I'm using kubios to analyze r-r intervals and it looks fine to me. However just few papers in the literature used kubios for Hrv.
I have used Kubios to analyze HRV and found that it is both useful and reliable. We published a recent paper where Kubios analysis was applied for estimation of statistical and spectral HRV indexes[González H et al, Medical Engineering & Physics 2013 (35): 178-187]
Thanks all. Has anyone compared Kubios with another software by any chance?
Yes, Kubios is a good software, but I use Nevrokard to compare the results with Kubios. The results of comparison were very similar. You can use Kubios with no problem.
Kubios is fine, we have used it for several projects.
However, the flexibility of other software packages is preferred.
There are (at least) four MatLab packages for detection/cleaning/processing RR intervals:
ARTiiFact - http://www.artiifact.de/
KARDIA - http://sourceforge.net/projects/mykardia/
HRVAS - http://sourceforge.net/projects/hrvas/
ECGLab - http://pgea.unb.br/~joaoluiz/
Not only are they all free, but you can do batch processing, and modify any necessary parameters of the input/output if you can get into the code.
Kubios is a good software in my opinion but I would also invite you to explore other recent approaches. Please, may you visit the site www.saistmp.com for exploring also a new method in HRV analysis? The download software is free of charge
I agree with all others, also used this software for publication e.g.: Weippert, M., Behrens, K., Rieger, A., Stoll, R., & Kreuzfeld, S. (2013). Heart rate variability and blood pressure during dynamic and static exercise at similar heart rate levels PLoS One, 8(12), e83690. doi: 10.1371/journal.pone.0083690
A brief suggestion. We are accustomed to investigate HRV by the methods also well indicated by the well known Task Force. Usually we use Poincaré Plot , we use linear indexes in time and FFT with estimation in the three bands , VLF,LF,HF (short time recording). R-R is a signal that is intrinsecally non periodic, non stationary and , the most important thing, NON LINEAR so that the use of the non linear methodologies would be also of importance. Poicaré plot is non linear in its representation but SD1 and SD2 relate still linear measures. FFT is linked to a linear transformation, time linear indexes relate assumption of linearity. Kubios estimate and perform a very good analysis by using such methods. But not only. Kubios also estimates non linear indexes and also consider the detrended fluctuation analysis that is of basic importance, ApEn, SamEn.. and still mnore A possible limit? Kubios also performs Recurrence Quantification Analysis and estimation of correlation dimension. We are entering in this case in a new methodlogy ( RQA) that does not induce any trasformation or particular elaboration of the data. According to the basic foundations of non linear methodlogies RQA first of all realizes a reconstruction of the given R-R time series in phase space and than calculates Recurrences , Determinism , Laminarity , entropy , Trapping time and still more . RQA is a method that , following the first indication of Eckmann, was quantified for physiological measurements by Zbilut , dead in 2009, and Webber and gives excellent results when the aim of the investigation is to inspect the inner structure of the given R-R signal. Webber and Zbilut also released a free software on RQA and the results of Kubios should be always at least compared with those obtained by the Zbilut and Webber software . Thre is also another software called Visula Recurrence Analysis that may be used just to compare the results obtained by such different methods.
Finally there is the problem for which I insist to visit our site in order to inspect the method that we propose. Zbilut and I worked so much time in collaboration and , in order to avoid the limitations deriving from the FFT, we published a new method called CZF method that presently one may also find on the Nevrokard site. Nevrokard is another excellent software but it does not use RQA. The aim of the CZF was to study the VARIABILITY of the R-R signal obtaining always the three bands that are of our interest in the short term ( 5-6 minutes) recording but overcoming this time the basic limits of the FFT. The result is that the CZF method estimates Variability in the three bands VLF,LF ,HF and enables also to compare the results about variability by this method with those given in the standard FFT. I repeat : CZF does not suffer of the limitations of the FFT: Where is the problem! It is that CZF estimates variability extending the analysis to the whole set of given R-R data. This is to say that estimates variability betwenn two adjacent RR values , between two spaced intervals , by three spaced and so on. The arising graph may result rather complex to be interpreted by a clinician . Consequently we have decided to give a new version of this method where from one hand we may obtain estimation of variability in the three bands VLF,LF,HF and at the same time we have comparison with standard FFT. The arising interpretation obviously no more gives the previously mentioned difficulties. This new method also includes the possibility of analysis relating some recent advances in HRV analysis as the generalized fractal dimension, the asymetry and an estimation of the BRS. This is the reason because I invite all you to visit www.saistmp.com , to download the softwares that every day we are updating to complete the method and to use it to investigate its new potentialities. Before publishing the software we have experienced the method for a long time and we obtain satisfactory results.
