The Real Revolution of Medvice


Recent years have seen an explosion of AI use in medical applications, as researchers and innovators find ever more ways to explore the incredible potential of machine learning to improve patient care and health outcomes. However, many of these applications, however sophisticated their algorithms, are doomed to limited success, because they are built on limited datasets.

In fact, the medical datasets that are currently being mined for artificial intelligence applications are deeply flawed. Data scientists end up spending up to 80% of their time simply acquiring and preparing data for use (source: https://hbr.org/2016/12/breaking-down-data-silos), rather than on its application or analysis. Medical data is often derived from “textbook” examples when real-world medical symptoms are more complex and less linear, or based on datasets that are old, messy, and inaccurate. Despite millions of dollars in development, and years of training, when IBM's Watson, arguably the most famous AI in the world, attempted to understand and interpret medical acronyms, abbreviated phrases, and different writing styles, it struggled to extract valuable information from medical records, and has yet to become a trusted partner in medical diagnosis and treatment plans (source: https://www.statnews.com/2017/09/05/watson-ibm-cancer). Ultimately, current medical databases simply do not have data that is properly cleaned and optimized for machine learning, and are often not based on validated cases.

While low quality datasets impede effective machine learning systems everywhere, they are potentially devastating in medical applications. These decisions can be life-changing, or even life-threatening. A medical AI that is biased, incomplete, or simply inaccurate is a risk that no patient or health care provider should have to accept.

Medvice is changing all that, from the ground up. Our groundbreaking partnerships with patients and clinics allow us to create a dataset that is clean, validated, and accurate from the very start. The Medvice.Ai predictive model is built on real-world data that has been optimized for machine learning from the outset, and developed to meet the highest medical standards. Rather than spending time cleaning, validating, and grooming data, the team at Medvice can focus on refining and improving AI decision making, and on data analysis and application. The entirely new data-capturing and validating methodology of Medvice opens up revolutionary new applications for machine learning and AI-assisted medical decision making.


Share this article