Did you know that acute pulmonary embolism remains one of the leading causes of cardiovascular mortality? Millions of patients present to the emergency department with severe chest pain and shortness of breath every year. If your clinical team is struggling with delayed imaging interpretations, you are not alone. Diagnostic bottlenecks are a very common type of clinical delay that hospitalists and pulmonologists experience on a daily basis. However, new technology is offering an all-rounded solution. Viz.ai has introduced a specialized AI-powered pulmonary care platform focused strictly on optimizing acute clinical workflows. This system is designed to accelerate the detection and triage of acute pulmonary conditions by analyzing complex imaging data in real-time. It builds upon their established track record of FDA-cleared algorithms that facilitate rapid cross-disciplinary communication. In this blog post, we will discuss the clinical evidence behind artificial intelligence in pulmonary care, how it impacts risk stratification, and what limitations you need to consider before implementing it in your facility.
What is an AI-powered pulmonary care platform?
When a patient undergoes a CT pulmonary angiogram in the emergency department, the images must be routed to a server, placed in a queue, and eventually read by an attending radiologist. This standard process takes valuable time. An AI-powered pulmonary care platform actively monitors the hospital imaging network to identify suspected pulmonary emboli the moment the scan is completed. By analyzing the complex imaging data in real-time, the algorithm flags positive cases instantly. It then sends immediate alerts directly to the mobile devices of the pulmonary embolism response team. Yes, the entire cross-disciplinary team gets the notification at the exact same time.
This simultaneous alerting is essential to reducing the time from diagnosis to life-saving intervention. Research by de Jong CMM et al, *Thrombosis research* 2024 confirms that modern imaging of acute pulmonary embolism benefits heavily from automated detection systems. These algorithms do not replace the radiologist. They simply pull high-risk scans to the top of the reading queue. They ensure the interventional team is aware of a potential mechanical thrombectomy candidate minutes after the scan is performed rather than hours later. You can rest assured that this technology integrates smoothly with existing hospital infrastructure. The software typically lives on a secure server and communicates seamlessly with the imaging modalities.
How the notification system changes behavior
The traditional method involves a radiologist making a phone call to the emergency physician, who then pages the pulmonologist, who then contacts the interventional radiologist. This chain of communication is prone to significant delays. However, a mobile triage platform puts the CT images in the hands of all specialists simultaneously. The interventionalist can review the clot burden on their phone while walking to the emergency department. This completely changes how quickly a treatment decision is made.
How is acute pulmonary embolism diagnosed using artificial intelligence?
The traditional diagnosis of a pulmonary clot relies on clinical gestalt, scoring systems such as the Wells criteria, and D-dimer testing followed by computed tomography. However, adding machine learning to this established workflow changes the speed and accuracy of confirmation. Douillet D et al, *Seminars in respiratory and critical care medicine* 2021 evaluated suspected acute pulmonary embolism and found that integrating artificial intelligence with standard scoring systems can drastically improve diagnostic speed.
There are many different diagnostic algorithms currently in development and clinical use. Some of them focus entirely on image pattern recognition within the contrast-filled pulmonary arteries. Others incorporate electronic health record data to provide a pre-test probability score before the scan is even ordered. Puchades R et al, *Thrombosis research* 2024 reviewed various machine learning approaches and their performance evaluation. They noted that AI algorithms consistently demonstrate high sensitivity for detecting filling defects in the pulmonary vasculature. This high sensitivity is essential for preventing missed diagnoses in busy emergency departments.
The algorithmic lung detective
Finding large saddle emboli is relatively easy for any trained physician. However, small peripheral clots are notoriously difficult to spot. Allena N et al, *Cureus* 2023 described artificial intelligence as the algorithmic lung detective in this context. Their research emphasizes how AI can recognize tiny subsegmental emboli that are easily overlooked during a rapid imaging read. When a patient is scanned at three in the morning, the fatigue of the on-call radiologist is a real clinical risk factor. The AI acts as an untiring second pair of eyes. It ensures that even small peripheral clots are flagged for review and appropriate management.
