{
  "id": 651505,
  "title": "frequency range?",
  "url": "/competitions/physionet-ecg-image-digitization/discussion/651505",
  "author_name": "",
  "post_date": "2025-12-03T23:02:18.038304600Z",
  "votes": null,
  "comment_count": 5,
  "views": 0,
  "content": "<p>I found the frequency range in the reference signals of the training set from 250 to 1025. \nIs there a specific frequency range limit for the evaluation stage? Can all questions be answered with a frequency of 250?</p>",
  "messages": [
    {
      "id": "3361872",
      "postDate": "12/03/2025 23:02:18",
      "content": "<p>I found the frequency range in the reference signals of the training set from 250 to 1025. \nIs there a specific frequency range limit for the evaluation stage? Can all questions be answered with a frequency of 250?</p>",
      "rawMarkdown": "I found the frequency range in the reference signals of the training set from 250 to 1025. \nIs there a specific frequency range limit for the evaluation stage? Can all questions be answered with a frequency of 250?",
      "votes": null
    },
    {
      "id": "3363264",
      "postDate": "12/05/2025 20:48:39",
      "content": "<p>No, not really. You'll need to read the sample frequency from the CSV file and resample your predictions accordingly, the only thing fixed in this dataset is the duration of the lead signals</p>",
      "rawMarkdown": "No, not really. You'll need to read the sample frequency from the CSV file and resample your predictions accordingly, the only thing fixed in this dataset is the duration of the lead signals",
      "votes": null
    },
    {
      "id": "3367305",
      "postDate": "12/08/2025 10:39:11",
      "content": "<p>Thank you for your answer, but is it technically possible for us to always and accurately detect frequencies?</p>",
      "rawMarkdown": "Thank you for your answer, but is it technically possible for us to always and accurately detect frequencies?",
      "votes": null
    },
    {
      "id": "3367490",
      "postDate": "12/08/2025 13:52:40",
      "content": "<p>You can't necessarily detect frequencies, but you can sample your predictions at pretty much any frequency that you've been asked to use</p>",
      "rawMarkdown": "You can't necessarily detect frequencies, but you can sample your predictions at pretty much any frequency that you've been asked to use",
      "votes": null
    },
    {
      "id": "3367886",
      "postDate": "12/08/2025 19:45:51",
      "content": "<p>No, the target sampling frequency should always be read from the test CSV file because the effective sampling frequency of the image-based time-series is imposed by the printing and scanning process, which are independent from the original signal's sampling frequency. It's just like converting a digital signal back to an analog signal and then resampling it again at another rate. You need to know the original sampling frequency to be able to get the signal length correct and compare the waveforms sample-by-sample.</p>",
      "rawMarkdown": "No, the target sampling frequency should always be read from the test CSV file because the effective sampling frequency of the image-based time-series is imposed by the printing and scanning process, which are independent from the original signal's sampling frequency. It's just like converting a digital signal back to an analog signal and then resampling it again at another rate. You need to know the original sampling frequency to be able to get the signal length correct and compare the waveforms sample-by-sample.",
      "votes": null
    },
    {
      "id": "3368524",
      "postDate": "12/09/2025 09:45:03",
      "content": "<p>Thank you for your explanation.</p>",
      "rawMarkdown": "Thank you for your explanation.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 3363264,
      "author_name": "henrychibueze",
      "author_url": "",
      "post_date": "12/05/2025 20:48:39",
      "content": "<p>No, not really. You'll need to read the sample frequency from the CSV file and resample your predictions accordingly, the only thing fixed in this dataset is the duration of the lead signals</p>",
      "votes": null,
      "replies": [
        {
          "id": 3367305,
          "author_name": "yasbolaghy",
          "author_url": "",
          "post_date": "12/08/2025 10:39:11",
          "content": "<p>Thank you for your answer, but is it technically possible for us to always and accurately detect frequencies?</p>",
          "votes": null,
          "replies": [
            {
              "id": 3367490,
              "author_name": "henrychibueze",
              "author_url": "",
              "post_date": "12/08/2025 13:52:40",
              "content": "<p>You can't necessarily detect frequencies, but you can sample your predictions at pretty much any frequency that you've been asked to use</p>",
              "votes": null,
              "replies": []
            },
            {
              "id": 3367886,
              "author_name": "r2241272",
              "author_url": "",
              "post_date": "12/08/2025 19:45:51",
              "content": "<p>No, the target sampling frequency should always be read from the test CSV file because the effective sampling frequency of the image-based time-series is imposed by the printing and scanning process, which are independent from the original signal's sampling frequency. It's just like converting a digital signal back to an analog signal and then resampling it again at another rate. You need to know the original sampling frequency to be able to get the signal length correct and compare the waveforms sample-by-sample.</p>",
              "votes": null,
              "replies": [
                {
                  "id": 3368524,
                  "author_name": "yasbolaghy",
                  "author_url": "",
                  "post_date": "12/09/2025 09:45:03",
                  "content": "<p>Thank you for your explanation.</p>",
                  "votes": null,
                  "replies": []
                }
              ]
            }
          ]
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "3361872": "I found the frequency range in the reference signals of the training set from 250 to 1025. \nIs there a specific frequency range limit for the evaluation stage? Can all questions be answered with a frequency of 250?",
    "3363264": "No, not really. You'll need to read the sample frequency from the CSV file and resample your predictions accordingly, the only thing fixed in this dataset is the duration of the lead signals",
    "3367305": "Thank you for your answer, but is it technically possible for us to always and accurately detect frequencies?",
    "3367490": "You can't necessarily detect frequencies, but you can sample your predictions at pretty much any frequency that you've been asked to use",
    "3367886": "No, the target sampling frequency should always be read from the test CSV file because the effective sampling frequency of the image-based time-series is imposed by the printing and scanning process, which are independent from the original signal's sampling frequency. It's just like converting a digital signal back to an analog signal and then resampling it again at another rate. You need to know the original sampling frequency to be able to get the signal length correct and compare the waveforms sample-by-sample.",
    "3368524": "Thank you for your explanation."
  },
  "source": "meta"
}