The Challenge of Post-Operative Infections

Post-operative infections represent a meaningful hurdle in patient recovery, impacting both individual well-being and healthcare system efficiency.Currently, a substantial percentage of surgical patients, ranging from 5% to 20%, experience complications such as pulmonary infections, urinary tract infections, and, in more severe instances, sepsis. These infections often lead to prolonged hospital stays, readmissions, and the need for extensive treatment, placing a considerable burden on resources and diminishing patient quality of life.

Introducing Periscope: An AI Solution

A groundbreaking artificial intelligence (AI) model, dubbed “Periscope,” is poised to transform how medical professionals identify and manage patients at elevated risk of post-operative infections. Developed and rigorously tested across hospitals in the Netherlands and Belgium, Periscope offers the potential for early intervention, optimized clinical decision-making, and a significant reduction in post-surgical complications.

How Periscope Works: Data-driven Insights

Periscope’s development stemmed from the critical need to improve the accuracy of predicting which patients are most likely to develop infections following surgery. The AI model was trained using a decade’s worth of anonymized patient data (Electronic Patient Data – EPD) from over 250,000 surgical procedures across three hospitals, including Leiden University Medical Center (LUMC), Radboudumc, and Oost-Limburg Hospital in Belgium.

The algorithm analyzes a range of factors, including:

• Patient history of infections
• Pre-existing conditions (comorbidities) such as diabetes
• Vital signs like heart rate, blood pressure, and weight

By correlating this facts with the actual occurrence of post-operative infections, Periscope learns to identify patterns and predict risk with increasing accuracy.

Periscope’s Performance: Comparable to Experienced Clinicians

During trials at LUMC, Periscope’s predictions were compared against those made by experienced doctors. The results demonstrated that the AI model performed comparably to seasoned clinicians and even surpassed them in cases where the medical staff had less experience. This suggests that Periscope can serve as a valuable tool, especially for less experienced medical professionals, in identifying high-risk patients.

Impact on Clinical Practice: Streamlining Workflow and Enhancing Decision-Making

Periscope is designed to be integrated seamlessly into the clinical workflow. Doctors, residents, and nurses in departments such as general surgery, orthopedics, and neurosurgery will have access to the tool. The system will display a patient’s infection risk as a percentage – categorized as low, medium, or high – alongside other relevant data, all consolidated in one accessible location.

This streamlined approach eliminates the need for medical staff to sift through extensive medical records to assess infection risk, saving valuable time and improving efficiency. As Dr. Siri Van der Meijden, the lead researcher behind Periscope, explains:

The fact that we certainly no what patients are at a higher risk of infections will allow us to monitor them more carefully and intervene earlier. This will help improve the quality of life of patients and reduce the impact of infections on the health system.
Dr. Siri Van der Meijden, Leiden University Medical center

Periscope is expected to aid in decisions regarding patient discharge and outpatient care, ensuring that resources are allocated effectively to those who need them most.

Implementation and Future Development

While Periscope is ready for deployment,integrating it into existing electronic health record systems presents a significant challenge. LUMC specialists anticipate that the tool will be implemented in clinical practice by mid-2026. plans are also underway to introduce Periscope to other hospitals.

The development team is committed to continuously improving the model. Unlike some AI tools that learn automatically, Periscope requires the addition of new data to enhance its predictive capabilities. Future enhancements may include the ability to estimate the probability of specific types of infections and to make predictions before surgery.

Looking ahead, the team envisions expanding Periscope’s capabilities to predict other post-operative complications, such as the risk of bleeding, re-intervention, or even mortality.

Significant Considerations

It is crucial to emphasize that Periscope is intended to augment, not replace, clinical judgment. As Dr. Van der Meijden stresses, The clinical protocols and the doctor’s opinion will continue to have priority. Periscope serves as a valuable tool to inform decision-making, but the ultimate obligation for patient care remains with the medical professionals.