{
  "id": 346540,
  "title": "My PhD was on injury prevention based on biomechanics",
  "url": "/competitions/big-data-derby-2022/discussion/346540",
  "author_name": "",
  "post_date": "2022-08-20T06:03:23.955417200Z",
  "votes": 15,
  "comment_count": 5,
  "views": 0,
  "content": "<p>Hi everyone,</p>\n<p>During my project, I always suggested someone should start something similar in horse racing, less ethical issues and an abundance of money. And data, lots of data. But before you go in heads first, please read this one:</p>\n<p><a href=\"https://pubmed.ncbi.nlm.nih.gov/27095747/\" target=\"_blank\">https://pubmed.ncbi.nlm.nih.gov/27095747/</a></p>\n<p>For our studies, we did not use ML, but factor and regression analyses in order to identify risk factors.</p>\n<p>Anyway, looking forward to see where this goes!</p>\n<p>Duncan</p>",
  "messages": [
    {
      "id": "1906680",
      "postDate": "08/20/2022 06:03:23",
      "content": "<p>Hi everyone,</p>\n<p>During my project, I always suggested someone should start something similar in horse racing, less ethical issues and an abundance of money. And data, lots of data. But before you go in heads first, please read this one:</p>\n<p><a href=\"https://pubmed.ncbi.nlm.nih.gov/27095747/\" target=\"_blank\">https://pubmed.ncbi.nlm.nih.gov/27095747/</a></p>\n<p>For our studies, we did not use ML, but factor and regression analyses in order to identify risk factors.</p>\n<p>Anyway, looking forward to see where this goes!</p>\n<p>Duncan</p>",
      "rawMarkdown": "Hi everyone,\n\nDuring my project, I always suggested someone should start something similar in horse racing, less ethical issues and an abundance of money. And data, lots of data. But before you go in heads first, please read this one:\n\nhttps://pubmed.ncbi.nlm.nih.gov/27095747/\n\nFor our studies, we did not use ML, but factor and regression analyses in order to identify risk factors.\n\nAnyway, looking forward to see where this goes!\n\nDuncan",
      "votes": null
    },
    {
      "id": "1906702",
      "postDate": "08/20/2022 06:11:55",
      "content": "<p>Wonderful article, great topic! </p>",
      "rawMarkdown": "Wonderful article, great topic!",
      "votes": null
    },
    {
      "id": "1913738",
      "postDate": "08/25/2022 13:31:36",
      "content": "<p>And today’s post (with comments) re: an update. Spoiler: still not possible.</p>\n<p><a href=\"https://twitter.com/RoaldBahr/status/1560960990300225539\" target=\"_blank\">https://twitter.com/RoaldBahr/status/1560960990300225539</a></p>",
      "rawMarkdown": "And today’s post (with comments) re: an update. Spoiler: still not possible.\n\nhttps://twitter.com/RoaldBahr/status/1560960990300225539",
      "votes": null
    },
    {
      "id": "1913899",
      "postDate": "08/25/2022 15:09:03",
      "content": "<p>I come from a sport background as well Duncan. Roald's stuff is great isn't it? I think the problem isn't necessarily that injuries can't be predicted, it's that the methods employed in sport (and possible to employ with the dataset we have here) can't do it. I reckon there is a much greater likelihood that injuries could be predicted with some computer vision/stride data within horses, which isn't much help for this competition, but is definitely a future avenue that the organisers could consider.</p>",
      "rawMarkdown": "I come from a sport background as well Duncan. Roald's stuff is great isn't it? I think the problem isn't necessarily that injuries can't be predicted, it's that the methods employed in sport (and possible to employ with the dataset we have here) can't do it. I reckon there is a much greater likelihood that injuries could be predicted with some computer vision/stride data within horses, which isn't much help for this competition, but is definitely a future avenue that the organisers could consider.",
      "votes": null
    },
    {
      "id": "1914737",
      "postDate": "08/26/2022 11:23:12",
      "content": "<p>Hi Tom, thanks for chiming in!</p>\n<p>Imho the intrinsically biomechanical factor is just one of many of this multifactorial problem. The others being - but not limited to - biochemical, psychological, extrinsically biomechanical, environmental, and just plain bad luck. And of course, each all interact. As Bahr argues, the identification of risk factors that correlate with the unwanted outcome (i.e., injury) is only worth it, if adequate preventive measures exist. On the other hand, you'll need to identify risk factors first to start working on prevention. Chicken - egg.</p>\n<p>In human subjects, previous injury is often reported as the most important risk factor. In horses, it seems (see other thread: <a href=\"https://www.kaggle.com/competitions/big-data-derby-2022/discussion/347128\" target=\"_blank\">https://www.kaggle.com/competitions/big-data-derby-2022/discussion/347128</a>) that once they're injured, it's more 'efficient' to euthanize than to rehabilitate?</p>\n<p>And I'm not entirely sure, but did the subtitle (aim) of this competition change? It seems that the F1 approach (<a href=\"https://www.kaggle.com/competitions/big-data-derby-2022/discussion/347249\" target=\"_blank\">https://www.kaggle.com/competitions/big-data-derby-2022/discussion/347249</a>) perhaps is better suited than a biomechanic injury (n=17!, not exactly big data) approach.</p>",
