{
  "id": 169995,
  "title": "Did any one use CT scans?",
  "url": "/competitions/osic-pulmonary-fibrosis-progression/discussion/169995",
  "author_name": "",
  "post_date": "2020-07-26T04:30:13.779112Z",
  "votes": 5,
  "comment_count": 5,
  "views": 0,
  "content": "<p>The leaderboard score is pretty close to the current score using only tabular data(age, weeks, smoking history, etc.). So I feel like people is just finetuning on using only tabular data. </p>\n\n<p>I'm wondering if any one is using CT scans? How did it score?</p>\n\n<p>Thanks in advance.</p>",
  "messages": [
    {
      "id": "945665",
      "postDate": "07/26/2020 04:30:13",
      "content": "<p>The leaderboard score is pretty close to the current score using only tabular data(age, weeks, smoking history, etc.). So I feel like people is just finetuning on using only tabular data. </p>\n\n<p>I'm wondering if any one is using CT scans? How did it score?</p>\n\n<p>Thanks in advance.</p>",
      "rawMarkdown": "The leaderboard score is pretty close to the current score using only tabular data(age, weeks, smoking history, etc.). So I feel like people is just finetuning on using only tabular data. \n\nI'm wondering if any one is using CT scans? How did it score?\n\nThanks in advance.",
      "votes": null
    },
    {
      "id": "946656",
      "postDate": "07/26/2020 17:57:13",
      "content": "<p>It worked poorly for me, but I didn't put much work into it.</p>",
      "rawMarkdown": "It worked poorly for me, but I didn't put much work into it.",
      "votes": null
    },
    {
      "id": "947030",
      "postDate": "07/27/2020 03:01:27",
      "content": "<p>I tried to use  original CT scans, my best model got 6.90x lb score.   </p>",
      "rawMarkdown": "I tried to use  original CT scans, my best model got 6.90x lb score.",
      "votes": null
    },
    {
      "id": "947038",
      "postDate": "07/27/2020 03:18:10",
      "content": "<p>Are you using both augmentations and pixel spacing adjustments? I am curious to what architecture too, I mostly boost quantiles and ensemble the resulting models. I tried image and csv together, but results were around 7. </p>",
      "rawMarkdown": "Are you using both augmentations and pixel spacing adjustments? I am curious to what architecture too, I mostly boost quantiles and ensemble the resulting models. I tried image and csv together, but results were around 7.",
      "votes": null
    },
    {
      "id": "947375",
      "postDate": "07/27/2020 08:18:13",
      "content": "<p>No augmentations and pixel spacing adjustments,just original pictures.  </p>",
      "rawMarkdown": "No augmentations and pixel spacing adjustments,just original pictures.",
      "votes": null
    },
    {
      "id": "947504",
      "postDate": "07/27/2020 10:03:30",
      "content": "<p>From previous related competitions it seems like quite a bit of work is needed on the preprocessing of the images.</p>",
      "rawMarkdown": "From previous related competitions it seems like quite a bit of work is needed on the preprocessing of the images.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 946656,
      "author_name": "eladwar",
      "author_url": "",
      "post_date": "07/26/2020 17:57:13",
      "content": "<p>It worked poorly for me, but I didn't put much work into it.</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 947030,
      "author_name": "qinhui1999",
      "author_url": "",
      "post_date": "07/27/2020 03:01:27",
      "content": "<p>I tried to use  original CT scans, my best model got 6.90x lb score.   </p>",
      "votes": null,
      "replies": [
        {
          "id": 947038,
          "author_name": "eladwar",
          "author_url": "",
          "post_date": "07/27/2020 03:18:10",
          "content": "<p>Are you using both augmentations and pixel spacing adjustments? I am curious to what architecture too, I mostly boost quantiles and ensemble the resulting models. I tried image and csv together, but results were around 7. </p>",
          "votes": null,
          "replies": [
            {
              "id": 947375,
              "author_name": "qinhui1999",
              "author_url": "",
              "post_date": "07/27/2020 08:18:13",
              "content": "<p>No augmentations and pixel spacing adjustments,just original pictures.  </p>",
              "votes": null,
              "replies": []
            }
          ]
        }
      ]
    },
    {
      "id": 947504,
      "author_name": "jameschapman19",
      "author_url": "",
      "post_date": "07/27/2020 10:03:30",
      "content": "<p>From previous related competitions it seems like quite a bit of work is needed on the preprocessing of the images.</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "945665": "The leaderboard score is pretty close to the current score using only tabular data(age, weeks, smoking history, etc.). So I feel like people is just finetuning on using only tabular data. \n\nI'm wondering if any one is using CT scans? How did it score?\n\nThanks in advance.",
    "946656": "It worked poorly for me, but I didn't put much work into it.",
    "947030": "I tried to use  original CT scans, my best model got 6.90x lb score.",
    "947038": "Are you using both augmentations and pixel spacing adjustments? I am curious to what architecture too, I mostly boost quantiles and ensemble the resulting models. I tried image and csv together, but results were around 7.",
    "947375": "No augmentations and pixel spacing adjustments,just original pictures.",
    "947504": "From previous related competitions it seems like quite a bit of work is needed on the preprocessing of the images."
  },
  "source": "meta"
}