What is McCoy?
One small step for scientist, one giant leap for science-kind
It is no secret that predictive analytics will play a large role in the future of healthcare delivery. Techniques that utilize deep learning and artificial intelligence show great promise as technological aids in preventing errors, reducing over-utilization, screening for fraud, and accomplishing the most important goal: making better medical decisions. The medical hierarchy exists to funnel patient management toward best medical practices, and machine learning will increasingly contribute to the efficiency and effectiveness of that process.
Journals were created to disperse knowledge by disseminating well-supported hypotheses throughout the scientific community. Within healthcare, medical journals predominantly function to distribute best practices for drug development, device innovation, disease education, and treatment guidelines. The McCoy platform is an innovative medium for distributing scientific knowledge in the form of well-functioning algorithms, within healthcare, while protecting the intellectual property rights of institutions and authorship rights of researchers. That’s one small step for scientist, one giant leap for science-kind.
Explaining McCoy is easiest using a simpler algorithm than deep learning. The same architecture principles apply for algorithms, such as the familiar linear regression. Linear regression analysis fits a line or curve to demonstrate a relationship between variables. Imagine we are fitting a line that represents the relationship between the number of grams of sugar consumed per day and the risk of developing type II diabetes. If there is a linear relationship between these two variables, one can publish the coefficients (m and b) associated with the regression analysis linear fit of “y = mx + b” where x is the number of grams of sugar consumed per day and y is the risk of developing type II diabetes. The McCoy platform allows the researcher who authored the study to deploy a version of the algorithm as a web application with one small step. Any researcher, institution, or entity can then use the algorithm by submitting an x through the application and getting the output y without ever gaining access to m or b, or even seeing the overall structure of the equation.
With McCoy, the hypothetical diabetes and sugar consumption research can now be used everywhere within the scientific community, through an easy interface, without anyone having access to private information or methodology. This approach safeguards intellectual property rights while dramatically decreasing barriers to scale and test predictive analytics. For example, the groundbreaking predictive model determining the risk of type II diabetes can now be validated prospectively with cohorts of patients around the country, without additional software development.
Media contact: Misha Herscu, firstname.lastname@example.org, 413 320–6636