Author(s):
Healthcare organizations across the country continue to face growing challenges related to patient no-shows. In imaging services, missed appointments create operational inefficiencies, reduce provider utilization, impact revenue, and delay important preventive care. At St. Luke’s University Health Network, a high rate of mammography no-shows at one location prompted a search for a more proactive and data-driven solution.
Recognizing that traditional scheduling interventions did not sufficiently address the problem, the organization launched a pilot program utilizing predictive analytics to identify patients at risk for nonattendance. The goal was to improve schedule utilization, increase patient engagement, and recover lost capacity while maintaining human oversight of scheduling decisions.
Missed mammography appointments have consequences that extend beyond empty time slots. Unused imaging capacity reduces operational efficiency and can leave providers underutilized. At the same time, missed screenings may delay preventive care and increase the likelihood that patients seek care outside the network, leading to patient leakage and long-term revenue loss.
For an organization committed to both patient access and operational excellence, reducing no-shows became a strategic priority for St. Luke’s University Health Network. However, the underlying causes of nonattendance are often complex, making it difficult to address the issue through standard reminder calls or broad overbooking practices alone. A more targeted approach was needed—one that could identify high-risk appointments before they became missed opportunities.
To address this challenge, the team implemented an Epic-integrated predictive risk-scoring tool. Every mammography screening appointment received a no-show probability score generated from historical attendance patterns and utilization data. These scores enabled scheduling teams to identify appointments with a higher likelihood of nonattendance and take proactive action when appropriate.
Importantly, the predictive model served only as a decision-support tool. Scheduling decisions remained controlled by staff members, ensuring that human judgment, clinical considerations, and operational context continued to guide patient scheduling activities. This approach combined the advantages of data-driven insights with the safeguards necessary to maintain governance, compliance, and patient-centered care.
The pilot ran from June 2025 through December 2025 and focused specifically on reducing mammography no-show rates. A multidisciplinary team consisting of Operations, Information Technology, Finance, and Analytics collaborated throughout the initiative to oversee implementation and evaluate outcomes. Several controls were established to ensure responsible deployment. These included:
The structured governance framework helped ensure that patient experience remained protected while allowing the organization to test innovative scheduling practices.
Traditional scheduling strategies often rely on blanket reminder calls and generalized overbooking practices. While useful, these methods can be inefficient and may negatively affect patient satisfaction.
The predictive analytics model enabled a more focused strategy. High-risk appointments were flagged for proactive outreach, allowing staff to confirm attendance, reschedule when appropriate, and better manage scheduling resources. St. Luke’s also introduced controlled double-booking practices. Initially, scheduling teams double-booked one appointment in the morning and one in the afternoon using a combination of high-risk and low-risk patients. As confidence in the model and workflow grew, the process expanded to two morning and two afternoon double-booked appointments. This measured approach allowed the team to maximize schedule utilization while minimizing disruption to patient flow and staff workload. Throughout the process, scheduling staff retained full decision-making authority, using predictive scores as informational guidance rather than automated directives.
A key component of the initiative was adherence to responsible AI principles. The pilot did not use generative AI technologies or perform automated scheduling or decision-making. Instead, the pilot relied exclusively on predictive risk modeling. By limiting automation, the team reduced operational risk and maintained clinician and staff trust in the technology. Clear communication, stakeholder engagement, and thorough documentation ensured alignment with organizational governance standards and compliance requirements.
This approach demonstrated that AI can support healthcare operations effectively while remaining transparent, accountable, and centered on patient care.
The pilot produced significant improvements in scheduling performance and operational efficiency. Schedule utilization exceeded 100% through controlled double-booking and proactive outreach, enabling more effective use of available appointment capacity.
Additional key outcomes included:
These results demonstrated not only operational gains but also tangible financial benefits, validating the effectiveness of predictive analytics as a scheduling support tool.
The success of the mammography pilot has created a strong foundation for broader adoption across the organization. A phased rollout strategy would allow additional sites to benefit from lessons learned while maintaining appropriate governance and oversight. Beyond mammography, the predictive model has potential applications across other imaging modalities, including:
As adoption expands and workflows mature, predictive accuracy and operational effectiveness are expected to improve further. Each additional implementation creates opportunities to reclaim lost capacity, improve efficiency, and increase patient access to care.
The Mammography No-Show Predictive Analytics Pilot demonstrates how healthcare organizations can leverage data-driven decision support to address long-standing operational challenges. By combining predictive risk scoring with human oversight, St. Luke’s University Health Network successfully reduced no-shows, increased provider utilization, recovered lost revenue, and improved access to preventive care.
The pilot’s success illustrates the value of responsible AI implementation in healthcare operations. Rather than replacing human decision-making, predictive analytics enhanced staff effectiveness and enabled more proactive scheduling interventions. As healthcare systems continue to seek greater efficiency and improved patient outcomes, predictive scheduling models offer a scalable approach for reducing missed appointments and maximizing the value of available clinical resources.
Judy Mushlitz is Director of Central Scheduling, and Tara Calixto is Manager of Central Scheduling at St. Luke’s University and Health Network, Bethlehem, Pennsylvania.