Is artificial intelligence effective for risk stratification?
Identifying a clot is only the first step of the clinical pathway. You also need to know if the patient is stable, intermediate-risk, or high-risk. Henkin S et al, *The American journal of cardiology* 2025 reviewed the expanding toolkit for risk stratification. They noted that artificial intelligence can automatically calculate the right ventricle to left ventricle diameter ratio directly from the CT scan.
This automated right heart strain measurement is an essential indicator of intermediate-high risk pulmonary embolism. Usually, a radiologist has to manually measure these ventricles on an axial slice, which adds time and variability to the report. The AI system provides an exact ratio instantly.
Besides this, AI can assist in calculating the total clot burden within the pulmonary arterial tree. Naser AM et al, *Diagnostics (Basel, Switzerland)* 2025 reviewed the role of artificial intelligence in the diagnosis and management of pulmonary embolism. They highlighted that this quantitative analysis gives the clinician a much clearer picture of the disease severity compared to subjective visual estimates. Knowing the exact clot volume helps the team decide whether aggressive intervention is necessary.
Role of artificial intelligence in analyzing electrocardiograms
Risk stratification is not limited to CT imaging alone. The algorithm can combine the imaging findings with other data sources to build a complete profile. Ose B et al, *Current cardiology reports* 2024 published a review on artificial intelligence interpretation of the electrocardiogram. They found that machine learning can detect subtle signs of right heart strain on a standard 12-lead ECG that a human reader might miss entirely.
Changes such as the classic S1Q3T3 pattern are actually quite rare in clinical practice. However, AI can detect minute shifts in the T-wave morphology and axis that correlate with right ventricular overload. If your team can integrate both CT analysis and ECG interpretation into one automated workflow, you will have a highly effective method for identifying patients who need immediate escalation of care. This multi-modality approach is the future of acute cardiovascular triage.
What are the limitations and false positives of AI detection?
Although it may seem like AI is the perfect diagnostic tool, there are real-world limitations you must consider. Artificial intelligence algorithms are trained on specific datasets, and their performance can drop when applied to a new hospital population with different imaging protocols. Silva LOD et al, *PloS one* 2024 validated an AI-based pulmonary embolism classification system using real-world data and found that false positive rates remain a significant clinical challenge.
Motion artifact, poor contrast timing, and chronic pulmonary hypertension can all trick the algorithm into flagging a scan as positive for acute embolism. In some cases, the AI might highlight a respiratory motion artifact as a clot. This leads to unnecessary alerts and alarm fatigue among the pulmonary response team. When the phones ring constantly for false positives, physicians eventually start ignoring the alerts altogether.
What’s more, AI does not always beat a trained human eye. Li Y et al, *Frontiers in medicine* 2025 highlighted the research progress of machine learning in pulmonary embolism, pointing out that AI does not always outperform expert clinical gestalt. A seasoned radiologist will easily dismiss a streak artifact from the superior vena cava that an algorithm might flag with extremely high confidence. Therefore, these tools must be used as a triage assistant rather than a definitive diagnostic authority. Your clinical judgment is still the most essential component of patient care.
How does rapid triage affect interventional treatments?
When the AI system alerts the team of an intermediate-high risk pulmonary embolism, the interventional radiologist or cardiologist can review the images on their phone instantly. This rapid communication directly impacts the time to reperfusion therapy. The faster you can get the patient to the catheterization lab, the better their chances of surviving without long-term right heart failure.
Mechanical thrombectomy outcomes
Jaber WA et al, *Circulation* 2025 published the primary results of the PEERLESS randomized controlled trial. They compared large-bore mechanical thrombectomy against catheter-directed thrombolysis in the management of intermediate-risk pulmonary embolism. The study showed that mechanical clot extraction provides rapid hemodynamic improvement. However, these advanced therapies require a coordinated team response. If you want to deploy mechanical thrombectomy effectively, you need a system that alerts the interventionalist before the patient decompensates hemodynamically.
