Rachael Fleurence, Ph.D.’s Post

💡 Harnessing AI for Enhanced Evidence Generation: NICE UK and Cochrane's Perspectives The integration of Artificial Intelligence (AI) in evidence generation is reshaping how we approach health technology assessment (HTA) and systematic reviews. Both NICE - National Institute for Health and Care Excellence UK and Cochrane are at the forefront of this transformation, each providing valuable guidance on the responsible use of AI. 📄 NICE UK’s recent position statement emphasizes the importance of balancing innovation with the need for transparency, trustworthiness, and ethical considerations. AI methods, from established machine learning to generative models, offer the potential to efficiently process large datasets, revealing patterns and insights that might be missed by human analysts. As NICE continues to explore AI’s role in evidence generation, the focus remains on ensuring that these technologies augment, rather than replace, human expertise. 🔍 Cochrane is similarly committed to the responsible use of AI in systematic reviews. They advocate for transparency in AI-assisted research and stress that AI tools, while powerful, cannot replace human authorship. Cochrane encourages research into AI tools, endorses the public sharing of this research, and is actively working to define standards for AI evaluation within their processes. Both organizations highlight the need for ✅ clear justifications, 🔬 robust validations, and 📊 transparent reporting when using AI. This ensures that AI-driven insights are reliable, unbiased, and ethically sound, supporting more effective and informed decision-making in healthcare. Over the past few years, Pharmacoevidence has been dedicated to enhancing human expertise with AI-driven approaches that are both transparent and compliant with established checklists and guidelines. We have successfully achieved up to 95% accuracy in automated systematic literature reviews (SLRs), all while being guided by subject matter experts. 📞 Connect with Pharmacoevidence for more insights on leveraging AI in evidence synthesis, meta-analysis, health economics modeling, and automated reporting. Reach out to our experts Dr. Rajdeep Kaur and Nicola Waddell to discuss how AI-based solutions can elevate your HEOR, Market Access, and life science projects. 🌟 Embrace the future of data-driven insights and elevate your decision-making process. #AI #EvidenceGeneration #Innovation #NICE #DataScience #slr #systematicliteraturereview #literaturereview #nma #metaanalysis Pankaj Rai Shubhram Pandey Gagandeep Kaur Sumeet Attri Akanksha Sharma

Pharmacoevidence is integrating AI-driven approaches and LLM models to advance HEOR applications

How can we ensure that by AI-Driven advancements in HEOR are acceptable to stakeholders and regulatory bodies? As AI continues to demonstrate its reliability and precision as a viable tool for conducting systematic literature reviews and meta-analyses, Barinder Singh shares our approach. Reach out to Dr. Rajdeep Kaur & Nicola Waddell to discuss the potential of augmenting your Health Economics Outcomes Research with Generative AI. #HEOR #GenAI #SLR

📢 Masterclass Announcement!!!! Brenda Dooley, AXIS CEO and AXIS Head of HTA Consulting, UK, Carl Prescott will take part in the Life Science Access Academy's upcoming Masterclass ‘The Use of AI Tools in a NICE Submission – Reality or Pipedream?’ 📅 Tuesday 14th May 2024 13:00 - 14:00 (BST).   This exciting 60-minute masterclass for LSAA members will look at the opportunities for using AI in HTA and reflect on what can be done to overcome the barriers.   Artificial intelligence offers transformative opportunities for the way health technology assessment (HTA) is conducted. Many of the benefits, such as increased speed and accuracy of data analysis, are easy to envision, however, there are many barriers when it comes to embedding AI into routine HTA practice. Key Learning Objectives: 💡 Gain an understanding of the potential opportunities for AI in HTA, particularly for HTA bodies and for the Life Science industry. 💡 Learn about the potential barriers to embedding AI in HTA. 💡 Find out what can be done to try to overcome these barriers. For more information or to book your place please contact The Life Science Access Academy at this link https://lnkd.in/ed63pWVZ #masterclass #AI #AIinHTA #HTA #MarketAccess #thoughtleaders

