SWARM Community

Responsible Adoption of AI: A Live Interview with Sima Yazdani Don’t miss this live interview with Sima Y., Executive Director: AI Strategy at ELYON International , as we discuss the essential principles for responsible AI adoption across industries. Sima will share her insights on integrating ethical practices, data governance, and transparency into AI implementation, addressing both challenges and opportunities. Learn how responsible AI can drive positive impact while ensuring accountability and trust in technology. Register here to attend: https://lnkd.in/eNQb62Hw

GenAI Adoption Alongside Graph: a Live Interview with Charles Ivie Join us for a live interview with Charles Ivie, Senior Architect at Amazon Web Services (AWS), as we explore the intersection of Generative AI (GenAI) and graph technology. We’ll dive into the latest trends in GenAI adoption and how combining it with graph or vector technologies can unlock powerful new insights and capabilities. Discover key implementation strategies, potential challenges, and the future of this dynamic technological landscape. Register here to attend: https://lnkd.in/eNQb62Hw

AI Revolution: Driving Transformation in Financial Services Explore the transformative impact of AI in the financial services sector! In this session, Satya Swarup Das, Director of solutions: finance at Unisys, will dive into how AI is perceived, its growing power, and practical applications across the banking value chain. He will highlight five key use cases and discuss the future of AI, focusing on the next steps for integrating AI innovations responsibly and effectively.

From Data to Decisions: The Impact of AI on Supply Chain Analytics The rapid evolution of artificial intelligence (AI) has brought about significant changes in various industries, with supply chain management standing out as a key area experiencing transformative shifts. This presentation explores the profound impact of AI on supply chain analytics, highlighting how AI technologies are reshaping the way companies manage, analyze, and optimize their supply chain operations. By leveraging AI, businesses can transform vast amounts of data into actionable insights, enabling smarter, faster, and more informed decision-making across all levels of supply chain management. Join Ramakrishna Garine, MSAI Student at The University of Texas at Austin, as he shares insights in the journey from data to decisions within supply chain analytics. AI’s ability to analyze and interpret complex data patterns enables supply chain managers to predict demand more accurately, optimize inventory levels, and improve forecasting. For instance, AI-powered predictive analytics can analyze historical sales data, weather patterns, and market trends to forecast future demand with a high degree of accuracy. This reduces the likelihood of overstocking or stockouts, which in turn improves customer satisfaction and reduces operational costs. AI enhances supply chain agility by allowing companies to respond more quickly to changes in demand or disruptions in supply. With AI-driven insights, businesses can make real-time adjustments to their supply chain strategies, ensuring a more resilient and responsive operation. Additionally, AI helps in improving efficiency by automating routine tasks such as order processing and inventory management, freeing up human resources for more strategic activities. However, the implementation of AI in supply chain analytics is not without its challenges. Looking towards the future, the presentation identifies emerging trends that will shape the next phase of AI in supply chain analytics. These include the rise of autonomous supply chains, where AI-driven automation handles everything from procurement to delivery, and the growing emphasis on sustainability, where AI helps optimize resource usage and reduce waste. By staying ahead of these trends, companies can not only enhance their operational efficiency but also build a more sustainable and ethical supply chain. This presentation aims to provide a comprehensive understanding of the impact of AI on supply chain analytics, from its benefits and challenges to future trends and practical implementation strategies. By the end, attendees will be well-equipped to harness the power of AI to drive smarter decisions, enhance efficiency, and build more resilient, agile supply chains that can adapt to the demands of an increasingly complex and dynamic global market. Register here to attend: https://lnkd.in/eNQb62Hw

AI & Cybersecurity Risk Mitigation In Healthcare Management In this presentation, Dr. Shaista Hussain, CEO at SAIF CHECK, will highlight ten key risk scenarios and mitigation processes related to cybersecurity safety protocols in healthcare settings. Each scenario is accompanied by case examples and best practices to address cybersecurity challenges effectively. The importance of maintaining regulatory compliance certifications, implementing transparent AI systems, ensuring human oversight in security protocols, and providing ongoing technology education to staff are highlighted as essential components of a robust cybersecurity framework in healthcare organizations. Register here to attend: https://lnkd.in/eNQb62Hw

The Cutting-Edge in Human Feedback Enabling Development of Frontier Multi-Modal LLMs In this presentation Manas Talukdar, Director of Engineering at Labelbox, will cover the latest in industry trends as well as academic research in the area of human feedback and its role in the development of frontier LLMs. The audience will learn about the various techniques and strategies that are being employed to leverage human feedback to develop powerful multi-modal LLMs that can be used in a variety of generative AI applications in the enterprise as well as consumer space. This includes cutting-edge research such as EvalGen, reward-model based training, AutoQA, etc. Register here to attend: https://lnkd.in/eNQb62Hw