      "rawMarkdown": "Hi Tom, thanks for chiming in!\n\nImho the intrinsically biomechanical factor is just one of many of this multifactorial problem. The others being - but not limited to - biochemical, psychological, extrinsically biomechanical, environmental, and just plain bad luck. And of course, each all interact. As Bahr argues, the identification of risk factors that correlate with the unwanted outcome (i.e., injury) is only worth it, if adequate preventive measures exist. On the other hand, you'll need to identify risk factors first to start working on prevention. Chicken - egg.\n\nIn human subjects, previous injury is often reported as the most important risk factor. In horses, it seems (see other thread: https://www.kaggle.com/competitions/big-data-derby-2022/discussion/347128) that once they're injured, it's more 'efficient' to euthanize than to rehabilitate?\n\nAnd I'm not entirely sure, but did the subtitle (aim) of this competition change? It seems that the F1 approach (https://www.kaggle.com/competitions/big-data-derby-2022/discussion/347249) perhaps is better suited than a biomechanic injury (n=17!, not exactly big data) approach.",
      "votes": null
    },
    {
      "id": "1915643",
      "postDate": "08/27/2022 06:24:28",
      "content": "<p>Hi mate,</p>\n<p>Firstly, competition rules state people can answer whatever question they want essentially so injury prediction is a possible question to answer, but it's not the route I'm going down for the reasons we both understand.</p>\n<p>The multifactorial nature is so true, my thought was that with the other data I mentioned, it might be easier to predict an inviting event. I don't think it's possible to do that with the data given, but I'd love to be proven wrong because it would help a lot in human sport!</p>\n<p>With regards to horses getting injured, mostly to my knowledge they are euthanised from catastrophic injuries (e.g. bone breaks/tendon injuries). These types don't fall under the \"risk factor associated with future injury\" in humans do they? I thought it was mainly soft tissue injuries that suffered from that and those injuries in horses don't tend to be widely reported (in the UK) so it's difficult to know if the same association holds.</p>\n<p>Essentially, what I'm saying is that I agree that injury prediction is unlikely to work in this dataset but I feel it will be possible in the future and have managed to turn that into two longish comments. No wonder I've spent some time in academia eh?</p>",
      "rawMarkdown": "Hi mate,\n\nFirstly, competition rules state people can answer whatever question they want essentially so injury prediction is a possible question to answer, but it's not the route I'm going down for the reasons we both understand.\n\nThe multifactorial nature is so true, my thought was that with the other data I mentioned, it might be easier to predict an inviting event. I don't think it's possible to do that with the data given, but I'd love to be proven wrong because it would help a lot in human sport!\n\nWith regards to horses getting injured, mostly to my knowledge they are euthanised from catastrophic injuries (e.g. bone breaks/tendon injuries). These types don't fall under the \"risk factor associated with future injury\" in humans do they? I thought it was mainly soft tissue injuries that suffered from that and those injuries in horses don't tend to be widely reported (in the UK) so it's difficult to know if the same association holds.\n\nEssentially, what I'm saying is that I agree that injury prediction is unlikely to work in this dataset but I feel it will be possible in the future and have managed to turn that into two longish comments. No wonder I've spent some time in academia eh?",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 1906702,
      "author_name": "ravi20076",
      "author_url": "",
      "post_date": "08/20/2022 06:11:55",
      "content": "<p>Wonderful article, great topic! </p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 1913738,
      "author_name": "fransz",
      "author_url": "",
      "post_date": "08/25/2022 13:31:36",
      "content": "<p>And today’s post (with comments) re: an update. Spoiler: still not possible.</p>\n<p><a href=\"https://twitter.com/RoaldBahr/status/1560960990300225539\" target=\"_blank\">https://twitter.com/RoaldBahr/status/1560960990300225539</a></p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 1913899,
      "author_name": "tomsawczuk",
      "author_url": "",
      "post_date": "08/25/2022 15:09:03",
      "content": "<p>I come from a sport background as well Duncan. Roald's stuff is great isn't it? I think the problem isn't necessarily that injuries can't be predicted, it's that the methods employed in sport (and possible to employ with the dataset we have here) can't do it. I reckon there is a much greater likelihood that injuries could be predicted with some computer vision/stride data within horses, which isn't much help for this competition, but is definitely a future avenue that the organisers could consider.</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 1914737,
      "author_name": "fransz",
      "author_url": "",
      "post_date": "08/26/2022 11:23:12",
      "content": "<p>Hi Tom, thanks for chiming in!</p>\n<p>Imho the intrinsically biomechanical factor is just one of many of this multifactorial problem. The others being - but not limited to - biochemical, psychological, extrinsically biomechanical, environmental, and just plain bad luck. And of course, each all interact. As Bahr argues, the identification of risk factors that correlate with the unwanted outcome (i.e., injury) is only worth it, if adequate preventive measures exist. On the other hand, you'll need to identify risk factors first to start working on prevention. Chicken - egg.</p>\n<p>In human subjects, previous injury is often reported as the most important risk factor. In horses, it seems (see other thread: <a href=\"https://www.kaggle.com/competitions/big-data-derby-2022/discussion/347128\" target=\"_blank\">https://www.kaggle.com/competitions/big-data-derby-2022/discussion/347128</a>) that once they're injured, it's more 'efficient' to euthanize than to rehabilitate?</p>\n<p>And I'm not entirely sure, but did the subtitle (aim) of this competition change? It seems that the F1 approach (<a href=\"https://www.kaggle.com/competitions/big-data-derby-2022/discussion/347249\" target=\"_blank\">https://www.kaggle.com/competitions/big-data-derby-2022/discussion/347249</a>) perhaps is better suited than a biomechanic injury (n=17!, not exactly big data) approach.</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 1915643,
