Healthcare analytics models: Build, buy, or both?

Healthcare payers face a rapidly changing marketplace and continually evolving regulatory landscape. These factors have driven an increased need to develop an information-centric business strategy and to swiftly deploy advanced, best-in-class analytic methods. For both enterprise data warehouses (EDWs) and insightful analytics, payers must evaluate whether it is best to build in-house or license pre-configured analytics and point solutions. Distinguishing which of these options is optimal depends on three key factors: time-to-value, resource availability, and integration. When faced with build or buy decisions, sometimes a third option emerges; do both.

Time-to-value

As business needs are identified, it is important for payers to reduce analytic latency and expedite return on investment. Keeping solutions in-house can result in customized models or point solutions for payers with the right expertise, resources, and budget to do so. With an intimate knowledge of end user needs and well-defined definitions and criteria, time-to-value can be accelerated.

Conversely, the decision to buy can also expedite time-to-value if the right vendor is selected. Vendors with flexible solutions and a proven track record of successful and timely implementations can negate issues related to hiring new personnel or re-allocating existing staff. These internal challenges can strain resources and expand budgetary requirements. In-house projects that are not fully realized can cause payers to pivot to a buy option, resulting in sunk costs and prolonged timelines.

Many payers find that buying models or partnering on innovative solutions with the right vendor can allow them to meet the full scope of their business needs. Pairing internal business knowledge and niche skill sets with external expertise can allow payers to leverage existing resources while filling internal gaps, resulting in more flexibility, less resource strain, and an expedited timeline.

Resource availability

Building a solution can require a variety of resources ranging from certified coders, data scientists and clinicians to technical writers and data architects, among other skill sets. These talents can be contracted in or refocused from other areas of the business. In addition to talent, a robust research database may be needed to support specific business activities such as testing, benchmarking, or developing statistically valid methodologies. Payers need to evaluate their internal capabilities against business needs and potential gaps when considering in-house options.

The decision to buy point solutions, methods, or datasets can free up internal staff to focus on value-add activities. A third-party vendor can relieve resource burdens by allowing existing teams to focus on critical business objectives, all the while providing an organized implementation plan and best practices. Payers should also consider the need to alleviate ongoing support and maintenance long-term.

Payers who are better equipped to handle tasks within their own organization may also consider working with a vendor partner with complimentary skill sets to produce solutions. This hybrid approach can produce a mutually beneficial relationship while allowing for a feedback loop. A third-party vendor can supplement existing resources by filling in the gaps with data sets or expertise working in conjunction with current infrastructure.

Integration

Integration with current solutions and existing infrastructure is an important consideration in build or buy decisions. Algorithms, models, and BI tools that work together allow for more robust insights. Consistency in defining metrics and methodology will mitigate downstream discrepancies.

When making the decision to buy, a quality vendor should provide well-documented methodologies aligned with national reporting standards. Vendors with an existing suite of interconnected models and analytic expertise can ensure solutions work together, remain consistent, and are scalable and adaptable as business needs change.

Specifications must be clearly defined and present in all solutions. Well integrated solutions can benefit from sharing standard input layouts and consistent definitions, while reducing or eliminating manual intervention. A lack of automation is more likely to introduce errors and result in rework. Users must be able to derive actionable insights quickly and easily while allowing for consistent tracking of outcomes over time.

Value of analytic accelerators

After careful review and consideration of these factors, payers may decide to build. This requires best-in-class talent, data resources, and any necessary licensing and infrastructure to create new models. Throughout this process, deploying talent efficiently on value-added work and differentiating products in the market are still top priorities. Licensing foundational analytic models to build upon can accelerate time to market and allow internal teams to focus on creating products and offerings that impact the bottom line. For example, there are approximately 70,000 ICD-10-CM codes. Many models utilize disease severity and comorbidities as the underpinning of predictive models. Licensing a disease severity model that classifies these codes into manageable groupings of conditions and disease severities can accelerate model creation and decrease ongoing maintenance. By building upon industry-accepted foundational models, payers can increase market acceptance of products further developed in-house.

Summary

In a rapidly evolving market, analytic tools need to be on par with industry standards and constantly evolving to meet accelerating objectives. The time-to-value, resource availability, and ability to integrate efficiently into existing infrastructure are key priorities in making the right decision. Payers making build and/or buy decisions should focus on options that reduce latency, expedite ROI, and provide well-defined criteria to increase time-to-value. A robust research database is necessary for the creation of new models, as are resources with robust, complimentary skill sets. It is equally important to incorporate well-defined best practices that easily integrate with existing infrastructure for meaningful insights. Interconnected models with clearly defined specs can reduce manual intervention.

The decision to build can be a good use of existing talent and an opportunity for innovation but choosing to buy, in particular with the added benefit of accelerating analytics with underlying foundational models, places clients at the center of an industry built upon delivering the highest grade of analytic tools that are consistently upgraded to exceed organizational goals with wide market acceptance.

Explore healthcare analytics

Schedule a chat