Biohacking to Improve Everyone's Health Reaches Top Ten in 2nd Round of Biomarkers of Aging Competition
Biohacking to Improve Everyone's Health, the team of Biohackers intending to compete in XPRIZE Healthspan, have reached the top ten in prediction of mortality using methylation data in the second round of the Biomarkers of Aging competition.
The contest aims to use data on methylation of DNA sites and other biomarkers to determine the biological age of individuals and predict outcomes such as mortality.
The Biohacking team, including software developer and machine learning expert Samuel Collingwood Smith, used their own proprietary software called LearnSilver to develop their prediction models, in combination with other tools. LearnSilver is a .NET C# / C++ / CUDA software library that allows execution of complex neural networks, including recurrent networks, on consumer hardware along with efficient serialisation. The library can execute in single-threaded mode on a CPU, or multi-threaded mode and it can also leverage nVidia hardware for massively parallel operations.
Sam Smith said, "It was extremely satisfying to come 8th on one of the leaderboards with a new product and a shoestring budget against long-standing experts and large corporations. Although methylation of DNA sites might not seem obviously to correlate to age at death, in fact there is a connection because methylation correlates to current biological age which in turn can be used to predict life expectancy in conjunction with existing life-expectancy data."
John Hemming, Biohacking CEO and Team Leader said, "Anyone can produce AI if they are willing to spend a $100 billion dollars as Microsoft and OpenAI are doing, or even a humbler $100 million per training round . The challenges in leveraging these technologies are about reducing costs and increasing efficiencies. Our models were developed, trained and executed on a single machine using a single Intel i9 processor and an nVidia GTX 1080 Ti Graphics Card for CUDA. Thanks to LearnSilver, we came in the top ten using affordable consumer grade hardware."
The contest aims to use data on methylation of DNA sites and other biomarkers to determine the biological age of individuals and predict outcomes such as mortality.
The Biohacking team, including software developer and machine learning expert Samuel Collingwood Smith, used their own proprietary software called LearnSilver to develop their prediction models, in combination with other tools. LearnSilver is a .NET C# / C++ / CUDA software library that allows execution of complex neural networks, including recurrent networks, on consumer hardware along with efficient serialisation. The library can execute in single-threaded mode on a CPU, or multi-threaded mode and it can also leverage nVidia hardware for massively parallel operations.
Sam Smith said, "It was extremely satisfying to come 8th on one of the leaderboards with a new product and a shoestring budget against long-standing experts and large corporations. Although methylation of DNA sites might not seem obviously to correlate to age at death, in fact there is a connection because methylation correlates to current biological age which in turn can be used to predict life expectancy in conjunction with existing life-expectancy data."
John Hemming, Biohacking CEO and Team Leader said, "Anyone can produce AI if they are willing to spend a $100 billion dollars as Microsoft and OpenAI are doing, or even a humbler $100 million per training round . The challenges in leveraging these technologies are about reducing costs and increasing efficiencies. Our models were developed, trained and executed on a single machine using a single Intel i9 processor and an nVidia GTX 1080 Ti Graphics Card for CUDA. Thanks to LearnSilver, we came in the top ten using affordable consumer grade hardware."
Comments