{"id":61752,"date":"2022-03-28T10:30:40","date_gmt":"2022-03-28T17:30:40","guid":{"rendered":"https:\/\/cm-edgetun.pages.dev\/en-us\/industry\/microsoft-in-business\/?p=61752"},"modified":"2023-06-16T08:46:07","modified_gmt":"2023-06-16T15:46:07","slug":"how-cloud-based-math-modeling-can-reduce-organ-transplant-risk","status":"publish","type":"post","link":"https:\/\/cm-edgetun.pages.dev\/en-us\/industry\/microsoft-in-business\/business-transformation\/2022\/03\/28\/how-cloud-based-math-modeling-can-reduce-organ-transplant-risk\/","title":{"rendered":"How cloud-based math modeling can reduce organ transplant risk\u00a0"},"content":{"rendered":"\n

\"MedicalSome of the most interesting scientific discoveries happen at the intersection between disciplines\u2014and sometimes those discoveries can even save lives. That\u2019s what two researchers at the University of North Carolina learned with organ transplantation, when they set out to find a better way to match donors with recipients. <\/p>\n\n\n\n

The organ transplant process is time-sensitive and high-stakes. There\u2019s a moment when you need to quickly figure out who\u2019s available to receive a transplant and who has the best chance at a successful match based on available medical data. <\/p>\n\n\n\n

The process can be very time-consuming and is typically done by hand, which can create a backlog that results in organ waste. In fact, the US has the highest rate of organ waste in the world. <\/p>\n\n\n\n

Math meets biology for better matchmaking <\/h2>\n\n\n\n

Dr. Eric Weimer, associate professor at the UNC Department of Pathology and Laboratory Medicine, knew that there must be a more efficient process for analyzing the factors known to impact transplant outcomes. \u201cI had the idea many years ago that someone who understands math at a far better level than myself could probably help,\u201d says Weimer.  <\/p>\n\n\n\n

Dr. Katharine Newhall, associate professor of mathematics at UNC Chapel Hill, shared Weimer\u2019s enthusiasm. But the two soon realized that the project needed to move to the cloud. That simple idea\u2014to port the project to Azure\u2014soon became a wider collaboration, leading to unexpected benefits and significant clinical insights. <\/p>\n\n\n\n

\u201cDon\u2019t worry, it\u2019s machine learning\u2014it\u2019ll work\u201d <\/h2>\n\n\n\n

\u201cSo initially, working with Microsoft, it was, \u2018Here\u2019s the existing code, please just port it onto the cloud, translate it over,\u2019\u201d says Newhall. \u201cBut very quickly, Microsoft said, \u2018Maybe we should go back and look at the whole problem. Maybe there\u2019s a more efficient way to handle the data and build some machine learning models.\u201d So Newhall and Weimer began teaching a team of Azure data scientists some of the science and logic behind their project. <\/p>\n\n\n\n

\u201cOur collaboration really came out of that,\u201d says Newhall, \u201cof how we can meet in the middle between mathematics and biology.\u201d The Azure data scientists soon identified opportunities where Newhall and Weimer could take advantage of data sets that they weren\u2019t using because of gaps in the data. \u201cAnd Eric and I go, \u2018Yeah, but scientifically, that doesn\u2019t really make sense,\u2019\u201d says Newhall. \u201cAnd they\u2019re going, \u2018No, don\u2019t worry, it\u2019s machine learning. It\u2019ll work.\u2019\u201d <\/p>\n\n\n\n

Then over the course of eight weeks, the data scientists helped Newhall and Weimer operationalize and automate the transplant matching process. Rather than the current standard of one-recipient, one-donor comparisons, the new process allows for evaluation of an essentially unlimited number of recipient-donor pairs.  <\/p>\n\n\n\n

The gold standard for the semi-manual matching process currently used by every US transplant center today predicts donor-recipient matches with 91 percent accuracy. The new models the team developed have increased that accuracy to 98 percent, and seven hospitals are already planning to pilot the team\u2019s work. Those better match-finding results are helping save both money and time\u2014and with organ transplantation, saving time can mean saving lives. (For more details on the improved matching process, see this presentation<\/a> by the UNC team.)<\/p>\n\n\n\n

Prebuilt for progress <\/h2>\n\n\n\n

As the UNC team looks ahead to future projects, they reflected on their experience with Azure, sharing thoughts for fellow scientists who might be considering this sort of research and the role that cloud, AI, and machine learning can play.  <\/p>\n\n\n\n

Newhall said she especially appreciated the prebuilt nature of many Azure tools. \u201cIt\u2019s much faster to learn how to use if it\u2019s sort of already there,\u201d says Newhall. \u201cYou\u2019re like, \u2018OK, I can just tweak this one thing. Oh, I\u2019ll put in this piece and take that piece out, rather than having to build the whole thing from scratch.\u201d She also liked how easy it was for multiple people to use the same code across different systems.  <\/p>\n\n\n\n

Weimer appreciates that you don\u2019t have to be a developer to take advantage of many Azure tools. \u201cThe Microsoft cloud environment spanned the gamut from drag-and-drop options\u2014for the non-coders out there\u2014to customized, write-your-own-Python-level code,\u201d says Weimer. \u201cIt\u2019s important, especially as a biologist, to understand that you don\u2019t have to come in being a programming specialist to start utilizing these resources.\u201d <\/p>\n\n\n\n

If you\u2019re a researcher hoping to learn more about Azure, we encourage you to check out the many resources tailored for scientific research:<\/em> <\/p>\n\n\n\n