be highly transferrable to cats. The lack of data for cat seizures is a challenge in this endeavour. However, there seems to be quite some videos of such events on Youtube. At the very least, they can be used for qualitative insights. But an idea would be to use motion tracking on the images to simulate an accelerometer, and generate a dataset from that. I have seen a paper on this kind of approach in another setting, but I cannot quite recall where to find it. |
|
|
Not so much as you might think.
The approach I'd take (from airborne geophysics) is to treat the datasets as "environmental normal" and pull out tens of thousand of (overlapping) 5 minute data runs and treat those as input vectors to an SVD (Singular Value Decomposition) reduction which becomes the kernal of monitoring going forward.
Next rig up the cat in question with accelorometers and record data just as was done in the prior datasets.
Your input now is a continuous pipeline (say every 20secs) of "the last five minutes of data" as a vector - reduce each vector to kernal (spanned by the basis for the "normal" dataset) + noise (doesn't match the normal span).
There will be a regular amout of "noise"; seizures and unusual behaviour should spike the amount of noise and deserve attention.
After a bit, you'll know what you're looking for (/cough /handwave /details).
This, more or less, is how "out of band" signal is found in 256 channel radiometric spectrometer surveys - primed with a back catalog of hours of regular boring survey data and trained to look for the abnormal.