Jan Beger’s Post

View profile for Jan Beger, graphic

Global Head of AI Advocacy @ GE HealthCare

This paper discusses the current and future state of #AI interpretation of medical images. It emphasizes the potential of #ArtificialIntelligence models to interpret a wide range of findings and generate full radiologic reports. It highlights the need for better generalization checks, clinician-AI collaboration, transparency, and post-deployment monitoring. The paper also mentions the development of new foundation models that could lead to broader adoption of AI in #Healthcare. 1️⃣ The future of AI in medical imaging is promising, with the development of generalist medical AI models that can tackle the entire task of radiologic image interpretation and more. These models could accurately generate full radiologic reports by interpreting a wide range of findings with degrees of uncertainty and specificity based on the image. 2️⃣ Clinician-AI collaboration is crucial for the successful use of AI in #Radiology. AI assistance in interpreting medical images has been found to be more beneficial to less experienced clinicians. However, the decision-making processes of many AI methods are not easily interpretable by humans, posing challenges for clinicians trying to understand and trust AI recommendations. 3️⃣ Transparency is a major challenge in evaluating the generalization behavior of AI algorithms in medical imaging. Many commercially available AI products lack scientific, peer-reviewed evidence of efficacy. There is a need for more rigorous transparency measures, such as the use of checklists and public release of medical-imaging datasets. 4️⃣ Post-deployment monitoring is essential for AI systems in radiology. This involves mechanisms to incorporate feedback from clinicians and continual learning strategies for regular updating of the models. This paper provides valuable insights into the future of AI in medical imaging, highlighting the potential of AI models to revolutionize the field. It also underscores the importance of clinician-AI collaboration, transparency, and post-deployment monitoring, which are crucial for the successful adoption and integration of AI in healthcare. ✍🏻 Pranav Rajpurkar, Matthew Lungren MD MPH. The Current and Future State of AI Interpretation of Medical Images. N Engl J Med. 2023 May 25;388(21):1981-1990. doi: 10.1056/NEJMra2301725. PMID: 37224199. 💡 How do you see the role of AI in your current medical practice and what steps can you take to further integrate AI into your workflow for improved patient care?

  • No alternative text description for this image
Dr KAYODE OGUNLEYE

I AM A PASSIONATE NIGERIAN HEALTHCARE PROFESSIONAL & ENTREPRENEUR, DREAMING OF AN AFRICAN UTOPIA WHERE HEALTHCARE IS ACCESSIBLE TO ALL.

1y

Microsoft has coined the term "CO-PILOT" to describe the role of AI in healthcare delivery, emphasizing its capacity to enhance and support human professionals rather than supplanting them. This concept underscores the approach of leveraging AI as a collaborative tool that works alongside healthcare providers, serving as a valuable partner to elevate patient care and outcomes. By harnessing the capabilities of AI technologies, healthcare professionals can achieve heightened efficiency, improved decision-making, and enhanced patient experiences, all while upholding the indispensable human element that is essential in healthcare delivery.

Marcel Boller

Pharma and Medtech Expert | Counsel at Wenger Vieli Ltd

1y

In my opinion, this paper effectively outlines the immense potential of AI in medical imaging and the need for robust clinician-AI collaboration, transparency measures, and post-deployment monitoring. It sheds light on the challenges and opportunities in this domain, paving the way for the successful integration of AI into healthcare and improving patient outcomes.

Issmaeel Lawendy

Founder and Chief Executive Officer | Health Care Access, Confidence Building

1y

Jan Beger great read!  We need to allow AI to play a role in the health care sector but with oversight as AI can reveal findings that clinicians may miss.  However, I would say that as great as AI is, it still needs to be supervised by an expert to make sure that the findings are correct and not generating false positives. When human lives are at stake, it's imperative to have human oversight before these technologies are implemented and depended on.

AI has come to disrupt the Health sector for good.

Like
Reply

At TMC we have a team dedicated to identifying, deploying and monitoring AI solutions to provide the best care for our patients always under the supervision of our clinicians. 

Coordonateur des communications RSADC

Chercheurs en intelligence artificielle appliquée aux enfants gravement malades chez CHU Sainte-Justine

1y

If you like this subject I strongly suggests that you follow our brand new page Regroupement en IA appliquée en soins aigus de l'enfant. It is a bilingual page (hence the french name, but publications will be posted in english as well).We will publish regularly posts, conferences and scientific articles on the topic of AI in the pediatric ICU setting. We are a group of researches from Montreal mostly based in the Sainte-Justine's hospital PICU.

Like
Reply

Jan Beger Great paper discussing the potential of AI interpretation of medical images & emphasising the importance of generalization checks, clinician-AI collaboration, transparency, and post-deployment monitoring. These principles align with the mindset and passion of OmniversalisDAO, which aims to leverage cutting-edge technology, including AI, to revolutionize healthcare. We recognize the promise of AI in medical imaging and emphasizes collaboration, transparency, and continuous improvement for the successful integration of AI into healthcare workflows, ultimately leading to improved patient care.

Like
Reply
Roman V. Dvorak, MD, PhD

Serial chief medical officer and healthspan enthusiast

1y

I see many companies pushing their own "latest & greatest" product in this space, given that the quality of the results is directly dependent on the number of images the system can learn from, wouldn't it make most sense to develop a few truly big ones, which would use millions of images for learning, be extensively validated and rigorously tested and then deployed globally?

Like
Reply

This compelling article captures the transformative potential of AI in the realm of medical imaging, a vision that we at Memori.ai wholeheartedly share and strive towards. We firmly believe in the extraordinary potential of AI and medicine working in unison, especially in the field of radiology, where advanced models could provide comprehensive radiologic reports, enhancing diagnostic precision and accelerating patient care. The focus on clinician-AI collaboration aligns perfectly with our ethos at Memori.ai. We champion the view of AI as an augmentative tool, complementing rather than replacing human expertise. AI has the power to aid interpretation, but the clinician's knowledge and contextual understanding of the patient remains indispensable. We also concur that transparency and post-deployment monitoring are essential components in ensuring the reliability and safety of AI systems. As we progress in this field, we must balance technological advancement with rigorous standards of validation and accountability. Thank you for sharing this enlightening paper. It serves as a timely reminder of the revolutionary potential of AI in healthcare, and the responsibilities we bear as we chart this exciting path forward.

See more comments

To view or add a comment, sign in

Explore topics