Earlier Intervention Powered by AI Agents
.png)
Northwestern Medicine Uses Hippocratic AI to Relieve Administrative Burden and Increase Access to Health Screenings
Each year, Northwestern Medicine nurses make tens of thousands of phone calls to prepare patients for procedures, to follow up after hospital stays and to check in between visits.
These calls are important for keeping patients safe and informed. However, they can be hard to complete as they often happen between in-person care activities. Sometimes, patients are not reached, or nurses cannot always answer nonmedical questions from patients.
What Are AI Agents?
In other words: these AI voice agents can perform tasks in real time, transferring calls to human agents when appropriate. Northwestern Medicine tested these agents over four months in two use cases.
Discover how Northwestern Medicine used the agents for lung cancer screenings and annual wellness visit reminders, and what the potential impact could be for patients.
Case Study No. 1: Lung Cancer Screening
The goal of lung cancer screening is to identify cancer at an early stage, when it is more likely to be curable. Nearly 80% of cases aren’t discovered until advanced stages, when symptoms begin to appear.
Among those screened for lung cancer, 40% have findings that require follow-up or monitoring, while 2% to 3% of the screenings result in a cancer diagnosis.
Therefore, one opportunity to use generative AI agents was calling to remind patients with open orders to schedule their lung cancer screening. More than 1,700 patients had outstanding orders for chest CT scans that would expire at the end of 2025.
The generative AI agents called all 1,705 patients over five days and transferred them to the scheduling line when appropriate. Patients were also given the option to speak directly with staff. As a result:
250
patients scheduled a lung cancer screening
7.4 out of 10
stated they would be likely to recommend talking to an AI agent again
3.2%
of patients opted to speak with staff
Early intervention
allows clinicians to identify and monitor issues early, improving the likelihood of successful treatment if lung cancer is found at an earlier stage.
Case Study No. 2: Wellness Screening Visits
When it comes to scheduling, staff call patients directly. And 80% of those calls go to voicemail. The staff member then documents this in the patient’s electronic medical record, a process that takes approximately seven minutes total.
Northwestern Medicine used Hippocratic AI’s agents to call 2,446 patients to remind them of their annual wellness visit. The agents then transferred patients to a staff member to schedule their appointment. Patients had the option to connect directly with staff at the beginning of the call.
The results:
50
patients scheduled visits
2.25%
of patients opted to speak with staff
While proactive health management is important, it also creates an opportunity to identify concerns that may have gone unnoticed. Of those who had their wellness visit, two cases of suicidal ideation and one case of suspected elder abuse were identified. In each instance, patients were immediately connected with a social worker who provided resources and interventions to help address these concerns.
Improving Patient Experience With AI
This workflow also significantly reduces administrative work for clinicians so they can focus on direct patient care. Implementation of the two pilots resulted in a combined savings of over 120 hours.
Based on the positive outcomes from both cases, Northwestern Medicine continues to use Hippocratic AI’s agents, expanding it where appropriate, including using it for post-surgical and post-discharge follow-up calls.
Generative AI in the healthcare field continues to show promise. Hippocratic AI raised more than $126 million in late 2025, bringing its total funding to $404 million. The company plans to use this money to expand its capabilities and invest in product innovation.
AI can enhance clinicians’ abilities. It can’t replace them. In these case studies, that early intervention could make all the difference.
