As hospitals and health systems intensify their financial planning efforts, revisiting and refining the forecasting techniques and tools that guide their decisions has become paramount. In today's rapidly evolving healthcare landscape, the traditional estimates that many hospitals and health systems rely on are increasingly inadequate. These outdated methods are no longer sufficient for addressing complex, near and long-term challenges, and can hinder effective decision-making and the creation of sustainable value.
However, organizations can employ predictive analytics and modeling techniques—already common in other industries—to provide more accurate estimates of future market-specific and macroeconomic activity than the models that are commonly used by providers today. Importantly, predictive analytics and artificial intelligence (AI) are already being used by many healthcare organizations in their clinical enterprises.
These emerging approaches can help organizations understand the probabilities of changing macroeconomic, market, and organizational conditions on financial planning in real time. For example, organizations might use predictive analytics to analyze inflation or the impact of strategic and operating initiatives with more efficiency than traditional methodologies.
Providers have a significant opportunity to use predictive analytics across their corporate functions as well. For instance, analytics can connect financial plans to budgets, by illuminating the potential impact of changes in volumes on overall financial performance and producing assumptions that can be applied to different time frames.
These approaches have several benefits over existing financial planning and forecasting approaches:
- Predictive analytics can test strategic assumptions quickly—for instance, exploring the impact of a capital project for a new tower and its incremental impact on the financial plan in the context of both expected performance and the market’s competitive landscape.
- Predictive analytics can illustrate complex feedback loops that traditional approaches are not able to simulate. Performance monitoring can occur in real-time—and can help assess what components or assumptions of a financial plan no longer hold true, and when a financial plan needs to be refreshed.
- Democratization of strategic planning and the accessibility of predictive analytics tools can empower organizations of all sizes to engage in sophisticated strategic and financial planning. This democratization can lead to more innovative and competitive smaller organizations.
- Improved resource allocation: By reducing the time and resources needed for strategic financial planning, predictive analytics can enable personnel to focus on higher-value activities, potentially leading to increased productivity and innovation.
- Enhanced risk assessment provides the ability to conduct detailed scenarios and sensitivity analyses for capital projects. This approach can also lead to more informed investment decisions and better capital allocation.
Case Study
Situation:
A large hospital system faced significant challenges in developing accurate forecasts for its annual budget and resource planning. Each year, managers relied heavily on manual methods to estimate patient volumes and resource needs, leading to a time-consuming process that produced consistently inaccurate results. These inaccuracies complicated effective planning, resource allocation, and financial management across the system.
Managers typically based their volume estimates on personal experience and recent trends, often overemphasizing the last few weeks of the year. Additionally, they struggled to account for external factors—such as weather, seasonal illnesses, holidays, and population changes—that can substantially impact patient demand. Without a systematic way to incorporate these variables, the hospital’s volume projections for the upcoming year were regularly off by nearly 20% each month.
Solution:
Kaufman Hall partnered with the hospital system to develop and implement AI-driven predictive analytics models tailored to the unique needs of each department. By integrating historical data spanning three to four years and accounting for external influencing factors, the new bespoke models allowed the hospital to shift from subjective estimates to data-backed forecasts. These models evaluated variables such as weather patterns, respiratory illness trends, and holidays to generate more comprehensive and precise projections.
Results:
The implementation of AI and machine learning models delivered transformative improvements in forecast accuracy and efficiency (Figure 1). The hospital system saw a significant increase in forecast accuracy across all departments. Other results included:
- Sixty departments achieved monthly forecasts that were within just a few percentage points of actual volumes
- For departments with extremely poor budget accuracy through traditional approaches, predictive models improved accuracy by 80%
- Departments with average budget accuracy improved by 29%
- Departments whose accuracy was already classified as good improved by 19%
Figure 1: Improvement in Budget Accuracy by Predictive Models from Traditional Approach to Predictive Approach
These advancements streamlined the forecasting process, reduced the time spent on budget planning, and enabled the hospital to allocate resources more effectively. By adopting predictive analytics, the hospital system gained the ability to respond proactively to anticipated patient demand, driving more strategic and financially sound decision-making.
Beyond the gains from the improvement in forecast accuracy, these models (see Figure 2 for a sample visualization of this approach) were able to produce yearly forecasts at a monthly level for more than 300 units within 30 minutes, greatly reducing the effort required to produce these. Given the speed at which these models can be trained and produce forecasts, the organization is working to implement these models so that they are rerun right before the fiscal year, allowing even more accurate forecasts to be produced. The ability to create these forecasts using current data would have otherwise been impossible to do given the manual effort and time required.
Figure 2: 3 Years of 365-Day Forecasts
The organization is integrating these predictive models into many of their processes to create better hiring plans, manage productivity, estimate the impact of the flu season, negotiate and plan for contract labor, build more accurate budgets and monthly budget spreads, and equip their leaders with more actionable forecast data.
The benefits produced from this work only scratch the surface of how these models can be used to inform planning, decision-making, and ongoing management. Importantly, most organizations today already have the data needed to leverage these types of approaches, so the potential return on investment once predictive analytics and process capabilities are in place is quite significant.
Conclusion: Get started now, regardless of where you are
Regardless of the current level of complexity in an organization’s approach to financial planning, there are steps that leaders can take immediately to improve their efficiency and accuracy. Predictive modeling is not a “one size fits all” proposition. While some larger systems may have more predictive capabilities than smaller providers currently, all organizations have room for improvement.
We advise organizations seeking to build successful analytics initiatives (see Figure 3 for key components of these initiatives) to consider the following key principles:
- Start small but broadly and move in a trust-inducing way
- Begin to apply predictive analytics to projections but use this to supplement not replace current practice
- Ensure that your organization can communicate how well these models work and gather “buy in” from key constituents and stakeholders
- Over time, demonstrate how capable these models are, highlight advanced areas of focus, develop a roadmap for increased sophistication, and begin to replace traditional approaches
- Monitor and review past analyses compared to actual results
Figure 3: Key Components of Predictive Analytics Initiatives
