The Data Science Adoption Model guides organizations through five critical levels of capabilities—guided analytics, deploying predictive models, building predictive models, retrospective comparison, and prescriptive optimization. By filling common gaps in a traditionally linear analytic adoption framework (e.g., ineffective process or insufficient healthcare how-to), the Data Science Adoption Model provides non-sequential steps to help data science practitioners and leaders direct their analytic investments and deliver real value. And with COVID-19 driving the demand for predictive models to improve the effectiveness of organizational response plans, actionable data science has rapidly become a healthcare imperative.
With its five-level approach, the Data Science Adoption Model (Figure 1) bridges the gap between interest in data science and its real-world application. The model gives a framework for using data science, breaking down its components so users know where they are on the adoption continuum, where they need to go, and how they can progress over time. It matches questions and recommendations with capabilities to resolve particular challenges and differentiate steps that require predictive models.
Level 1—Guided Analytics: Answer questions with more precision to share interpretations and have discussions.
Level 2—Deploying Predictive Models: Explain predictive models to help leaders build trust among team members in model output.
Level 3—Building Predictive Models: Positioned outside of self service, encouraging users to consider how to optimize models over time versus rushing through the automated building process.
Level 4—Retrospective Comparison: Rigorous, plausible data assessments to drive solid conclusion and next steps.
Level 5—Prescriptive Optimization: Solid groundwork to set up for analytics for success from the front end of the data science journey.