Two final observations. I mentioned the RQA analysis of Zbilut and Webber. This method holds about the notion of recurrence in phase space for the gievn R-R time series. The concept of Variability that we recall in our new method as well as in the previous CZF method no other is that the counterpart of the notion of recurrence.
The final observation. In the estimation of the Variability by the new method we use still the FFT. This is because clinicians use FFT from years in their HRV analysis and thus are accustomed to use this technique. And we continue to use this method but within a brief time we will load also a software where the new method is applied overcoming the standard FFT and continuining in any case to have estimation in the three bands of interest, VLF,LF,HF.
Kubios is fine (especially automation of analysis across several time windows), and results are comparable to other software packages for HRV analysis like Nevrocard. However, IBI corrections delivers sometimes strange results. I just want to test
ARTIIFACT now. Does anyone have already experiences with this new tool?
p.s. Dear Elio, you are right, Kubios implemented RCA as well, however parameters are not to change and resulting recurrence has been always above 99% with my HRV data (measured in goats). Any idea why this happens?
As I said.. I appreciate Kubios that I do not use but it is a good software . In particular in the past I had some contacts with the people that realized Kubios and I attest that they are very serious and have great competence . They have all my consideration. The kubios software. Certainly it moves in the direction of the Tsk Force indications. It calculates PSD with a given window and obviously the results that it gives are very often similar to those of the Nevrokard that of course contains also the first version of my CZF method. You certainly may understand that the limit is not in the kubios software but in the fact that we use the Fourier. This is a linear transformation and it is erroneously applied to a R-R signal that instead it is intrinsecally non linear, non stationary and non periodic. It is true that we resample at 2 or 4 Hz( the situation does not change so much) but obviously resampling we lose basic information. In front of such limitations Fourier coninues every day ti be applied and you find VLF,LF and HF estimations in every published journal also of international value. The crucial limit is that it is also applied in analysis of pathologies where the situation becomes still more complex. Still. The problem of the window . If you change the window kind obviously obtain differnt values. I have a work in my RG list where it is shown how , obviously, all the results are window dependnet. This limit could be overcome by decidng all to use the same window but often this does not happen. Poinacre Plot. This is not a great advantage to have a software calculating Poicare Plot. It is certainly a great step and it gives a good representation in a non linear manner. Sd1 and SD2 returning in kubios an in all other softwares is not so difficult .... SD1 and SD2 have standard formula to be estiamted . We cannot have difefrent results in different softwares. The limit of the Poincaré Plot is that you force an R-R time series in a 2D representation while instead a correct representation in phase space requires so frequently more and more dimensions. The linear indexes! Jan ... they are so simple to be estimated that you never can find difefrences by diffeernt softwares. The problem again is different. What is their actual adavtage. Really do you think to examine and to perform an accurate HRV analysis evaluating SDNN? I have all my more profound reservations. R-R is one of the most complex signals in nature Should SDNN solve the problem? Unfortunately HRV studies finsih to have little consideration , in particular from cardiologists, because the used method induce so strong resrevations. SDNN ... really? The RQA analysis. It was realized by Zbilut and Webber. This is the first example of serious analysis. I have worked several years with Zbilut that unfortunately died in 2009 and I have published with him a lot of papers and both we conceived the first version of the CZF method. It is rather difficult and in particular it does not give values in the three required VLF,LF,HF bands. It requires a detailed reconstruction of the R-R siganl in phase space , an accurate evaluation of the selected distance and still still more. Recurrences should not exceed usually 10-12 % . Knubios finds more elevated values. One shoudl be care since increasing recurernce can produce as consequence to add noise and this may give unreasonable results. Following the line of the initial CZF method I have preapred a new version that as you possibly now is given as theoertcila method and software in www.saistmp.co. I am convinecd , following Zbiluit that may be considered a founding father of linear and non linear methodologies in