Stanford Center for Artificial Intelligence in Medicine and Imaging (AIMI)’s Post

Our systematic review from 2022 is still one of the top 10 most read papers in BMJ Open about AI in healthcare!

A thought-provoking piece on the promise and potential of AI in the life sciences and healthcare from California Life Sciences (CLS). California is a leader in innovation and a leading voice in the integration of AI across industries. We're already seeing how AI is changing the game in the life sciences and healthcare in areas like drug discovery, medical research, minimizing bias in clinical trials and more. But with the vast promise of AI comes a critical need to responsibly deploy this technology with the appropriate guardrails in a way that doesn't impede potential advancements.

Love seeing the ongoing conversation around AI in healthcare gaining traction. AI has the potential to change the way we diagnose and treat patients by providing quick and accurate insights❗ However, it's essential to remember that AI should complement, not replace, human expertise. The ability of AI to explain its decision-making process is crucial for building trust and ensuring patient safety. As AI continues to evolve, its role as a supportive tool for clinicians will be key to improving outcomes and enhancing patient care 📈 #HealthcareAI #MedicalInnovation #DigitalHealth

What is the impact of AI models on clinician diagnostic accuracy? A recent study in the JAMA, Journal of the American Medical Association, titled "𝐌𝐞𝐚𝐬𝐮𝐫𝐢𝐧𝐠 𝐭𝐡𝐞 𝐈𝐦𝐩𝐚𝐜𝐭 𝐨𝐟 𝐀𝐈 𝐢𝐧 𝐭𝐡𝐞 𝐃𝐢𝐚𝐠𝐧𝐨𝐬𝐢𝐬 𝐨𝐟 𝐇𝐨𝐬𝐩𝐢𝐭𝐚𝐥𝐢𝐳𝐞𝐝 𝐏𝐚𝐭𝐢𝐞𝐧𝐭𝐬", sheds light on the matter and reveals intriguing findings. 🔎 📈 Clinicians experienced a notable 4.4% increase in diagnostic accuracy when powered by standard AI model predictions...   📉 ...but they also faced an 11.3% drop in accuracy when encountering systematically biased AI predictions.    While #AI is surely enhancing #healthcare, addressing systematic biases is crucial to maintain patient care standards. 👉 𝐖𝐡𝐚𝐭 𝐢𝐬 𝐭𝐡𝐞 𝐰𝐚𝐲 𝐟𝐨𝐫𝐰𝐚𝐫𝐝 𝐭𝐨 𝐢𝐦𝐩𝐫𝐨𝐯𝐞 𝐫𝐞𝐬𝐮𝐥𝐭𝐬? Read more in the article by Sara Scarpinati, Marketing & Communications Manager at EVERSANA: https://lnkd.in/dQEgMHh3 

These are sound principles for Medical AI. I applied them to my PhD thesis on AI for growth diagnosis 30 years ago, TrenDx. Lets keep the foundations of rigorous AI testing.

Bridging the Gap: AI in Healthcare from Research to Reality A thought-provoking NEJM AI paper highlights a crucial issue: while AI research in healthcare is abundant, real-world evidence remains scarce. Key needs identified: • Comprehensive clinical effectiveness evaluations • Focus on real-world impact, not just model validation • Consideration of local healthcare contexts • Adoption of implementation science principles I'm excited to be part of a project addressing these very challenges in emergency care. We're focusing on translating AI innovations into practical improvements in ED efficiency and patient outcomes. Looking forward to sharing more as our work progresses. How do you see AI reshaping emergency medicine? #HealthcareAI #EmergencyMedicine #ClinicalEffectiveness https://lnkd.in/dBjYBN_x

AI promises to revolutionize healthcare by enhancing diagnosis, treatment, and patient safety. Yet, a critical gap exists between AI research and its real-world impact. We urgently need comprehensive clinical effectiveness evaluations to bridge this divide. My colleagues at UC San Diego Health, UC Davis Health, and I published this article in the NEJM AI to establish boundaries for using AI in the healthcare setting, bringing together research and real-world evidence for comprehensive clinical effectiveness. AI is bound to revolutionize our industry that was traditionally adverse to change, yet it is essential to collaborate on the shared goal of safer, more effective, and equitable care for all patients. Christopher Longhurst Karandeep Singh Aneesh Chopra Ashish Atreja, MD, MPH Our call to action: (1) Move Beyond Model Validation: It's essential to validate AI models not just through simulations but in real-world clinical settings, akin to clinical trials for new drugs. (2) Focus on Local Context: The importance of local health care contexts in AI model validation cannot be overstated. Different settings require tailored approaches for optimal outcomes. (3) Implement Science Principles: Adopting implementation science principles will ensure that AI tools are effectively integrated into health care workflows, resulting in tangible improvements in patient outcomes. A robust network of healthcare delivery organizations is crucial to focus on the clinical effectiveness of AI models in real-world settings. This collaborative approach will help us achieve safer, more effective, and equitable care for all patients. #HealthcareInnovation #AIinHealthcare #ClinicalEffectiveness #PatientSafety #ImplementationScience #AIValidation #EquitableCare #HealthTech

Improving communication between patients and physicians is paramount for enhancing patient involvement in their own care and fostering confidence in the care provided. The integration of generative AI in healthcare communication, as demonstrated by the research led by NYU Langone Health, holds significant implications for the future of patient-physician interaction and overall healthcare delivery. The study's findings indicate that AI-driven translation of patient discharge summaries into a more readable format achieved notable improvements in readability and understandability compared to the original summaries. This enhancement could potentially empower patients to actively engage in their care decisions and follow-up recommendations more effectively. While the widespread use of generative AI in healthcare communication may not be immediately feasible due to current limitations, such as accuracy issues, the research highlights the potential of AI as a valuable tool in enhancing healthcare communication. With further advancements and improvements in technology and oversight, AI-driven solutions could play a significant role in improving patient outcomes and healthcare delivery in the future. #GenerativeAI #FutureOfHealth #HealthAI

This article highlights the formation of Artificial Intelligence Implementation Science Centers as pivotal in evaluating AI tools' clinical effectiveness in healthcare. By focusing on real-world applications, these centers aim to ensure that AI advancements truly enhance patient care and safety, bridging the gap between technology and healthcare efficacy. #AIHealthcare #ClinicalEffectiveness #HealthTech

One of the areas where new #AI tools are doing the most is in the world of #medicine. Partly that’s because this is a #datadriven enterprise – big data has the capability to help heal disease, discover new drugs, and even work on the human genome itself. So this kind of work started prior to the AI revolution, but AI supercharges what scientists are able to do. Read about the intersection of #genAI and #healthcare: https://lnkd.in/g3fpZ3SZ