Managing bleeding risks
One of the primary reasons to favor mechanical thrombectomy over systemic or catheter-directed thrombolysis is the bleeding risk. If your patient has a history of bleeding complications, giving them thrombolytics can be incredibly dangerous. Murthy SB et al, *Stroke* 2025 evaluated outcomes following minimally invasive surgery for intracerebral hemorrhage. While their study focused on stroke registries, it serves as a stark reminder of the devastating consequences of intracranial bleeding. When an AI system rapidly identifies a pulmonary embolism, it gives the team the necessary time to carefully review the patient’s contraindications and opt for a mechanical intervention rather than risking a catastrophic brain bleed with thrombolytic drugs.
Vacuum thrombectomy and right ventricular recovery
Similarly, Moriarty JM et al, *Journal of the American Heart Association* 2025 presented an interim analysis of 300 patients from the STRIKE-PE study. They looked at the periprocedural results and right ventricular outcomes of computer-assisted vacuum thrombectomy. The data indicated substantial improvements in right heart strain following the procedure. An AI triage platform ensures that these patients are identified and evaluated for vacuum thrombectomy as quickly as possible. This seamless pipeline from diagnosis to intervention is exactly what these new software platforms aim to provide.
Role of artificial intelligence in managing chronic pulmonary embolism
Most AI platforms focus heavily on the acute presentation, but chronic conditions also benefit from machine learning. Chronic thromboembolic pulmonary hypertension can be incredibly difficult to diagnose because the imaging findings are often subtle and easily missed on a routine scan. Abdulaal L et al, *Frontiers in radiology* 2024 conducted a systematic review of artificial intelligence tools for chronic pulmonary embolism on CT pulmonary angiography.
They found that AI can successfully detect webs, bands, and subtle vessel narrowing that indicate chronic fibrotic disease. If a patient presents with unexplained shortness of breath and an AI system flags chronic changes on a CT scan, your team can refer them for specialized pulmonary hypertension evaluation much earlier in the disease process. Early referral for pulmonary endarterectomy or balloon pulmonary angioplasty can significantly improve long-term survival for these complex patients.
New diagnostic tools and healthcare costs
The technology surrounding pulmonary embolism care is changing rapidly. Shapiro J et al, *Methodist DeBakey cardiovascular journal* 2024 outlined several new diagnostic tools for pulmonary embolism detection. They emphasized that advanced software is pushing the boundaries of what we can see on standard imaging. What’s more, these technologies have significant financial implications for the hospital system as a whole.
Mohr K et al, *European heart journal. Acute cardiovascular care* 2024 modelled the costs of interventional pulmonary embolism treatment. They analyzed the implications of United States trends for a European healthcare system. While advanced algorithms and mechanical thrombectomy devices have a high upfront cost, they can drastically reduce the length of stay in the intensive care unit. Getting a patient out of the intensive care unit one or two days earlier will easily offset the initial software licensing fees. Hospitals must look at the total episode of care costs rather than just the price of the AI software itself.
Conclusion
Undoubtedly, acute pulmonary embolism is a highly dangerous condition that requires immediate and coordinated medical attention. The longer a patient waits for a definitive diagnosis, the higher their risk of right heart failure and death. The new Viz.ai platform and similar technologies provide an essential bridge between the radiology suite and the interventional team. By analyzing scans in real-time, they ensure that high-risk patients are not lost in the busy workflow of a modern emergency department.
There is no specific cure-all for the complex logistics of hospital medicine, but these tools can definitely be managed to optimize your resources. If you integrate an AI triage system properly and train your response team to handle the alerts, you can rest assured that your patients will receive faster, more effective care for their acute pulmonary conditions.
