Problem to recreate the result

[agrans]

Hi, I have some problems recreating the result from the article when following the instruction.

• "download_minic_iii_records.py" crashes when it tries to download the files in "MIMIC-III_ppg_dataset_records.txt" with the error: Traceback (most recent call last): File "download_mimic_iii_records.py", line 111, in download_mimic_iii_records(RecordsFile, OutputPath) File "download_mimic_iii_records.py", line 73, in download_mimic_iii_records abp_FidPoints = hp.process(abp, fs) File "/home/agrans/Documents/non-invasive-bp-estimation-using-deep-learning/venv/lib/python3.8/site-packages/heartpy/heartpy.py", line 289, in process working_data = fit_peaks(hrdata, rol_mean, sample_rate, bpmmin=bpmmin, File "/home/agrans/Documents/non-invasive-bp-estimation-using-deep-learning/venv/lib/python3.8/site-packages/heartpy/peakdetection.py", line 305, in fit_peaks raise BadSignalWarning('\n----------------\nCould not determine best fit for \ heartpy.exceptions.BadSignalWarning:

• "download_minic_iii_records.py" seams to not create ".mat"-types which seams to be required for "prepare_MIMIC_dataset.py" and all data is skipped.

What am I doing wrong when following the instructions in the README?

[Fabian-Sc85]

Hi, could you give me the name of the particular record at which the script fails so that I can reproduce this error? Thanks.

[agrans]

The first four it crashes on are: 39/3982843 38/3889764_0025 38/3879205_0027 31/3168979

[Fabian-Sc85]

It seems the problem arises if the heartpy package is unable to extract systolic and diastolic peaks from the ABP/PPG signals. I have "solved" the problem by skipping such records. However, I am going to publish the MIMIC-B dataset used in the paper as soon as possible

[Fabian-Sc85]

The dataset created by the script "download_minic_iii_records.py" is now available for download via Zenodo