physiology, that in order to start to ahve methods and softwares indicating a new perspective we have to restart not from the R-R signal but from its VARIABILITY. It is the VARIABILITY the key in HRV and or R-R studies. So I have realized the software and the method totally based on the concept of variability. It travels with the RQA since the recurernce concept in RQA is just the logic counterpart of the cocept of variability. This was my initail idea and of Zbilut when we formulated the first versionof the CZF and this is the conceptual framework of the new version in www.saistmp.com. Analysis of variablity gives again results in VLF ,LF and HF bands and thus one has the advantage also to compare the results with the standard Fourier. Obviously my curernt version of the new CZF method uses still the Fourier but does not use windows. This is a limit? See. The problem is that we are in presence of a non linear , possiblyu chaotic , signal... as Zbilut always learmned to me..... on a series of a non linear process we never should transform ten dtata and instead you see that we go on manipulations by manipulations... corerction of artefacts , linear transfiormation, resampling.... how do we expect to obatin valuable results? I must say that we have applied and still continue to apply the new CZF methdo in difefernt cases of normal and pathological results and the results are very encouraging. Certainly there is still the limit that we continue to use the Fourier .... but we are preparing , as in the standard CZF old version, also the new version in which the Fourier is not used and we obtain at the same time the values in the three bands VLF,LF,HF as required for ouyr studies. In thsi first version we are still using the Fouirier since some clinicians expect to use Fourier, expect to see the Fourier, do not see other than the Fourier .. just where instead , as Zbilut, had and have continue reservations when we ask to ahve detailed quantitative results. Inbstead it runs to have phenomenological indications.
Sorry Jan , I do not remember well but may be I was not complete in my previous exposition. When using Poicare Plot, I apologize if I do not rememebr corerctly, it seems to me that it represnets the distribution of the points , the fitted elipse and gives the values of SD1 and SD2 . Nevroakrd instead calculates alos the area of the elipse elipse, the centroid and add some statistical indexs. Area is important. But I apologize if I bad remember!
Dear Martino,
I can recommend Kubios Software for HRV analysis due to one reason: it has an outstanding R-peak detection algorithm that leaves other R-peak detection implementations far behind! It also offers the possibility to define several segments for analysis, so you can split your raw data automatically into several epochs that will be analysed all on their own - everything is put together in the ouput file.
The main draw-back of Kubios software is the lack of batch processing. Although one always should take a look on the raw data and the found R-peaks, also in case of batch processing, the possibility to ad least load several files into one workspace is missing, especially when you are planning to analyse large cohorts.
I agree only partially. Any HRV software analysis must have a satisfactory R-peak identification system. It is satisfactory also Nevrokard as well as more sophisticated systems as Biopac . Define segments is also a possibility of nervrokard as well of other systems. I do not find so much difference . Of course I repeat ... such softwares , although satisfactory, are all arranged about standard indexes . I think that we must explore variability intended as counterpart of recurrence in R-R intervals. This is the reason because we are proposing a new HRV analysis approach as you may see visiting our site www.saistmp.com. Finally one must account that for standard analysis, with the only exception of RQA, kubios is satisfactory and , this is also of importance, it is free of charge . On the contrary nevrokard needs to be purchased but obviously some basic differences exist.
sorry .. may be that I am not well informed and a new version of kubios is in use. As well as I know kubios requires as input file the R-R time series where the peaks have been just identified by another system. Sorry if it exists also a new version of kubios with identificationof the pekas ... the standard version that I knowledged , I repeat, always required an input R-R time series file. I apologize
Dear Elio Conte,
I very much appreciate your "call for new measures". However, the existing and validated parameters that can be assessed by Kubios and other free HRV software tools provide a solid ground for analysis of the autonomous nervous function.
Concerning the "new" version of Kubios: indeed there is an update available that is capable of r-peak identification. Also the output can now be selected to be a Matlab file or an text file besides the standard pdf outprint.
