JMP® Student Editionについてご質問があります。
このソフトで、RRI（心拍間隔）のエクセルのデータからストレス指標のLF/HF（低周波成分/高周波成分）を知ることは可能かどうということと、もし可能であれば、その手順もご教示頂きたいです。
I have a question about JMP® Student Edition.
Is it possible to use this software to determine the stress indicator LF/HF (low frequency component/high frequency component) from Excel data on RRI (heart rate interval)? If so, please explain the procedure.
My name is Ono and I work as a senior tester in the JMP Japan division of SAS Institue Japan.
I myself only learned about the "HF/LF ratio" from your question, so my answer will be quite vague.
Basically, calculating the HF/LF ratio is not provided as a standard feature of JMP, and requires processing that is not even available as a programming function, so I think it would be quite troublesome to do this in JMP.
First of all, there is no way to calculate the HF/LF ratio, a stress index, from RR interval data using just a single function or mouse operations in JMP or JMP Pro (hereafter collectively referred to as "JMP"). However, JMP Student Edition has the functions of JMP Pro. There are also other functions, such as BMI calculations, that are not provided in the GUI in JMP, and if you want to perform such calculations in JMP that are not provided as standard, you will need to program it yourself.
(There is no add-in that can perform the "HF/LF ratio" calculation, but there are several free programs available on the JMP Marketplace website ( https://marketplace.jmp.com/ja) that use a function called "add-in" and are provided as is.)
If you want to calculate the HF/LF ratio in JMP, you will need to program it (write the code yourself) in the following way:
- Programming is done in a language specific to JMP called JSL (JMP Scripting Language).
- Programming in Python (JMP 18 and later) within JMP.
There are several methods for processing and calculating the HF/LF ratio from RR interval data. There may already be a standard method that is best, but when I asked the following prompt on ChatGpt, several papers were introduced.
"There are various methods that can be considered for calculating the LF/HF ratio from the RR interval, such as (a) linear interpolation or spline interpolation to make the data evenly spaced, (b) whether to use the value as is or smooth it to calculate the power in the frequency domain, and (c) how to finally numerically integrate the power spectrum. Are there any papers that compare these methods?"
For the simple calculation methods used in the papers to respond to the above prompts (for example, creating equally spaced data by linear interpolation, performing a raw fast Fourier transform, and product distribution using the trapezoidal rule), there are several functions available in JSL (linear interpolation can be performed with the Interpolate function, and fast Fourier transform with FFT). However, for other processing, there are no specific functions available in JMP, and you will need to write those parts from scratch yourself. If you are not sure which calculation method is best, it is better to try all the methods and show that the results are the same, so I think it would be time-consuming to write it in JSL for JMP.
In addition to the details that need to be taken into account when making calculations like those mentioned above, you should first consider whether the LF/HF ratio itself is a valid indicator and whether it matches the subject of your estimation. I think it would be quicker to seek the opinion of an expert who has looked at actual ECG data in the frequency domain (not just the time domain) (if you do not know how to interpret data in the frequency domain and simply calculate the LF/HF ratio using any method, you may be left with concerns about whether the calculation is correct, etc.).
(Note that if you want to fit an ARIMA model and look at the power spectral density, you can do so in the Time Series platform. However, this platform does not calculate LF power, HF power, LF/HF ratio, etc.)
I'm sorry that this isn't a positive answer, but if you have any questions, please let me know.
This is not my area of expertise and I think you need quite a bit of background to properly measure the LF/HF ratio from the data. I found this which describes many measures: https://pmc.ncbi.nlm.nih.gov/articles/PMC5624990/. I also found a number of references citing issues with the use of this ratio. But, regarding JMP's capabilities, it looks like building any of these measures is straightforward - it looks like they can all be accomplished with relatively simple JMP formulas. Given that the data is time series data, then a number of built in methods could be used, including simple splines or more complex time series models, and the results of these can easily be saved as variables in the data set. So, while the appropriate calculation and its interpretation seem to require a lot of subject matter knowledge, the JMP implementation doesn't see complex to me.
So, my answer to your question would be: yes, JMP can easily be used to measure LF/HF from your heart rate data. But to explain the procedure, you need to provide more background on exactly what measure you want to use. I don't really understand the yusuke_ono response which makes the JMP calculation seem complex, but that might be because I understand even less about LF/HF than they do.
I apologize for the delayed response. Thank you for your reply.
I understand that there is no way to calculate the HF/LF ratio using only mouse operations. I also appreciate you informing me that programming is necessary and that there is a feature called 'add-in'.
It seems a bit challenging with JMP, so I think I will consider other methods as well.
