{
  "id": 149863,
  "title": "Predicting Gleason scores",
  "url": "/competitions/prostate-cancer-grade-assessment/discussion/149863",
  "author_name": "Son of Anton v3.0",
  "post_date": "2020-05-10T09:02:20.491000",
  "votes": 2,
  "comment_count": 5,
  "views": 0,
  "content": "<p>Has anyone got appreciable results by predicting Gleason score and getting isup_grade from it? The Gleason score looks more sensitive than isup grade. Is it a good approach in the first place?</p>",
  "messages": [
    {
      "id": 840669,
      "postDate": "2020-05-10T09:02:20.490Z",
      "content": "<p>Has anyone got appreciable results by predicting Gleason score and getting isup_grade from it? The Gleason score looks more sensitive than isup grade. Is it a good approach in the first place?</p>",
      "rawMarkdown": "Has anyone got appreciable results by predicting Gleason score and getting isup_grade from it? The Gleason score looks more sensitive than isup grade. Is it a good approach in the first place?",
      "votes": 2
    },
    {
      "id": 851802,
      "postDate": "2020-05-17T23:50:55.243Z",
      "content": "<p>I tried multiple times, both in classification and regression, as I was (wrongly) convinced it should have been better.\nThe results were always significantly worse. </p>",
      "rawMarkdown": "I tried multiple times, both in classification and regression, as I was (wrongly) convinced it should have been better.\nThe results were always significantly worse. "
    },
    {
      "id": 847389,
      "postDate": "2020-05-14T11:17:51.603Z",
      "content": "<p>I have got 0.67LB with this approach.</p>",
      "rawMarkdown": "I have got 0.67LB with this approach."
    },
    {
      "id": 840897,
      "postDate": "2020-05-10T13:28:12.090Z",
      "content": "<p>I tried at the very beginning of the competition, but very briefly. The results were quite bad, so I decided to drop the idea. Maybe didn't spend enough time to optimize it. I would be surprised if someone manages to make it work, it must be some smart approach I guess</p>",
      "rawMarkdown": "I tried at the very beginning of the competition, but very briefly. The results were quite bad, so I decided to drop the idea. Maybe didn't spend enough time to optimize it. I would be surprised if someone manages to make it work, it must be some smart approach I guess",
      "replies": [
        {
          "id": 840984,
          "postDate": "2020-05-10T14:29:02.500Z",
          "content": "<p>Same here. I got around 0.04 LB. So I decided to bin it.</p>",
          "rawMarkdown": "Same here. I got around 0.04 LB. So I decided to bin it.\n"
        },
        {
          "id": 841104,
          "postDate": "2020-05-10T15:46:24.670Z",
          "content": "<p>me too. I trained my model to predict the Gleason scores and then compute isup_grade. \nnot working. I guess this is because the model doesn't understand the differences between these two Gleason scores and sometimes predict a wrong combination, for example, 0+3.</p>",
          "rawMarkdown": "me too. I trained my model to predict the Gleason scores and then compute isup_grade. \nnot working. I guess this is because the model doesn't understand the differences between these two Gleason scores and sometimes predict a wrong combination, for example, 0+3."
        }
      ]
    }
  ],
  "comments": [
    {
      "id": 851802,
      "author_name": "Peter Cnudde",
      "author_url": "",
      "post_date": "2020-05-17T23:50:55.243000",
      "content": "<p>I tried multiple times, both in classification and regression, as I was (wrongly) convinced it should have been better.\nThe results were always significantly worse. </p>",
      "votes": 0,
      "replies": []
    },
    {
      "id": 847389,
      "author_name": "Nishant Singh",
      "author_url": "",
      "post_date": "2020-05-14T11:17:51.603000",
      "content": "<p>I have got 0.67LB with this approach.</p>",
      "votes": 0,
      "replies": []
    },
    {
      "id": 840897,
      "author_name": "abzaliev",
      "author_url": "",
      "post_date": "2020-05-10T13:28:12.090000",
      "content": "<p>I tried at the very beginning of the competition, but very briefly. The results were quite bad, so I decided to drop the idea. Maybe didn't spend enough time to optimize it. I would be surprised if someone manages to make it work, it must be some smart approach I guess</p>",
      "votes": 0,
      "replies": [
        {
          "id": 840984,
          "author_name": "Son of Anton v3.0",
          "author_url": "",
          "post_date": "2020-05-10T14:29:02.500000",
          "content": "<p>Same here. I got around 0.04 LB. So I decided to bin it.</p>",
          "votes": 0,
          "replies": []
        },
        {
          "id": 841104,
          "author_name": "Zhijian Li",
          "author_url": "",
          "post_date": "2020-05-10T15:46:24.670000",
          "content": "<p>me too. I trained my model to predict the Gleason scores and then compute isup_grade. \nnot working. I guess this is because the model doesn't understand the differences between these two Gleason scores and sometimes predict a wrong combination, for example, 0+3.</p>",
          "votes": 0,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "840669": "Has anyone got appreciable results by predicting Gleason score and getting isup_grade from it? The Gleason score looks more sensitive than isup grade. Is it a good approach in the first place?",
    "851802": "I tried multiple times, both in classification and regression, as I was (wrongly) convinced it should have been better.\nThe results were always significantly worse. ",
    "847389": "I have got 0.67LB with this approach.",
    "840897": "I tried at the very beginning of the competition, but very briefly. The results were quite bad, so I decided to drop the idea. Maybe didn't spend enough time to optimize it. I would be surprised if someone manages to make it work, it must be some smart approach I guess"
  }
}