Imagine a world where clinical trial timelines are predicted with unparalleled accuracy and based in historical patterns. Now stop imagining, because Kognitic Predict Next-Gen is here to make it a reality. 🚀 At Kognitic, we are not just reporting, we are predicting- and our latest release is a testament to our commitment to pioneering a new era of AI-powered intelligence. Here are the highlights of our updated Prediction module: 🎯 Enhanced Accuracy: Kognitic Predict Next-Gen demonstrates superior accuracy compared to Sponsor estimations, particularly at trial launch. Our new model uses neural networks, which when combined with Kognitic’s AI NLPs, can analyze vast clinical trial datasets to identify the patterns within. While sponsor predictions can be off by over 1 year, our forecasts exhibit a deviation of less than 1 quarter, showcasing our enhanced predictive capabilities. 📊 Revamped Visualization: Based on user feedback, we've revamped our prediction visuals within the Predict 2.0 module. The platform has a timeline dashboard to track multiple trial launches in one place. Our new launch features intuitive, customizable visuals including confidence scores and prediction summaries for maximal clarity. 🔍 Compare "Look-alike" Trials: Predict 2.0 allows you to view and compare complete and active "look-alike" trials, empowering you to make informed decisions based on historical data and patterns. Ready to explore the future of predictive analytics in clinical trials? Reach out to Chad Hinrichs ([email protected]) to learn more and unlock the power of Kognitic Predict Next-Gen! 💡 #Kognitic #PredictiveAnalytics #ClinicalTrials #AI #Innovation #HealthcareTech #DataScience

AI Tool “Foresight” Revolutionizes Healthcare Predictions NHS healthcare researchers collaborated to create Foresight, an AI application aimed at enhancing patient health outcome predictions. Developed by a team from London NHS Foundation Trust, including Kings College Hospital and Guy’s and St Thomas’, alongside academic institutions like King’s College London and University College London, Foresight marks a significant advancement in clinical decision support and health monitoring. Utilizing AI for clinical insights, Foresight operates on the Cogstack platform, employing cutting-edge natural language processing to extract valuable health data from NHS electronic health records. Unlike traditional methods, Foresight utilizes deep learning to discern complex patterns in both structured and unstructured data, thus improving its predictive accuracy. Recently featured in The Lancet Digital Health, Foresight demonstrated promising results, detecting ten upcoming disorders with high accuracy across various datasets. Achieving 68% accuracy at King’s College Hospital, 76% at Maudsley NHS Foundation Trust, and an impressive 88% with the US MIMIC-III dataset, Foresight’s effectiveness was further validated through comprehensive evaluations by healthcare professionals. Beyond clinical settings, Foresight holds potential for real-world risk forecasting and clinical research simulation, as noted in a report by the National Institute of Health Research Maudsley Biomedical Research Centre. Professor Richard Dobson, the lead researcher, emphasized the pivotal role of real-world data in enhancing Foresight's capabilities. Funding for Foresight's development was secured from various organizations including the NHS AI Lab, National Institute for Health and Care Research, and HDRUK, ensuring privacy protection and promoting the widespread adoption of technological solutions in healthcare. Continuously evolving, the Foresight team is working on Foresight 2, leveraging data from multiple hospitals to improve predictions and expand its applicability beyond healthcare settings. With a commitment to success, Foresight 2 aims to further revolutionize healthcare decision support systems. #ValueIcons #ValueIconsMagazine #VI #ValueIconsBusinessLeadership #ValueIconsExecutiveInsights #IndustryLeaders #ValueIconsCorporateVision #ValueIconsLeadershipPerspectives #ValueIconsBusinessMagazine #ExecutiveEdition #CEOInsights #IndustryMagazine #ValueIconsLeadershipJournal  #ProfessionalGrowthJourney #ArtificialIntelligenceHealthCare #AIHealthCare #AIHealth #healthcare #AIHealthEducation #predictions #healthcaretips #worldhealthorganization #aiforeveryone #health #hope #healtheducation #healthpoint