References
- Puchades R et al. Artificial intelligence for predicting pulmonary embolism: A review of machine learning approaches and performance evaluation. Thrombosis research 2024. doi:10.1016/j.thromres.2023.12.002 (PMID: 38113607)
- de Jong CMM et al. Modern imaging of acute pulmonary embolism. Thrombosis research 2024. doi:10.1016/j.thromres.2024.04.016 (PMID: 38703584)
- Douillet D et al. Suspected Acute Pulmonary Embolism: Gestalt, Scoring Systems, and Artificial Intelligence. Seminars in respiratory and critical care medicine 2021. doi:10.1055/s-0041-1723936 (PMID: 33592653)
- Ose B et al. Artificial Intelligence Interpretation of the Electrocardiogram: A State-of-the-Art Review. Current cardiology reports 2024. doi:10.1007/s11886-024-02062-1 (PMID: 38753291)
- Li Y et al. Research progress of artificial intelligence and machine learning in pulmonary embolism. Frontiers in medicine 2025. doi:10.3389/fmed.2025.1577559 (PMID: 40212275)
- Henkin S et al. Artificial Intelligence for Risk Stratification of Acute Pulmonary Embolism: Perspectives on Clinical Needs, Expanding Toolkit, and Pathways Forward. The American journal of cardiology 2025. doi:10.1016/j.amjcard.2025.05.025 (PMID: 40436309)
- Naser AM et al. Role of Artificial Intelligence in the Diagnosis and Management of Pulmonary Embolism: A Comprehensive Review. Diagnostics (Basel, Switzerland) 2025. doi:10.3390/diagnostics15070889 (PMID: 40218239)
- Silva LOD et al. Artificial intelligence-based pulmonary embolism classification: Development and validation using real-world data. PloS one 2024. doi:10.1371/journal.pone.0305839 (PMID: 39167612)
- Allena N et al. The Algorithmic Lung Detective: Artificial Intelligence in the Diagnosis of Pulmonary Embolism. Cureus 2023. doi:10.7759/cureus.51006 (PMID: 38259362)
- Abdulaal L et al. A systematic review of artificial intelligence tools for chronic pulmonary embolism on CT pulmonary angiography. Frontiers in radiology 2024. doi:10.3389/fradi.2024.1335349 (PMID: 38654762)
- Jaber WA et al. Large-Bore Mechanical Thrombectomy Versus Catheter-Directed Thrombolysis in the Management of Intermediate-Risk Pulmonary Embolism: Primary Results of the PEERLESS Randomized Controlled Trial. Circulation 2025. doi:10.1161/CIRCULATIONAHA.124.072364 (PMID: 39470698)
- Shapiro J et al. New Diagnostic Tools for Pulmonary Embolism Detection. Methodist DeBakey cardiovascular journal 2024. doi:10.14797/mdcvj.1342 (PMID: 38765212)
- Murthy SB et al. Outcomes Following Minimally Invasive Surgery for Intracerebral Hemorrhage in the AHA Get With The Guidelines-Stroke Registry. Stroke 2025. doi:10.1161/STROKEAHA.124.048650 (PMID: 40177744)
- Moriarty JM et al. Periprocedural Results and Right Ventricular Outcomes of Computer Assisted Vacuum Thrombectomy Treatment of Acute Pulmonary Embolism: Interim Analysis of 300 Patients From the STRIKE-PE Study. Journal of the American Heart Association 2025. doi:10.1161/JAHA.124.039975 (PMID: 40878986)
- Mohr K et al. Modelling costs of interventional pulmonary embolism treatment: implications of US trends for a European healthcare system. European heart journal. Acute cardiovascular care 2024. doi:10.1093/ehjacc/zuae019 (PMID: 38349225)
- https://www.fiercehealthcare.com/ai-and-machine-learning/vizai-launches-ai-powered-pulmonary-care-platform
Licensed physician and clinical AI specialist. Founder and Editor-in-Chief of ZayedMD, a physician-led medical publication covering clinical AI, neurology, metabolic health, and evidence-based patient guidance.