Sorry Sebastian , I cannot agree . There are not actually actually existing and validated measures in HRV analysis not in kubios. They are all non lineare indexes while it is well established that R-R is a non linear, intrinsecally non peridoci and non stationary signal One of the most complex signals existing in nature. FFT as measure in frequency domanin gives only qualitative indications and it is affected from alot of limitations. In addition it cannot be applied in pathology. Are you able to use it in presnece of pathologiacl caes ? are you able to indicate me what should be the most appropriate window to be seleccetd ? The Hamming ? some other? The time libear indexes .. where do you find a coherent picture ibn the situation ? Do you knoe a range of values for normal subjects? Often you find set of values whose standard deviation is three also four times greatre than the mean values. Is this the manner to hope to do HRV analysis by validated parameters.? Of course , infront of such difficulties we do not ask to dismiss suddently all the standard procedure. If you examine in detail the pdf of our publication we only sugegst that the current methodlogies are not sufficient and the the CONCEPT of variability must be added , quantified and analyzed. I cannot understand the reason to have profound resrevatio n on this sugegstion. Take a normal siubject and a pathologicla subject. Download our software , Do the analysis of the R-R siganls one time by using what you call validated parameters and the other time use our software and do the analysis of variability. Note .. the concept of variability is not a strange idea ... IT IS THE BASIC AND CENTRAL POINT of the physiology and of the cardiology if one intends to examine HEART RATE VARIABILITY. Use all the results obtained : I am sure since we are experiencing this method from long time, that you will find important innovations and you will thank me. Finally a little of open positions no? One cannot always says .. I feel myself very well in this house I do not see reasons to improve... it at the limit may be correct for an house not for science. Attempt first and after you will have new elements to evaluate.
Thank you all for the kind words about Kubios. We are working hard to make Kubios even better, including also the pre-processing steps like R-wave detection and artifact correction. Just few days ago, we released the a new version of Kubios (ver. 2.2) which is now available also for Mac OSX and includes support for Garmin FIT files. Of course there are many things that could be improved and sometimes I hope that I would have more time to give for Kubios.
I compared Kubios with my Matlab functions with satisfactory agreement. Excpet for DFA. But i think it demands large amounts of data and i didn't explored it so much. If researchers know what are tey doing and what do they whant, usually some free softwares are acceptable, mainly because the most important thing is the phsyiological data and information, the experimental design and what you can really obtain from spectral analysis, that requires many assumptions.
sorry I am not sure that I am understanding you correctly ... what do you mean exactly for plety!!
Plethysmographic.
Lihua, you should start a separate question with a description of your data, the platform you collected it on, a better description of what you've tried, etc.
Ok. I had only a doubt .... no more . First of all I do not see how Kubios may enter in the question you posed ( this was the reason of my dount) ...... Kubios accepts only R-R intervals as input txt file if some thing has not changed in some furthetr version and I do not knnow.
In principle : R-R intervals should be determined always by ECG and with a valuable software for peak identification. We cannot confuse ECG signal with recording peripheral circulation . Finger pletysmography should not be used for ANS analysis . I am certainly aware that more and more device are realized with finger plethysmography and peak identification and subsequent analysis in VLF,LF,HF bands but some physiology convinces us that we cannot obtain accurate results as well as when ECG is used, In any case kubios does not enter in this question if , as I know, it uses previously established R-R intervals. In any case finger plethysmography is very usefull but it cannot give accurate results in ANS , VLF,LF,HF bands , as well as by using starting with an ECG with an appropriate software for peak identification. As examle if one submits a paere for publication evidencing that in the tachogram analysis he used finger plethysmography usually the referee rejects the paper if it contains ANS analysis . R-R signal is by itself a so complex signal ... conseuqnently we must be care in principle to add further limits.
In this case it may be your Pulse wave system gives problems. How to save it , waht is the sampling frequency , may yo send us a recorded pulse wave? We will look at the problem. you may contact me at [email protected]
Yes,newer version of kubios support ECG data (as well as RR intervals) and a has reasonably peak detection algorithm. Unfortunately pletismography data don't generate useful tachogram on kubios. You may use invasive blood pressure traces with Labchart(isn't a free software) but the accuracy is still smaller compared to ECG.
