Components of a modern health-data strategy

Perspective – Focus on designing and creating stable, data-enabled offerings. Develop business requirements that support frontline workers, minimize data entry burden, and improve the patient experience. Create a safe environment for testing ideas, experimenting, and capturing lessons learned. Use these lessons to drive future iterations. Treat data as a critical organizational asset, and accord the same level of importance that is associated with other critical assets.

Ownership – Share ownership of problems and outcomes between business and technology leaders. They must define the strategic business objectives for the organization, including patient outcomes, cost efficiency, and regulatory compliance. For example, you can establish a Cloud Center of Excellence (CCoE) with engagement of both business and IT leadership. A CCoE helps to create joint responsibility for accelerating business adoption and value. At the same time, a CCoE embraces the innovation potential of the cloud and helps ensure a well-architected data solution.

Data literacy – Promote data literacy by establishing a data committee that includes clinical and operational representation. Committee leaders should commit to promoting agility, innovation, and a data-oriented mindset across the organization and within their respective business units. Create a roadmap that aligns data literacy and data-driven business transformation. Train and encourage the line-of-business leaders to use decision support systems and make data-based decisions.

Governance – Establish a data governance framework that outlines the policies, procedures, and standards for managing data within your organization. Develop guidelines for data quality, data privacy, data security, and data access. Design these guidelines to facilitate regulatory compliance. Implement the governance framework in stages as you implement business use cases. Create federated or distributed governance models to balance non-negotiable security, privacy, and regulatory concerns with the need to innovate. Identify central data management opportunities (for example, a central patient index, a unified data catalog). Assess potential impact on the enterprise in unifying multimodal data.

Simultaneously, governance should facilitate democratization of data for fast, intuitive access to data for those who need it, helping users feel empowered, not controlled. To meet governance requirements more efficiently and with less burden on frontline staff, use purpose-built AWS healthcare compliance tools and best practices. Wherever possible, provide self-service tools to reduce the impact on the data and analyst teams.

Artifacts – Define and use artifacts that improve collaboration and data sharing across different teams and departments. Key artifacts include data catalogs, data dictionaries, and data models. For example, use AWS Glue Data Catalog to catalog data. Use Amazon DataZone and AWS Clean Rooms to share specific data or data insights within and across healthcare organizations without compromising patient privacy or violating HIPAA compliance requirements.

Data architecture – Design and continuously refine your data architecture. An architecture that supports a modern health data strategy should embrace multimodal data assets. Adopt a domain-driven approach to handling multimodal data by decoupling data producers from consumers within the architecture. Consider storage, retention, and format. Place an emphasis on ease of access and use, facilitated by robust metadata management.

Healthcare-specific needs, such as regulatory compliance and consent management, should help define data-handling policies and procedures. Consider defining the central data standards that are required to uniquely define business entities such as patients, providers, and employees. Reduce process complexity by defining and creating de-identified datasets to help with accelerating use cases that do not require access to Protected Health Information (PHI).

Technology – Adopt a cloud-based architecture that uses purpose-built services based on the business needs at hand. Create solutions where your organization needs to innovate, but use off-the-shelf solutions and managed services when possible to reduce keep your teams focused on innovation. For example, use predictive analytics to identify vulnerable or at-risk patients for proactive outreach and care. Use Amazon Comprehend Medical to query and extract information from unstructured and semistructured data such as medical notes. Use AWS HealthImaging to help frontline workers process medical images more accurately and efficiently.

Democratized access to data – Promote transparency and visibility to organizational data by using cataloging tools such as Amazon DataZone. These tools provide the ability to search and explore available organizational data, to understand data definitions, lifecycle, and lineage, and to request access to data.

Ease of use – The success of your modern health data strategy depends on ease of use. Assess the different levels of data literacy within the organization, and develop a plan to address consumption across a spectrum of users. Assess current data literacy levels across the organization, devise a data literacy curriculum, and identify project opportunities to develop staff and training plans. Consider the following three broad user categories that your staff might fall into, focusing on their needs for training and adoption: