AI for Patient Risk Stratification

A groundbreaking study from Johns Hopkins Medicine, published in Nature Cardiovascular Research on July 2, shows how #AI can dramatically improve risk prediction for sudden cardiac death (SCD) in patients with hypertrophic cardiomyopathy (HCM) a leading cause of cardiac arrest in young people and athletes. The research team, led by Changxin Lai PhD, and senior author Natalia Trayanova PhD, developed a deep learning model called MAARS (Multimodal AI for Arrhythmia Risk Stratification). By integrating data from cardiac MRI, echocardiograms, and electronic health records, MAARS revealed critical, previously inaccessible insights into heart health. Today’s clinical guidelines can only identify high-risk HCM patients with around 50% accuracy barely better than a coin toss. This uncertainty leads to tragic outcomes: some patients suffer preventable cardiac arrests, while others undergo unnecessary implantable defibrillator surgeries. MAARS significantly outperformed current risk assessment methods: 89% accuracy across all patients 93% accuracy in adults aged 40–60, a high-risk group for HCM The study involved 837 patients from Johns Hopkins Hospital and the Sanger Heart & Vascular Institute. All were evaluated using both traditional guidelines and MAARS. Across demographics, the AI model showed robust performance and clear clinical potential. “This could save lives - and spare others from living with devices they don’t need,” said Trayanova. “We now have the ability to predict, with high accuracy, who is truly at risk.” Beyond HCM, the research team plans to extend MAARS to other heart conditions like cardiac sarcoidosis and arrhythmogenic right ventricular cardiomyopathy. This work highlights how multimodal AI is moving us closer to truly personalized medicine transforming raw data into life-saving insight. https://lnkd.in/eGCatbvw #cardiology #healthcare #cardiomyopathy

The headline that caught my eye this week was "AI Trial to Spot Heart Condition Before Symptoms." Here's my take: Artificial intelligence holds substantial promise to improve quality and reduce costs in healthcare. One example from Leeds involves an algorithm that scours medical records for early warning signs of atrial fibrillation (AF) before symptoms appear — potentially preventing thousands of strokes. The results suggest that by analyzing existing medical records for patterns that human physicians might miss, AI can flag high-risk patients for early intervention. The trial has already identified cases like a 74-year-old former Army captain who had no symptoms but can now manage his condition effectively. This is particularly significant given that AF contributes to around 20,000 strokes annually in the UK alone. As Professor Chris Gale notes, too often the first sign of undiagnosed AF is a stroke — an outcome this technology could help prevent. The broader implication here is about AI's role in healthcare: not replacing physicians but augmenting their ability to identify risks earlier and intervene before conditions become critical.  

🚀 AI in Medicine: A Lifeline or a Line of Controversy? 🏥 The NHS is using AI-driven patient prioritisation, tackling the 7.5 million-person waiting list. In a nine-hospital pilot, AI helped prevent post-op chest infections, halved complication rates, and reduced hospital stays by 4+ days. https://lnkd.in/dkuiqjHQ 🤖 How? AI trained on 200 million records analyses patient risk—factoring in age, vitals, and even postcode. The result? More urgent cases get prioritised, optimising care and saving lives. ⚖️ The Ethical Dilemma: 🔹 Could AI unintentionally widen health inequalities? Studies show some AI systems favoured certain patients over others, based on the existing training datasets. 🔹 Should black box scoring, using AI, help us made decision on economic productivity—prioritising a 55-year-old sole breadwinner over a 75-year-old retiree needing a hip replacement? 🇬🇧 The UK’s Opportunity The UK is uniquely positioned to lead ethical, evidence-driven AI healthcare policies. If we get this right, we don’t just build faster systems—we build fairer ones. 💡 Your thoughts? Should AI play a bigger role in who gets treated first? 🩺 AI is evolving healthcare—redefining hospitals, empowering doctors, and transforming patient care. Let’s strive for excellence in the future of medicine.

🚨 New Publication Alert! 🚨 Excited to share our latest work published in Nature Portfolio journal npj Precision Oncology: "Computationally integrating radiology and pathology image features for predicting treatment benefit and outcome in lung cancer" In this study, we tackled one of the biggest challenges in lung cancer care—the lack of robust biomarkers for guiding treatment decisions. By combining radiomics from CT scans and pathomics from H&E slides, we developed integrated AI models that significantly improved risk stratification and prediction of treatment benefit in: 🔹 Early-stage NSCLC – where our integrated model predicted disease recurrence with a hazard ratio of 8.35 (C-index: 0.71) 🔹 Advanced-stage NSCLC – achieving improved immunotherapy response prediction (AUC: 0.75) 🔹 Small Cell Lung Cancer (SCLC) – outperforming individual models in predicting chemotherapy response (AUC: 0.78) This work underscores the power of computationally fusing imaging data across scales—from radiology to pathology—to advance precision oncology. Grateful to have collaborated with a fantastic multidisciplinary team: Pranjal Vaidya, Mohammadhadi Khorrami, Kaustav Bera, Pingfu Fu, Lukas Delasos, Amit Gupta, Cristian Barrera, Nathan Pennell MD, PhD, FASCO, Vamsi Velcheti MD MBA FASCO Read the full paper here: https://rdcu.be/eppuU Our commercial partner Picture Health is working to translate these multimodal #AI tools to the clinic. #AIinHealthcare #LungCancer #PrecisionOncology #Radiomics #Pathomics #AnantMadabhushi #EmoryAI4Health #npjPrecisionOncology #ComputationalPathology #MultiscaleImaging #CancerResearch

AI could help identify high-risk heart patients Artificial intelligence is stepping up in healthcare, particularly in identifying patients at risk of serious heart conditions. A team at the University of Leeds has developed an AI system called Optimise, which examined the health records of over two million individuals to find those most vulnerable to conditions like heart failure, stroke, and diabetes. The findings were significant: more than 400,000 people were identified as high-risk, representing a staggering 74% of patients who later died from heart-related issues. In a pilot study of 82 high-risk patients, Optimise revealed that 20% had undiagnosed moderate to high-risk chronic kidney disease. Additionally, over half of those with high blood pressure were prescribed different medication to better manage their heart risk. Dr. Ramesh Nadarajah from the University of Leeds highlighted the potential of AI in offering timely care, which is often more cost-effective than treating advanced conditions. This proactive approach not only benefits patients but also helps ease the burden on healthcare systems like the NHS. The research team's next step is to conduct a larger clinical trial to assess the full impact of doctor-led care supported by AI insights. The British Heart Foundation's Chief Scientific and Medical Officer, Prof. Bryan Williams, emphasized the importance of early diagnosis, noting that a quarter of all deaths in the UK are due to heart and circulatory diseases. By harnessing AI, this study opens new avenues for detecting and managing these life-threatening conditions, offering hope for improved patient outcomes and reduced hospital admissions. Now they plan to carry out a larger clinical trial to prove the AI's worth and efficiency.

The AI hype vs. reality gap in healthcare - 3 practical ways we're actually using AI to improve patient care today While tech headlines promise AI doctors replacing humans, the real revolution is happening quietly behind the scenes. After implementing AI across multiple healthcare organizations, I've seen firsthand: the most powerful AI applications are the ones patients never see. 1/ Clinical documentation is being transformed ↳ Doctors spend 2 hours on documentation for every 1 hour with patients ↳ Our AI-powered ambient listening tools cut documentation time by 63% ↳ Notes are more accurate, capturing nuances human memory often misses ↳ Physicians regain 1-2 hours daily for direct patient care or personal time ↳ The impact: reduced burnout and restored physician satisfaction without changing the patient experience 2/ Risk stratification is becoming proactive ↳ Traditional risk models identify ~40% of high-risk patients ↳ Our AI systems correctly identify 78% of patients who will need acute intervention ↳ Models analyze thousands of variables across structured and unstructured data ↳ Flagging happens automatically, without requiring additional physician time ↳ The impact: earlier interventions for patients most likely to deteriorate, often before clinical symptoms are obvious 3/ Clinical workflow automation is eliminating waste ↳ Average physician receives 77 EHR notifications daily ↳ AI systems filter these to the ~20% requiring human attention ↳ Intelligent routing ensures tasks reach appropriate team members ↳ Smart scheduling optimizes patient flow based on real visit durations ↳ The impact: reduced cognitive load on providers and staff while delivering better care The most effective healthcare AI isn't replacing clinicians—it's removing the administrative burden that prevents them from practicing at the top of their license. While startups pitch expensive AI chatbots directly to patients, we're investing in AI tools that amplify human clinicians' capabilities without disrupting the therapeutic relationship. I've seen health systems chase flashy AI applications that patients can see, while ignoring the unsexy back-office applications that actually move the needle on outcomes, clinician satisfaction, and costs. The future won't be AI doctors. It will be human doctors empowered by AI systems that patients never need to see or interact with. ⁉️ What administrative tasks in healthcare do you think AI should tackle first? What work should remain firmly in human hands? ♻️ Repost to help cut through the AI hype and focus on practical applications that are working today. 👉 Follow me (Reza Hosseini Ghomi, MD, MSE) for more insights on the intersection of technology, neuroscience, and healthcare operations.

DO RISK STRATIFICATION ALGORITHMS WORK? That is the question. Here is our answer: https://lnkd.in/eYgRvcbs with Christopher Oddy, Joe Zhang and Hutan Ashrafian TL;DR: Results from the real-world application of risk-prediction models are disappointing with equal weight of evidence suggesting a harmful effect as a beneficial one. Specifically: 🤖Systematic review: 51 studies evaluating 28 Risk Strat models 📈Half were externally validated, but only 2 validated internationally 📊Model discrimination is satisfactorily robust to application context BUT ... 📉Little evidence that accurate identification of high-risk individuals can be translated to improvements in service delivery/ morbidity. 📉Real-world evals report no change, or significant increases, in healthcare utilisation. 📉Only 1/3 of reports demonstrate some benefit. SO ... ⚕️There's an URGENT need to independently appraise the safety, efficacy & cost-effectiveness of risk prediction systems already widely deployed in primary care. 🏥Deployment of individual-level risk prediction must be subject to the same controls as other medical technologies. ⚖ National bodies involved in the procurement of commercial risk stratification services must review the cost-effectiveness and systemic implications of adjusting the likelihood of individuals within the population they serve accessing care based on personal predicted risk. 🚧 Regulatory bodies, including the Medicines and Healthcare products Regulatory Agency and the US Food and Drug Administration, must either confirm that risk stratification algorithms fall within their purview and are thus subject to the same regulation as other technologies defined as a ‘Software as a Medical Device’, or clarify why these algorithms do not fall into this category.