Technical Feasibility of Implementing and Commercializing a Machine Learning Model for Rare Disease Prediction

Wild-type transthyretin amyloid cardiomyopathy is an under-recognized cause of heart failure. Pfizer previously developed a machine learning model that performed well in identifying wild-type vs. nonamyloid However, challenges exist when introducing applications into healthcare, mainly due to restrictions on sharing patient data. This requires the triggering and execution outside developer’s infrastructure using hosts with diverse information technology capabilities. With these barriers mind, we investigated architectural designs facilitate delivery customer. We considered manageability scalability model, host’s maturity, maintenance data privacy, protection Pfizer’s intellectual property. A container-based design, wherein application shipped as container image third-party platform, fulfilled criteria was piloted platform hosted by Philips. Objectives this pilot included defining/testing design technical parameters for image, creating scalable modular framework manage multiple different platforms, exploring communication pattern based clinical decision support Hooks/Cards representational state transfer calls within Philips platform. Implementing may enable earlier identification treatment cardiomyopathy, learnings from lead improved other models healthcare providers, thereby increasing utilization. article also presents overview help others adopt new methodologies ideas commercialization.