JMIR Publications

ABSTRACT

Machine learning applications promise to augment clinical capabilities, and at least 64 models have been already cleared by the FDA. These tools are developed, shared, and used in an environment in which regulations and market forces remain immature. An important consideration when evaluating this environment is the introduction of open source solutions, in which innovations are freely shared, and which have long been a facet of digital culture. We discuss the feasibility and implications of open source machine learning in a healthcare infrastructure built upon proprietization. Decreased cost of development as compared to drugs and devices, a longstanding culture of open source products in other industries, and the beginnings of machine learning-friendly regulatory pathways together allow for the development and deployment of open source machine learning models. Such tools have distinct advantages, including enhanced product integrity, customizability, and decreased cost leading to increased access. However, significant questions regarding engineering concerns about implementation infrastructure and model safety, a lack of incentives from intellectual property, and nebulous liability rules significantly complicate the ability to develop such open-source models. Ultimately, the reconciliation of open source machine learning and the proprietization-driven healthcare environment requires that policymakers, regulators, and healthcare organizations actively craft a conducive market in which innovative developers will continue to both work and share.