{
  "id": 185824,
  "title": "Has anyone convincingly cracked the tabular data barrier?",
  "url": "/competitions/osic-pulmonary-fibrosis-progression/discussion/185824",
  "author_name": "Alexander Soare",
  "post_date": "2020-09-22T08:41:29.578000",
  "votes": 1,
  "comment_count": 0,
  "views": 0,
  "content": "<p>By that I mean, have you made use of the DICOMs to get a CV evaluation which is better than just using the tabular data with statistical significance?</p>\n<p>My comparison method has been to zero the DICOM data in whatever training scheme I've been using and compare the results (rather than removing the DICOM data input nodes altogether - which  wouldn't be a fair comparison).</p>\n<p>I've tried every which way, and been fooled to think I'd made a breakthrough several times, but in the end it's always been a false alarm.</p>",
  "messages": [
    {
      "id": 1021969,
      "postDate": "2020-09-22T08:41:29.580Z",
      "content": "<p>By that I mean, have you made use of the DICOMs to get a CV evaluation which is better than just using the tabular data with statistical significance?</p>\n<p>My comparison method has been to zero the DICOM data in whatever training scheme I've been using and compare the results (rather than removing the DICOM data input nodes altogether - which  wouldn't be a fair comparison).</p>\n<p>I've tried every which way, and been fooled to think I'd made a breakthrough several times, but in the end it's always been a false alarm.</p>",
      "rawMarkdown": "By that I mean, have you made use of the DICOMs to get a CV evaluation which is better than just using the tabular data with statistical significance?\n\nMy comparison method has been to zero the DICOM data in whatever training scheme I've been using and compare the results (rather than removing the DICOM data input nodes altogether - which  wouldn't be a fair comparison).\n\nI've tried every which way, and been fooled to think I'd made a breakthrough several times, but in the end it's always been a false alarm.",
      "votes": 1
    }
  ],
  "comments": [],
  "raw_markdown_by_id": {
    "1021969": "By that I mean, have you made use of the DICOMs to get a CV evaluation which is better than just using the tabular data with statistical significance?\n\nMy comparison method has been to zero the DICOM data in whatever training scheme I've been using and compare the results (rather than removing the DICOM data input nodes altogether - which  wouldn't be a fair comparison).\n\nI've tried every which way, and been fooled to think I'd made a breakthrough several times, but in the end it's always been a false alarm."
  }
}