What's the technology readiness level (TRL) of your clinical AI? That's one of the things I'll be covering during my session, "Moving Towards Trust, Validation and Generalizability in Clinical AI" at the International Conference on Complex Acute Illness (ICCAI) 2024 in September in Bethesda focused on "Maximizing Clinical Impact of Algorithmic Innovation (AI) in Critical Illness: Artificial Intelligence from Academia & Industry to Government". https://iccai.org/ For those who are not familiar with it, the NASA developed TRL scale is an easy tool for technologists, SMEs, admins and leadership alike to quickly get on the same page about the maturity level of a technology. As Fleuren et al (2020) demonstrated in their translation of the TRL scale for clinical AI, 93% of the algorithmic tools they surveyed for intensive care were stuck in TRL 3 or 4 - model prototyping and development. Only 5% had made it to TRL 5 - model validation. And only 1% had made it to TRL 8 - clinical outcome evaluation. None made it to TRL 9 and model integration. This was back in 2020 before the widespread push and publicity for generative AI and associated large language models, which seems to have only down shifted the numbers to an even less mature state. Moving fast and breaking things in health care generally means adverse events and negative outcomes, so an accurate TRL level and related model validation (TRL 5), real time model testing (TRL 6), workflow implementation (TRL 7), clinical outcome evaluation (TRL 8), and model integration (TRL 9) are critical to any clinical AI plan. If you are interested in improving health outcomes (and doing no harm) with clinical AI, come join us in Bethesda this September 4-6 for ICCAI 2024 (co-hosted by my grad school alma mater, Uniformed Services University of the Health Sciences). #medicine #health #tools #evaluation #technology #science #research Heather Leigh Flannery Hassan Tetteh MD MBA FAMIA Sandy Barsky Paul Tibbits, MD Paul Kavitz, Chris Regan, Seth Schobel, PhD Gilles Clermont Yoram Vodovotz Roger Boodoo, MD F. J. Haran Tamiko Eto

Introducing FoundationEHR: A Transformer-Based Model for the Utilization of Structured Electronic Health Records The healthcare industry's use of structured electronic health records (EHRs) spans various applications, from health-economic studies to marketing strategies. Integrating EHRs for personalized risk prediction, understanding disease trajectories, and simulating clinical trials is emerging as a critical area of focus. In this evolving landscape, the application of artificial intelligence (AI) and machine learning (ML) technologies is becoming increasingly prevalent. #FoundationEHR, a transformer-based neural network designed to utilize structured EHRs for applications such as personalized risk prediction, is now available. Our recent study (https://lnkd.in/e4tCdZnZ) describes FoundationEHR's pre-training on a large dataset comprising nearly a billion EHR entries from 3.5 million patients across the United States. Our findings show that FoundationEHR surpasses other ML models in predicting severe COVID-19 progression based on a patient's medical history. We are excited to provide healthcare professionals and institutions with FoundationEHR, allowing them to derive additional, task-specific AI/ML models through fine-tuning. For more details on FoundationEHR, please visit  https://lnkd.in/ebJeruAk.

@NYUStern Professor Natalia Levina has been awarded as a distinguished recipient of the “ISSIP 2023 Excellence in Service Innovation Award Program”. See below for the full details of the award, via ISSIP: Award Type: Impact to Innovation Innovation: To Engage or Not to Engage with AI for Critical Judgments: How Professionals Deal with Opacity When Using AI for Medical Diagnosis Summary: This study is the first in-depth, longitudinal investigation of how ML-based prediction tools are used for medical diagnosis decision making. AI tools may operate in similar ways technologically, but they are experienced in unique ways depending on the users, context, and work practices. This is a fundamentally new way to think about opacity and has implications for how ML tools are designed and deployed in critical decision making contexts. Organization: University of Virginia, Warwick University, New York University Primary Sarah Lebovitz, Hila Lifshitz-Assaf, Natalia Levina See more details regarding ISSIP and the award here: https://lnkd.in/gwFB-P5j #NYUStern #NYU #FubonCenter #ISSIP #Innovation #AI #Business

🔬 AI in Healthcare 🔬 As we continue to explore the intersection of AI and healthcare, we would like to share a groundbreaking study by a group of esteemed scientists: "A Scoping Review of Using Large Language Models (LLMs) to Investigate Electronic Health Records (EHRs)." Key Insights: 📈 Trends & Applications: The study analyzed 329 papers, highlighting LLMs' rapid advancements and diverse applications in EHR research. Key areas include: Named Entity Recognition Information Extraction Text Summarization Text Classification Dialogue Systems Diagnosis and Prediction 🚀 LLMs Transforming Healthcare: Models like BERT and GPT are revolutionizing how we interpret and utilize EHR data, enhancing clinical documentation and decision-making processes. 🔍 In-depth Analysis: The review categorizes research into seven critical topics and discusses LLMs' unique capabilities and challenges in each area. 💡 Implications for Future Research: The study offers valuable guidance for researchers and practitioners looking to leverage LLMs, focusing on data resources, prompt engineering, fine-tuning, performance measures, and ethical considerations. https://anablock.com