High-Risk Follow Through
Project status
Collaborators
Jennifer Lee, MD
Tessa Cook, MD, PhD
Ann Huffenberger, DBA
Hayley Cassidy
Briana Neely
Innovation leads
Awards
CIO 100 Award, 2026
Opportunity
Patients often receive important follow-up recommendations – such as imaging or specialist visits – but these next steps do not always occur. As a result, serious conditions can go undetected or untreated, even when patients have ongoing contact with the health system. This gap in follow-through can lead to worse health outcomes, increased burden on clinicians to track outstanding care, and greater legal and operational risk for the system.
This moment in the care journey is critical: Delays in follow-up can mean missed opportunities to intervene early, when conditions are more treatable. Improving how patients are identified and supported after a concerning finding creates an opportunity to close this gap, ensuring that important recommendations lead to timely care rather than getting lost over time.
Intervention
High-Risk Follow Through is a centralized program that helps identify patients with overdue, high-risk care recommendations and supports them in completing the next steps. It combines data, technology, and a care team to proactively find patients who need follow-up and guide them through scheduling and completing recommended care.
Once a patient is identified, High-Risk Follow Through uses a combination of outreach methods – including text messages, phone calls, and clinician reminders – to prompt and assist with scheduling. Patients who need additional support can be connected to a centralized clinic or offered virtual visits if they do not already have a care team. This coordinated approach helps ensure that patients receive timely follow-up while reducing the administrative burden on clinicians.
Impact
High-Risk Follow Through launched in April 2025 and operates across the health system. Initially we have focused on patients with needing follow-up imaging for lung nodules, and we are working to expand the program to more clinical areas such as surveillance colonoscopies.
As of May 2026, the program has supported nearly 8,000 follow-up interventions, leading to more than 1,300 completed CT scans for pulmonary patients and the detection of 12 cancer cases.
By improving how patients are identified and guided to complete recommended care, the program is increasing access to timely services, enabling earlier diagnosis of serious conditions, and helping the health system better capture and coordinate care across a growing set of clinical populations.
Way to Health Specs
Learn more about the platformInnovation Methods
Assumptions matrix
Assumptions matrix
Before the program was designed, we outlined assumptions to ensure the workflow was sound and the appropriate partners were being included.
Assumptions matrix
An assumption is a statement about something that must be true for your solution to work.
When you have defined a solution you'd like to test, ask yourself, "What must be true for this to work?" Once you have a complete list, plot your assumptions on a 2x2 matrix where one axis is how certain you are that your assumption is accurate and the other is how detrimental it will be if it is not.
Mapping assumptions will help you determine what you need to test to de-risk a potential solution. Assumptions that you are uncertain about and that are crucial for your solution to work are your riskiest assumptions.
Fake back end
Fake back end
Before automating identification and outreach, we completed outreach manually to fine-tune the workflow and ensure the efficacy of the intervention.
Fake back end
It is essential to validate feasibility and understand user needs before investing in the design and development of a product or service.
A fake back end is a temporary, usually unsustainable, structure that presents as a real service to users but is not fully developed on the back end.
Fake back ends can help you answer the questions, "What happens if people use this?" and "Does this move the needle?"
As opposed to fake front ends, fake back ends can produce a real outcome for target users on a small scale. For example, suppose you pretend to be the automated back end of a two-way texting service during a pilot. In that case, the user will receive answers from the service, just ones generated by you instead of automation.