This blog is a contribution from our customer, Personify Health, the first personalized health platform company. Learn how their team reduces AI model development time by 82% for personalized healthcare.
Personify Health is a global company empowering diverse and unique businesses — and diverse and unique people — to engage more deeply in health at a lower cost. We’ve brought together health plan administration, holistic wellbeing, and comprehensive navigation solutions to create the industry’s first and only personalized health platform.
We’re bridging the gap between the enormous potential of AI and the need for smarter healthcare. Our AI-powered platform helps our 19 million members across 190+ countries to proactively manage their health with highly personalized recommendations and content that spans mental health, nutrition, sleep, movement, exercise guidance, and more.
To achieve personalization at this level and drive efficiency throughout the company, we’re accelerating AI adoption with more than 15 predictive and generative AI projects currently underway. DataRobot has been essential for improving the efficiency and effectiveness of our expert team of data scientists, engineers, and data analysts.
Why accuracy and governance are essential for AI in healthcare
Personify Health has the best and brightest data scientists, engineers, and analysts on our team. But in order for them to do their best work, they needed a toolset that would reduce daily headaches like tool overload, clunky workflows, and duplicative work. As CTO, I want my people focused on delivering results, not managing bad processes.
I’m a strong believer in eliminating silos across the organization. Our engineers, data scientists, and analysts must come together on one platform to scale our development cycles and productivity gains. But we must be thoughtful with each decision, particularly as healthcare is a highly regulated field with numerous ethical concerns.
We need to blend AI with human expertise for a close personal touch along the entire customer journey. Our members trust us to guide their health and wellbeing through recommendations provided by advanced predictive analytics, while our health coaches rely on generative AI co-pilots to make them more effective partners to our community members.
Internally, we also use generative AI to drive productivity with helper bots for different business functions. Trust in each of these AI applications is absolutely essential for us to do our jobs and support our member community in continuing to improve their health.
Predictive models are critical to the personalized experience we provide on our platform. To inform Personify’s predictions, we consider a dataset that encompasses approximately 275 million people with 700 features. We need to train models on constantly changing data — at scale — across hundreds of models multiplied by millions of members.
DataRobot allows our team to train multiple models simultaneously and compare outputs for the greatest possible accuracy. Their end-to-end platform means we have the tools to quickly deliver results our customers can trust.
Supporting Personify’s model responsibility standards
Centralized tooling further supports our commitment to social and model responsibility with the right infusion of ethics, privacy, and governance. At Personify, here’s how we’re taking a thoughtful approach to AI adoption and use across the company:
A steering board
An AI executive steering board oversees all projects to help guide our decisions at the leadership level.
Research and development
An R&D team researches and builds the foundation for our generative AI-based frameworks for use cases in operations, commercial marketing, and content generation.
Governance team
AI policy governance is critical in how we work with data. Our AI policies propagate to our 70+ sponsors on the platform so everyone understands how we protect data. Our governance team guides AI policies with input from our legal counsel.
Privacy and bias mitigation
We work with anonymized data sets, which keep member data private. At the same time, it eliminates the possibility of bias in our models.
A unifying platform
To reduce silos and accelerate model time, we chose DataRobot to bring all our teams and predictive models in one place. We’ve been able to rip apart those silos across the organization, which is fundamental to our success.
The platform unifies our AI analytics across teams, technologies, and data sources, and speeds our time to production. Instead of taking three weeks to get an acceptable model, we can do it in 21 hours and create multiple models in parallel.
DataRobot also supports our governance framework with essential explainability. Having visibility behind models helps us quickly understand, choose, and gain acceptance of the right models.
Building momentum for generative AI in personalized healthcare
Looking ahead, generative AI will become a greater part of how we enable our organization. With the foundation we’ve already built in DataRobot across predictive AI and generative AI models, I’m excited for Personify Health’s future.
We’re already unifying all of our generative and predictive AI models within DataRobot to further improve how we manage and monitor them to drive meaningful results for the business and our members. This platform is core to many of our objectives as we continue to lead the future of AI in healthcare.
For example, we foresee having our own AI conversational hub on the backend that plugs into all our products to answer questions for even greater responsiveness.
We’ve also been experimenting with DataRobot Application Templates, which will give us repeatable use-case logic that the team can adapt for specific business problems. This solves a significant headache for our team and will let us move at an even faster pace.
Together, Personify Health is empowering consumers with smarter recommendations that are held to the highest ethical and legal standards. DataRobot makes it so much easier to build and manage AI models, which means my team can focus on delivering better health and wellbeing recommendations to our customers.
Check out Personify Health’s customer spotlight to learn more about how they reduced AI model development time by 82%.
About the author
As Chief Technology Officer at Personify Health, Amit Jain draws on more than 20 years of expertise in SaaS engineering leadership, cloud architecture, data science, and machine learning to create and execute a vision that impacts millions of people globally.