      "author_name": "tomsawczuk",
      "author_url": "",
      "post_date": "08/27/2022 06:24:28",
      "content": "<p>Hi mate,</p>\n<p>Firstly, competition rules state people can answer whatever question they want essentially so injury prediction is a possible question to answer, but it's not the route I'm going down for the reasons we both understand.</p>\n<p>The multifactorial nature is so true, my thought was that with the other data I mentioned, it might be easier to predict an inviting event. I don't think it's possible to do that with the data given, but I'd love to be proven wrong because it would help a lot in human sport!</p>\n<p>With regards to horses getting injured, mostly to my knowledge they are euthanised from catastrophic injuries (e.g. bone breaks/tendon injuries). These types don't fall under the \"risk factor associated with future injury\" in humans do they? I thought it was mainly soft tissue injuries that suffered from that and those injuries in horses don't tend to be widely reported (in the UK) so it's difficult to know if the same association holds.</p>\n<p>Essentially, what I'm saying is that I agree that injury prediction is unlikely to work in this dataset but I feel it will be possible in the future and have managed to turn that into two longish comments. No wonder I've spent some time in academia eh?</p>",
      "votes": null,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "1906680": "Hi everyone,\n\nDuring my project, I always suggested someone should start something similar in horse racing, less ethical issues and an abundance of money. And data, lots of data. But before you go in heads first, please read this one:\n\nhttps://pubmed.ncbi.nlm.nih.gov/27095747/\n\nFor our studies, we did not use ML, but factor and regression analyses in order to identify risk factors.\n\nAnyway, looking forward to see where this goes!\n\nDuncan",
    "1906702": "Wonderful article, great topic!",
    "1913738": "And today’s post (with comments) re: an update. Spoiler: still not possible.\n\nhttps://twitter.com/RoaldBahr/status/1560960990300225539",
    "1913899": "I come from a sport background as well Duncan. Roald's stuff is great isn't it? I think the problem isn't necessarily that injuries can't be predicted, it's that the methods employed in sport (and possible to employ with the dataset we have here) can't do it. I reckon there is a much greater likelihood that injuries could be predicted with some computer vision/stride data within horses, which isn't much help for this competition, but is definitely a future avenue that the organisers could consider.",
    "1914737": "Hi Tom, thanks for chiming in!\n\nImho the intrinsically biomechanical factor is just one of many of this multifactorial problem. The others being - but not limited to - biochemical, psychological, extrinsically biomechanical, environmental, and just plain bad luck. And of course, each all interact. As Bahr argues, the identification of risk factors that correlate with the unwanted outcome (i.e., injury) is only worth it, if adequate preventive measures exist. On the other hand, you'll need to identify risk factors first to start working on prevention. Chicken - egg.\n\nIn human subjects, previous injury is often reported as the most important risk factor. In horses, it seems (see other thread: https://www.kaggle.com/competitions/big-data-derby-2022/discussion/347128) that once they're injured, it's more 'efficient' to euthanize than to rehabilitate?\n\nAnd I'm not entirely sure, but did the subtitle (aim) of this competition change? It seems that the F1 approach (https://www.kaggle.com/competitions/big-data-derby-2022/discussion/347249) perhaps is better suited than a biomechanic injury (n=17!, not exactly big data) approach.",
    "1915643": "Hi mate,\n\nFirstly, competition rules state people can answer whatever question they want essentially so injury prediction is a possible question to answer, but it's not the route I'm going down for the reasons we both understand.\n\nThe multifactorial nature is so true, my thought was that with the other data I mentioned, it might be easier to predict an inviting event. I don't think it's possible to do that with the data given, but I'd love to be proven wrong because it would help a lot in human sport!\n\nWith regards to horses getting injured, mostly to my knowledge they are euthanised from catastrophic injuries (e.g. bone breaks/tendon injuries). These types don't fall under the \"risk factor associated with future injury\" in humans do they? I thought it was mainly soft tissue injuries that suffered from that and those injuries in horses don't tend to be widely reported (in the UK) so it's difficult to know if the same association holds.\n\nEssentially, what I'm saying is that I agree that injury prediction is unlikely to work in this dataset but I feel it will be possible in the future and have managed to turn that into two longish comments. No wonder I've spent some time in academia eh?"
  },
  "source": "meta"
}