A few years ago my friend John Simpson gave me a copy of his data records of epileptic seizures that he had over the period January 2004 to June 2011. For each seizure he recorded whether it was grande mal or lesser seizure, which were recorded as 2 or 1 respectively. All other days were recorded as 0. At the time I did some cycles analysis of his data and there were very significant cycles, and lots of them. Recently I asked him if I could put the information on Cycles Research Institute blog and he said yes.
On the face of the data record, the seizures seems to be rather randomly scattered in time. However a spectral analysis shows that there is considerable order in the timing. Here is the spectrum:
The figures in red are the cycles period which were significant at the 0.05 level. Probably many of these are not real, but many will be. The outstanding period is the 16.672 day one. It has p=0.000128 which means that such a cycle would be present only one time in 8,000 if the data were random.
The cycles periods found do not generally correspond to any well known cycles periods such as solar or lunar. Perhaps the cluster around 7 days is related to the Sun because the Earth is hit by more solar wind particles typically every 7 days due to the Sun rotating.
This would tend to suggest that the periods are mostly due to internal processes in John’s body. It is surprising that such a cycle could be very regular, and yet the 16.672 day cycle is so regular that the seizures are concentrated in half the cycle and especially in a few days, with one week of the cycle rather free of seizures.
This graph is prepared by wrapping time in a loop every 16.672 days as it were, so in 2004, January 1 is on the left through to 17th on the right. The 18th will appear about 1/3 of the way between 1 and 2 January and subsequent days at one day interval. After each multiple of 16.672 days the data is re-entered from the left. I call this type of analysis “wrap”. It is useful for seeing the average shape of a cycle.
The scale at the left is the probability of a seizure on that day of the cycle, without grand mal counted as two. It can be seen that the peak days are over 5 times as likely to have seizures as the trough days, and more than 3 times as likely as the whole trough week.
A search of the internet for epilepsy cycles brings up many pages. For women about 10% to 12% have a form called catamenial epilepsy which follows the hormonal cycle. No monthly cycle is noticed for men. It is thought that hormones play some part, but this is not fully understood.
It is reported that seizures are often clustered in time, there being a burst in a short period and then none for a while. Such distribution will make cycles appear to be more significant than they are, or even make them appear when they are not. However that is not true of the dominant cycle here.
I found a report of one young man having a cycle of 10-12 days in seizure clusters. That is not so greatly different from John’s. He did not mention to me that he had noticed any periodicity. It may be that the seizures are not happening every cycle so the intervals might be 17, 33, 17, 50, … days which would not be obviously apparent.
John reports to me that he no longer has any grande mal and no longer keeps a log. Thank you for sharing your data with us John.