I have some difficulty to retain that it is a problem of Kubios ... I more retain that it is a problem in Pulse Wave ... there are some different features one must look very well in Pulse Wave Recording
Obviously an ECG gives an R peak that is rather constant in amplitude and well discriminated respect to the other peaks .. in plety we have a pulse wave whose amplitude may vary from subject to subject also in a consistent manner and in addition may vary in the same subject from an instant to the other ... as example vasocostriction .. or seriously cardiac problems.
I can clarify that we have developed Kubios HRV software for heart rate variability analysis and the input data should be either beat-to-beat RR intervals or ECG data. The peak detection is designed to detect QRS complexes from ECG and you cannot assume that it would work on pulse wave data, which is a completely different signal. Lihua, you can contact me at [email protected] if you want to discuss more about this and possible solutions for your problem.
I expected what Mika has detailed. I have not taken vision of the last version of Kubios but he is right . If they realized a version also with R-R obtained from ECG , consequently it may be difficult that the same software may also run without problems in R-R obtaining time series since Pulse Wave in fact may have basic features that are different from ECG and possibly affecting the result. Of course if one has to perform an ACCURATE ANS analysis , an ECG must be used. It is important to follow this criterium.
I have used Kubios for analysis of RR intervals, but for greater reliability of my data and analysis, I use Kubios in situations of controlled rest and do visual analysis of the data to make sure that artifacts have not been taken into consideration during the analysis.
Hi - I have just started using Kubios for HRV analysis and I'm wondering if the is a way to automatically separate the data into different samples (conditions) before loading it into Kubios. Currently, I'm using the sliding windows to create the different samples. Is there a specific way to structure the data so that upon loading the different samples can be automatically detected. Currently we have time (s) and ECG (mV) as separate columns (we do something like this in Ledalab for the analysis of EDA). Is it possible to add an additional column singling the different samples? If not, what is the best way to do this without the sliding windows?
Thank you,
Victor
Artifact correction with Kubios is quite handy! Keep the good work Mika!
Feedback: It would be nice if it would be possible to analyse shorter intervals than 30s (e.g 15s) --> Might be useful for R-R intervals analysis during quick breaks of sedentary time.
Cheers!
Thanks for these comments and suggestions. We are working on the next release of Kubios now and we will take these into account as much as possible.
Does anyone know when the new version of Kubios will be available? I wanted to download it on a new PC but the website says it isn't available at the moment.
In contrast to the 2.0 version the 2.2 version doesn't open after download and installing (+ Matlab Compiler Runtime 7.17 32-bit).
Does anyone have experience with this kind of problem?
The system is a Windows 8.1 64-bit with MatlabR2016a running just fine.
There might be couple of reasons behind this. Do you receive any error message? Please send me an email (kubios-software (at)uef.fi)
kubios 2.2 seems to be there but when you go to its link it says something like to be available soon.
I had used Kubios in my studies and I like it. I am ansious to use new version.
I checked kubios.com there are both standard (free) and premium (for money) version but stating "To be released soon". Also, it looks like it is primarily for ECG signals. Does anyone know if it can be used with BCG signal?
I have been using Kubios for quite a while. Indeed, it is my favourite.
I particularly like recurrent analysis variables, an aspect poorly highlighted in HRV analysis literature.
I have recommended using Kubios to analyze the HRV and I used in some studies that I have conducted (https://www.researchgate.net/publication/261067212_Effects_of_Additional_Repeated_Sprint_Training_During_Preseason_on_Performance_Heart_Rate_Variability_and_Stress_Symptoms_in_Futsal_Players). But there is another good software, such as ECGlab (http://www.ene.unb.br/joaoluiz/pdf/icsp2002_ecglab.pdf).
Article Effects of Additional Repeated Sprint Training During Presea...
Hello ! What is a BCG Signal ?
I am using the Faros Senor with an ECG cable and the premium version
works fine for me. The analysis is not so easy.
best regards
- Badges
- Science topic
- Similar topics
- Phytochemistry
- Extracts