How to improve patient outcomes with next best actions using a Composable CDP on Databricks

Whether your organization directly delivers healthcare, or supports patients and providers in the larger medical and insurance ecosystem, the rapid expansion of patient data and new technologies (including AI) that leverage it creates an unprecedented opportunity to serve patients and optimize outcomes.

What makes data-driven healthcare journeys challenging and important

On one hand: Discuss challenging regulations around the use of PHI, and the need for BAAs across the technology ecosystem.

On the other hand: Talk about the real-world implications of reaching the right members with the right messages at the right time.

• This isn’t just marketing to sell a product–it’s critical communications that can change how patients address their health and wellness
• You have the chance to more intimately connect with patients to drive better outcomes while achieving key performance metrics like HEDIS and eCQMs

Patient-centered care is the concept of meeting an individual where they are and performing the appropriate activities to achieve the best possible outcomes for that person. The ability to personalize care is backed by data-driven decisions—but these efforts all too often fall short due to a few predominant challenges in healthcare.

The first is that although there has been a huge emphasis on interoperability from CMS (e.g. CMS Health Tech Ecosystem and Interoperability & Patient Access Rule), the systems of record do not contain all the information relevant to the patient, creating a gap in data-informed decision-making. Further complicating interoperability are specific HIPAA regulations around PHI and the need for tools regulated by BAAs to collaborate.

More nuanced than data interoperability are the practical challenges of reaching the right members with the right message at the right time. Unlike marketing in other industries, which might be focused on selling specific products, healthcare outreach is all about using information intelligently to directly address a member's health and wellness. The opportunity exists to connect more deeply with patients and members and partner with them to live healthier lives. This engagement also helps health plans and providers improve key performance metrics, such as HEDIS, STARS, and eCQMs.

Activating the right message at the right time presents another challenge. What channels of communication should you use, and how should you measure responsiveness and effectiveness? For most healthcare organizations, a single platform does not solve every permutation of outreach, so non-technical teams often need to coordinate across multiple systems.

How a Composable CDP for healthcare works

A Composable CDP is a HIPAA-compliant engagement layer on your data foundation.

• It allows you to build a complete patient profile in the data warehouse to power your patient comms and interactions by decoupling data warehouse operations and outreach operations
• Composable CDP 101:
  • Data lives in Databricks
  • Hightouch offers modules to collect data into Databricks; unify it into complete member profiles; activate it across approved/compliant channels.

Data access and usage are highly governed in Databricks

• Data remains in your cloud storage managed by Databricks
• PHI can be masked and managed at a fine-grained level, by row, column, or attribute tagging (see attribute based access controls)
• Controls ensure only the right people access data
• Controls ensure only permitted data flows to permitted BAA-signed tools in your ecosystem

Patient and Member services teams can self-serve within your governed parameters in Hightouch

• Create segments and evergreen journeys based on specific signals from the member journey
  • Analyze segment composition and adjust on-the-fly
• Take action on those segments/journeys/triggers
  • Lifecycle teams: Send emails, phone calls, text messages, and more
  • Paid media teams: Start ad campaigns, suppress ads from existing or repeat users
  • Digital teams: Personalize websites and mobile apps in real-time, show relevant messages and offers
  • Most importantly: You can make a single segment or journey that spans all of these teams and touchpoints. Ensure that your members have consistent experiences across every channel.
  • Use AI agents and AI Decisioning to automate and scale these interactions, get even more personalized for each user
  • Analyze results of each campaign, progression through journeys, and more– all from that source of truth data

A Composable CDP (Customer Data Platform) is a common approach to create data-driven decisions that power action. It is designed to solve the aforementioned challenges by leveraging your data as the primary driver and foundation for insights and decision-making, and operates upstream from the points of interaction and activation with customers. This allows organizations to take full advantage of the strengths of their data foundation to power all of their downstream customer interactions.

