{
  "id": 10267,
  "title": "Seizure Frequency",
  "url": "/competitions/seizure-prediction/discussion/10267",
  "author_name": "",
  "post_date": "2014-09-08T21:31:14.817Z",
  "votes": null,
  "comment_count": 1,
  "views": 1017,
  "content": "<p>Are the numbers of interictal and preictal data segments indicative of frequency of seizures? For example, Dog_1 has 480 interictal data segments and 24 interictal data segments. Does that mean that Dog_1 has leading seizures about 4.75% of the time? I realize that this is somewhat ambiguous, as it depends on the time units.</p>\n<p>Can seizures be treated as rare occurrences?</p>",
  "messages": [
    {
      "id": "53366",
      "postDate": "09/08/2014 21:31:14",
      "content": "<p>Are the numbers of interictal and preictal data segments indicative of frequency of seizures? For example, Dog_1 has 480 interictal data segments and 24 interictal data segments. Does that mean that Dog_1 has leading seizures about 4.75% of the time? I realize that this is somewhat ambiguous, as it depends on the time units.</p>\n<p>Can seizures be treated as rare occurrences?</p>",
      "rawMarkdown": "",
      "votes": null
    },
    {
      "id": "53450",
      "postDate": "09/10/2014 16:30:46",
      "content": "<p>Seizures are indeed rare events. Seizure frequency varies greatly across patients, but most patients spend a very small percentage of the time actually having seizures. If anything the data in the current contest underestimates the disparity between preictal and interictal data. The data size constraints and the need for as many preictal training segments as possible dictated this, but it is an important feature of the problem.</p>",
      "rawMarkdown": "",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 53450,
      "author_name": "bbrinkm",
      "author_url": "",
      "post_date": "09/10/2014 16:30:46",
      "content": "<p>Seizures are indeed rare events. Seizure frequency varies greatly across patients, but most patients spend a very small percentage of the time actually having seizures. If anything the data in the current contest underestimates the disparity between preictal and interictal data. The data size constraints and the need for as many preictal training segments as possible dictated this, but it is an important feature of the problem.</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "53366": "",
    "53450": ""
  },
  "source": "meta"
}