What is PE Risk Stratification?
Pulmonary embolism (PE) is a life-threatening condition where a blood clot blocks the pulmonary arteries, often due to deep vein thrombosis in the lower limbs. PE is the third leading cause of cardiovascular death, following heart attacks and strokes. Timely identification and categorization–or “stratification”–of patients by risk is essential for effective PE management, helping to prioritize diagnostic and therapeutic responses to prevent mortality.
Risk stratification in PE typically involves two key steps:
- Identification of circulatory failure: Patients with low blood pressure and other signs of circulatory instability face a high risk of immediate death and require urgent intervention.
- Estimation of 30-day mortality: For those without severe instability, the pulmonary embolism severity index (PESI score) or other scoring systems are used to assess 30-day mortality risks and guide treatment options.
PE patients are generally grouped into four categories based on their risk levels:
- High-Risk: Obstruction of the pulmonary arterial tree exceeds 50% of a cross-sectional area, causing acute and severe cardiopulmonary failure from right ventricular overload. These cases require immediate specialist intervention and warrant immediate Pulmonary Embolism Response Team (PERT) consultation.
- Intermediate-High Risk: Defined by the presence of both RV dysfunction (RVD) and elevated biomarkers (troponin or bnp) without hemodynamic instability. These patients are typically classified as PESI Class III-IV or have a simplified PESI (sPESI) score of ≥1. With increased experience in PE treatment, tailored care pathways for intermediate-high risk patients have evolved, offering multiple management options depending on patient stability and risk factors.
- Intermediate-Low Risk: Characterized by either RVD or elevated biomarkers but not both, and without hemodynamic instability. These patients also generally fall into the PESI Class III-IV or have an sPESI score of ≥1. Treatment often includes anticoagulants, with PERT consultations as needed, especially when the clinical presentation is ambiguous.
- Low-Risk: A PE that is stable with normal RV function. Low-risk PEs are less likely to require a PERT consultation, but often require anticoagulant therapy.
PE Score Criteria
Validated clinical scoring systems help clinicians estimate PE risk and determine the best diagnostic and treatment approaches. Some commonly used PE risk scores include:
- Modified Wells Scoring System: Uses criteria like clinical signs of DBT and tachycardia to assign risk points.
- Revised Geneva Scoring System: Based on a combination of clinical history and physical exam findings.
- Pulmonary Embolism Rule-Out Criteria (PERC): Helps exclude PE in low-risk patients.
Imaging Methods Used for PE Risk Stratification
Imaging plays a crucial role in PE risk stratification, particularly for intermediate-risk patients. Imaging techniques like echocardiography and CT pulmonary angiography provide insight into right ventricular function, while newer serologic biomarkers such as plasma lactate, troponin and bnp levels offer additional prognostic information. Identifying right ventricular dysfunction and myocardial injury helps predict 30-day mortality risk and indicates when aggressive treatments, like thrombolysis or mechanical thrombectomy, may be necessary.
How AI Can Be Used in Conjunction With Imaging for PE Risk Stratification
AI advancements in PE diagnostics have significantly improved the timeliness and accuracy of care. For instance, Sheba Medical Center, Israel’s largest medical center, used an AI algorithm capable of identifying high-risk PE patients upon hospital admission. This machine-learning model, trained on extensive medical records, predicted PE risk with notable accuracy, enabling early diagnosis and faster intervention.
Prof. Gad Segal, who led the study, highlighted the algorithm’s potential for transforming PE care. Despite only using arrival data, the AI model accurately flagged high-risk patients, improving survival rates by expediting care. These tools aim to further integrate with diagnostic AI models to create comprehensive risk assessment systems, elevating PE treatment to new standards.
Additionally, real-world experiences underscore AI’s value in PE intervention. Gustavo Meirelles, MD, shared a case where Aidoc’s AI flagged an incidental PE in an oncologic patient during a routine scan, accelerating her treatment from days to minutes and potentially saving her life.
How Aidoc Helps With PE Care
AI-based cardiovascular solutions like Aidoc offer profound benefits to patients and healthcare systems:
- Shortened ICU Stays: Aidoc’s AI implementation reduced ICU stays by 55.4%, from 80.2 hours to 35.8 hours1, demonstrating substantial improvements in patient management in the network.
- Cost Savings: A recent study showed an average cost savings of $10,500 per PE patient, amounting to over $500,000 annually for hospitals like Cedars-Sinai performing multiple thrombectomy procedures each year.2
Nate Mizraki, MD, from Cedars-Sinai emphasizes that AI isn’t a substitute for physicians, but rather a tool enhancing their efficiency. The benefits of AI allow doctors to remain focused on high-risk cases, leveraging AI’s predictive power to improve outcomes across emergency medicine.
The Future of AI in PE Risk Stratification
PE risk stratification is vital for reducing PE-related deaths and improving patient care. By combining clinical judgment, established scoring systems and innovative AI tools, healthcare providers can streamline PE treatment and improve prognosis for high-risk patients. AI-driven models are already reshaping PE care by improving early diagnosis, cutting ICU times and reducing healthcare costs, emphasizing their transformative potential across emergency and cardiovascular medicine.
PE risk stratification and AI together are opening doors to more effective, lifesaving interventions–pointing toward a future where data-driven tools can help standardize and enhance PE care worldwide.
Citations-
1. Burch, C. “Improving Patient Outcomes with an AI-Enhanced Pulmonary Embolism Response Team in a Large Healthcare Network” Presented at the 10th Annual Pulmonary Embolism Symposium 2024
2. Mizraki, N. “Cost-Effectiveness Analysis of AI-Driven Pulmonary Embolism Response Team Activation in Mechanical Thrombectomy” Presented at the 10th Annual Pulmonary Embolism Symposium 2024