For many healthcare organizations, Databricks serves as the enterprise data platform of choice. Built on the Lakehouse architecture, it unifies the capabilities of a traditional data warehouse and business intelligence with the advanced AI, machine learning, and real-time processing of a data lake under a single platform. Data is stored within the organization’s cloud tenant, with governance enforced through enterprise security groups to support HIPAA compliance and secure, role-based access. Capabilities such as dynamic row- and column-level masking enable teams to operate from a single source of truth while safeguarding sensitive PHI and PII.

With data secured and appropriate role-based access controls in place, Composable CDPs like Hightouch run on Databricks to enable business users to self-serve data access and usage through AI/BI Genie. These teams can create (and measure, and then adjust) cohorts and evergreen journeys based on specific signals from the member journey. Then, they can trigger actual customer communications in downstream tools based on those customer signals. Lifecycle marketing teams can create key cohorts, such as “patients who are due for an annual physical,” and trigger communications across channels like email, phone, and SMS. Paid media teams can upload cohorts to advertising platforms to launch new ad campaigns or suppress ads for existing members. Digital teams can use their Composable CDP to personalize websites and mobile apps in real-time, showing relevant offers or messages to each visitor. And, most importantly, these teams can build a single cohort or journey that spans all these channels to ensure members have a consistent experience across every touchpoint.

And of course, AI can accelerate all of these marketing workflows, but it’s critically important that it’s appropriately governed to prevent negative (or even dangerous, in the case of health misinformation) member experiences. Hightouch’s Agentic Marketing Platform starts by enabling marketers to analyze their data in Databricks, come up with suggestions, and create segments and journeys faster than ever before with a simple workflow. It goes further to create content for advertising and lifecycle marketing campaigns that operate with complete context of each brand’s style and requirements, while ensuring that humans are the ultimate approvers before anything reaches members. And AI Decisioning can automatically choose which message, channel, and timing to deliver to each member to maximize business-defined outcomes while operating within their guardrails.

Example use cases for data-driven customer engagements

New patient acquisition (paid channels)

• Known target audiences (e.g. people new to a region without a PCP yet) & Lookalike audiences
• Efficiency levers
  • Suppression audiences (prevent advertising to existing customers)
  • Conversion APIs to provide a signal for ad effectiveness
  • Measure the incrementality of ad campaigns

Triggered customer engagement for key moments in their lifecycle

• Appointment reminders
• Nudges towards preventative screening (“you’re due for a colonoscopy”)
• Reminders about taking & refilling medications

Evergreen/proactive member health management journeys

• Example: recommended reading sent to customers based on their health history
• Example: ongoing flow for management of a chronic condition like diabetes, or for an ongoing program like weight management while taking a GLP-1
• Example: post-surgery care sequence
• Example: maternity care sequence

Sync key patient and medication data into operational systems to determine cost savings

• Ex: Identify medications that might be out of policy, and identify lower-cost alternatives

Your organization can get started with simple, automated communications that have a positive impact on patients, like appointment reminders, medication pickup or refill notifications, and screening checkpoints. Use cases like these help lay the foundation for more complex workflows and potentially greater impact down the road.

Where a Composable CDP clearly provides an advantageous approach is in more complex workflows for episodes of care, such as pregnancies, and for managing chronic conditions like diabetes and hypertension, as well as for programs like behavioral health and weight management. These programs require a more sophisticated approach with higher levels of personalization and flexibility that account for the nuances of each patient’s medical history and preferences. These programs merit special focus due to the critical nature of the care required and the limited resources available from health plans and providers to support them. Using your data and AI, and empowering outreach teams with self-service data access, helps them maximize impact.

Conclusion

Many organizations are active in engaging with patients outside the clinical setting. A Composable CDP enhances these capabilities by giving organizations self-serve capabilities to access and leverage their data across channels, from their central source-of-truth patient data in Databricks. This approach allows organizations to move faster, to more flexibly tailor communications across multiple channels, and to better engage patients and members based on their preferences and history. Explore more today by reaching out.