TVG Hospital
Intelly Core + Consulting
The Background
Healthcare is a heavily regulated industry. So not only are hospitals conservatives about sharing EMRs, but AI adoption also requires an extra layer of privacy on training data. TVG Hospital has been looking into ways to help doctors identify high-risk patients by their medical histories and DNA results. To achieve this goal, TVG Hospital realizes they need to cooperate with other specialized hospitals and combine with their EMRs for better training data set.
The Challenge
TVG Hospital involves another two hospitals to get the training data. To ensure the training data remains anonymous, each hospital needs to scrub data first and then find a way to combine them. The stripped data then need to be hosted somewhere (preferably third-party place) to be trained, and the final product cannot provide any way to track back to a particular patient.
Why Intelly AI?
Intelly AI team was brought into the project to identify the information to be stripped away, and the proper data format to be imported into Intelly Core platform. Our team built a simple tool to process the EMRs from hospital databases and feed the data into Intelly Core platform automatically. Since Intelly Core is entirely cloud-based and the entity is set up to be accessed through a special VPN provided by Intelly AI, all participating hospitals are comfortable sharing the EMRs for the project. When the model is ready, it is deployed to each hospital's internal system to be used by the doctors.
The Impact
AI adoptions need extra steps to ensure data safety and user privacy in the healthcare industry. This is one of the main reasons that AI adoption rate is lower in this sector. With the Intelly Core platform's flexibility and the Intelly AI team's knowledge of infrastructure, we develop a customized AI adoption plan for healthcare's unique needs. The model has helped doctors to identify high-risk patients in their databases, and doctors can be proactive about their preventative care and tailored monitoring services. Other healthcare providers and EMRs system providers can use the same plan to train new models should there be a need for other purposes.