hrv_time of an epoch with 0 or 1 r_peaks and issue with sanitizing data.

[LaskawiecPiotr]

Hi,

I am not sure if this is a bug or actually intended. For some reason the first 70 epochs of my data have 0 r_peaks. This leads to an empty rri if I am not mistaken. Therefore when running hrv_time on one of the epochs it returns the error:

IndexError                                Traceback (most recent call last)
Cell In[102], line 1
----> 1 nk.hrv_time(epochs["70"])

File ~\anaconda3\Lib\site-packages\neurokit2\hrv\hrv_time.py:164, in hrv_time(peaks, sampling_rate, show, **kwargs)
    162 out["SDNN"] = np.nanstd(rri, ddof=1)
    163 for i in [1, 2, 5]:
--> 164     out["SDANN" + str(i)] = _sdann(rri, window=i)
    165     out["SDNNI" + str(i)] = _sdnni(rri, window=i)
    167 # Difference-based

File ~\anaconda3\Lib\site-packages\neurokit2\hrv\hrv_time.py:228, in _sdann(rri, rri_time, window)
    226     rri_time = np.nancumsum(rri / 1000)
    227 # Convert timestamps to milliseconds and ensure first timestamp is equal to first interval
--> 228 rri_cumsum = (rri_time - rri_time[0]) * 1000 + rri[0]
    229 n_windows = int(np.round(rri_cumsum[-1] / window_size))
    230 if n_windows < 3:

IndexError: index 0 is out of bounds for axis 0 with size 0

It would be nice if the code just returned NaN or issued a warning that this is what happened.

Also I think there is an issue with sanitizing data. I took my signal, I cleaned it using ecg_clean and then got the peaks=ecg_peaks. Therefore peaks is now in a tuple of df and a dictionary. According to the documentation ths is a valid input for hrv but I get an error. hrv(peaks[0]) does work. After following the code I think i realized what is the problem ( I might be wrong I am very new to this). peaks is a tuple so _hrv_santizie_tuple is called to sanitize them.

def _hrv_sanitize_tuple(peaks, sampling_rate=1000):

    # Get sampling rate
    info = [i for i in peaks if isinstance(i, dict)]
    sampling_rate = info[0]["sampling_rate"]

    # Detect actual sampling rate
    if len(info) < 1:
        peaks, sampling_rate = peaks[0], peaks[1]

    # Get peaks
    if isinstance(peaks[0], (dict, pd.DataFrame)):
        try:
            peaks = _hrv_sanitize_dict_or_df(peaks[0])
        except NameError:
            if isinstance(peaks[1], (dict, pd.DataFrame)):
                try:
                    peaks = _hrv_sanitize_dict_or_df(peaks[1])
                except NameError:
                    peaks = _hrv_sanitize_peaks(peaks[1])
            else:
                peaks = _hrv_sanitize_peaks(peaks[0])
    rri, rri_time, rri_missing = _hrv_get_rri(peaks=peaks, sampling_rate=sampling_rate)

    return rri, rri_time, rri_missing, sampling_rate

Therefore here, the input peaks will go into peaks = _hrv_sanitize_dict_or_df(peaks[0])

But hrv_sanitize_dict_df outputs rri and not peaks which leads to an error. Therefore shouldn't it be

def _hrv_sanitize_tuple(peaks, sampling_rate=1000):

    # Get sampling rate
    info = [i for i in peaks if isinstance(i, dict)]
    sampling_rate = info[0]["sampling_rate"]

    # Detect actual sampling rate
    if len(info) < 1:
        peaks, sampling_rate = peaks[0], peaks[1]

    # Get peaks
    if isinstance(peaks[0], (dict, pd.DataFrame)):
        try:
            rri, rri_time, rri_missing, sampling_rate = _hrv_sanitize_dict_or_df(peaks[0],sampling_rate=sampling_rate)
            return rri, rri_time, rri_missing, sampling_rate
        except NameError:
            if isinstance(peaks[1], (dict, pd.DataFrame)):
                try:
                    peaks = _hrv_sanitize_dict_or_df(peaks[1])
                except NameError:
                    peaks = _hrv_sanitize_peaks(peaks[1])
            else:
                peaks = _hrv_sanitize_peaks(peaks[0])
    rri, rri_time, rri_missing = _hrv_get_rri(peaks=peaks, sampling_rate=sampling_rate)

    return rri, rri_time, rri_missing, sampling_rate

I do not know what are the dependencies but this fixed it for me at least.

Update: I apologize I think i know what happened. I guess you want people to do peaks,info= ecg_peaks and then pass the peaks into hrv. I did peaks=ecg_peaks which led this problem. I guess this is a non issue if you make the documentation more precise.

[welcome]

Hi 👋 Thanks for reaching out and opening your first issue here! We'll try to come back to you as soon as possible. ❤️