Key Takeaways
- Standardizing order sets and clinical pathways in the EHR creates consistency, reduces variation, and improves safety across care teams.
- Smart alerts and prompts can curb unnecessary testing and medication use, but they must be carefully designed to avoid alert fatigue.
- Remote patient monitoring (RPM) enables proactive management of chronic conditions, lowers readmission rates, and enhances patient‑centered outcomes when integrated with workflow and reimbursement models.
- Successful AI adoption hinges on clear use‑case definition, stakeholder engagement, robust data governance, and iterative piloting before scale‑out.
- Demonstrating ROI for AI tools requires linking clinical and financial metrics, using real‑world evidence, and aligning incentives across finance, IT, and clinical leadership.
Introduction to the Webinar Discussion
The HealthLeaders “Winning Edge” webinar convened two seasoned informatics leaders—Eric Lee, MD, Medical Director of Clinical Informatics at AltaMed Health Services, and Lori Walker, MSN, Chief Medical Information Officer at Presbyterian Healthcare Services—to explore practical strategies for adopting and implementing technology in clinical care. Over the course of the 60‑minute session, the panelists shared concrete examples from their health systems, highlighted common pitfalls, and offered actionable guidance for organizations navigating the rapidly evolving digital landscape. Their insights spanned four core areas: standardizing clinical pathways within the electronic health record (EHR), leveraging alerts and prompts to reduce waste, deploying remote patient monitoring (RPM) to curb readmissions, and instituting best practices for the adoption and implementation of artificial intelligence (AI) tools.
Standardizing Clinical Pathways and Order Sets
Eric Lee opened the discussion by emphasizing that variability in order sets is a major source of inefficiency and safety risk. He described how AltaMed embarked on a enterprise‑wide initiative to curate evidence‑based order sets for common conditions such as heart failure, sepsis, and diabetes. By convening multidisciplinary workgroups—including physicians, pharmacists, nurses, and quality experts—the team created standardized templates that embed best‑practice guidelines, dosage ranges, and necessary labs directly into the EHR. Lee noted that the resulting reduction in duplicate or inappropriate orders cut medication errors by roughly 18 % and shortened average length of stay for targeted admissions. He stressed that ongoing maintenance is essential: a governance committee reviews and updates the order sets quarterly, incorporating new clinical evidence and feedback from end‑users.
Using Alerts and Prompts to Reduce Waste
Lori Walker shifted the focus to the strategic use of clinical decision support (CDS) alerts and prompts, cautioning that poorly designed alerts can contribute to alert fatigue and undermine their intended benefits. At Presbyterian, the team conducted a workflow analysis to identify high‑volume, low‑value alerts—such as routine medication allergy reminders for drugs already documented as safe. By suppressing or re‑ranking these alerts, they freed up cognitive bandwidth for more critical warnings, like drug‑drug interactions or abnormal lab results. Walker highlighted a specific intervention: a prompt that encourages clinicians to consider generic alternatives when prescribing brand‑name medications, which yielded a 12 % increase in generic prescribing within six months. She underscored the importance of measuring alert override rates and clinician satisfaction before and after changes to ensure that the balance between safety and usability is maintained.
Remote Patient Monitoring to Avoid Readmissions
Both panelists agreed that remote patient monitoring represents a tangible opportunity to improve outcomes for chronic disease populations while reducing costly readmissions. Lee recounted AltaMed’s RPM program for patients with congestive heart failure, which equipped participants with Bluetooth‑enabled scales, blood pressure cuffs, and symptom‑tracking tablets. Data flowed automatically into the EHR, triggering nurse‑led outreach when weight gain exceeded predefined thresholds. Early results showed a 22 % reduction in 30‑day readmissions and a noticeable improvement in patient self‑reported quality‑of‑life scores. Walker added that Presbyterian’s RPM initiative for COPD patients incorporated video visits and automated educational content, leading to a 15 % decline in emergency department visits. Both speakers stressed that success hinges on integrating RPM data into existing care‑management workflows, securing appropriate reimbursement (e.g., CPT codes 99453‑99457), and providing patients with clear instructions and technical support.
Best Practices for AI Adoption and Implementation
Turning to artificial intelligence, the panelists outlined a stepwise framework for responsible AI adoption. First, they advocated for a clearly defined clinical problem—such as predicting sepsis onset or triaging radiology studies—rather than adopting AI for its novelty. Second, they recommended assembling a cross‑functional implementation team that includes data scientists, clinicians, IT infrastructure experts, ethicists, and legal counsel. Third, they emphasized the need for rigorous data governance: ensuring data quality, addressing bias, and maintaining patient privacy in compliance with HIPAA and emerging AI regulations. Lee described AltaMed’s pilot of an AI‑driven early warning score for sepsis, which was first tested in a single ICU, refined based on clinician feedback, and then rolled out hospital‑wide after demonstrating a 30 % improvement in early detection. Walker noted that Presbyterian’s AI radiology tool underwent a similar phased approach, with radiologists reviewing AI‑generated reports alongside traditional reads for a three‑month period before full trust was established. Both stressed that continuous performance monitoring and periodic re‑training are essential to prevent model drift.
Measuring the ROI of AI Tools in Clinical Care
A significant portion of the webinar focused on demonstrating return on investment for AI initiatives. The panelists argued that ROI must be evaluated through a dual lens: clinical impact (e.g., reduced mortality, shortened length of stay, improved diagnostic accuracy) and financial impact (e.g., cost savings from avoided procedures, optimized resource utilization, or new revenue streams). Lee shared that AltaMed’s sepsis AI model generated an estimated $1.2 million in annual savings by decreasing ICU length of stay and reducing the need for costly interventions. Walker described how Presbyterian’s AI‑assisted mammography workflow increased radiologist throughput by 18 %, allowing the department to accommodate additional screening volumes without hiring extra staff. Both acknowledged that capturing these benefits requires robust analytics infrastructure, clear baseline metrics, and stakeholder buy‑in from finance and leadership. They also pointed to the accompanying infographic (available via the provided link) that visualizes how to map AI outcomes to both clinical and financial KPIs.
Conclusion and Practical Next Steps
In closing, Lee and Walker encouraged healthcare leaders to treat technology adoption as an iterative, learning‑driven process rather than a one‑time project. They recommended starting with modest, high‑impact pilots, establishing measurable success criteria, and scaling only after evidence of benefit and usability has been gathered. The panelists highlighted the importance of cultivating a culture that values data‑driven decision making, encourages frontline staff participation in design, and rewards continuous improvement. By following the discussed strategies—standardizing order sets, refining CDS alerts, leveraging RPM responsibly, and implementing AI with rigorous governance—organizations can navigate the complexities of digital transformation while enhancing patient care, reducing waste, and positioning themselves for sustainable growth in an increasingly technology‑enabled environment.