AI-Driven Clinical Decision Support Systems: Ensuring Responsible AI and Explainability in Healthcare As artificial intelligence (AI) continues to revolutionize healthcare, its application in clinical decision support systems (CDSS) holds significant potential to enhance diagnostic accuracy, optimize treatment plans, and improve patient outcomes. However, these advancements come with the responsibility to ensure that AI systems operate in a transparent, explainable, and responsible manner. In this presentation, Roberto Shimizu, CEO of IntelliDoctor.ai, explores the technological foundations and best practices for building explainable and reliable AI systems that healthcare professionals can trust in their clinical environments. AI systems in healthcare must not only provide accurate recommendations but also offer clear justifications for their outputs. This presentation will focus on how AI-driven clinical decision support systems can be designed to provide transparent explanations that support clinicians in understanding how a decision was reached. The discussion will include: • Explainable AI (XAI): Highlighting methods for making AI systems interpretable and understandable by healthcare professionals. This will cover techniques such as model transparency, decision-tracing, and how AI can provide evidence-backed explanations that align with clinical reasoning. • Human-AI Collaboration: Exploring the role of human-centered AI in healthcare, where clinicians can engage with the system’s recommendations, verify their accuracy, and integrate AI insights into their clinical workflow with confidence. One of the most significant risks in deploying AI in clinical environments is AI hallucination—the generation of inaccurate or fabricated information. In healthcare, the consequences of such errors are particularly serious. The future of AI in healthcare depends on its ability to be both powerful and explainable. By focusing on responsibility, explainability, and reliability, AI can become a trusted partner for clinicians, enhancing the quality of care while ensuring that recommendations are grounded in solid evidence. This presentation will offer practical insights into how AI developers and healthcare professionals can work together to implement transparent, explainable, and reliable AI systems that improve patient care and decision-making. Register here to attend: https://lnkd.in/eNQb62Hw

AI in Genomics - How Polygenic Modeling Will Change Human Healthcare The advent of artificial intelligence (AI) in genomics heralds a new era in healthcare, one where the predictive power of polygenic risk scores (PRS) can be fully realized. In this presentation, Edward Messick, CTO of PolyCypher Health, aims to introduce attendees to the concept of PRS, elucidate the current machine learning methodologies employed in their computation, and propose advanced AI techniques to enhance their accuracy and applicability. Polygenic risk scores aggregate the effects of numerous genetic variants to predict an individual's predisposition to various diseases. Despite their potential, traditional PRS models often fall short in terms of accuracy and portability across different ancestries. These limitations stem from the homogeneous nature of many genetic studies, which predominantly focus on populations of European descent. As a result, PRS models may not be as effective for individuals from diverse backgrounds, thus perpetuating health disparities. Our discussion will begin with an overview of PRS, detailing their calculation and current applications in healthcare and biopharma. We will then explore the existing machine learning approaches used to develop PRS, highlighting their strengths and limitations. This will set the stage for a deeper dive into how AI, particularly advanced machine learning and deep learning techniques, can address these limitations. AI offers the potential to enhance PRS in several key ways. By leveraging large-scale genetic data and sophisticated algorithms, AI can improve the accuracy of risk predictions, ensuring they are more reflective of an individual's true genetic predisposition. Furthermore, AI can facilitate the integration of PRS with other healthcare data, such as electronic health records (EHRs) and OMIC data, to provide a more holistic view of an individual's health. For instance, combining PRS with lifestyle and environmental data can lead to more precise and personalized health recommendations. In the realm of biopharma, enhanced PRS can accelerate drug discovery and development by identifying individuals who are more likely to benefit from specific therapies, thus improving clinical trial efficiency and success rates. In conclusion, this presentation will provide a comprehensive overview of the current state of PRS and the transformative potential of AI in enhancing their accuracy and applicability. Attendees will gain insights into the practical applications of PRS in healthcare and biopharma, and understand the importance of equitable AI-driven approaches in advancing personalized medicine. Participants will be equipped with the knowledge to contribute to the development and implementation of improved PRS models, ultimately driving better health outcomes for diverse populations. Register here to attend: https://lnkd.in/eNQb62Hw

Knowledge Augmented Graph Reasoner - A Neuro-Symbolic Reasoning Framework Clinicians are increasingly using applications like ChatGPT and Perplexity AI to get answers to their queries with respect to their patients, signifying a clear demand for tools that can readily answer clinical questions, provide personalized responses based on specific patient details. However, such applications have significant limitations to be clinically applicable - adherence to guidelines being one major challenge for current language models. Instruction tuned models and reasoning enabled inference models like "Strawberry" O1 model from OpenAI are current state of the art methods when it comes to helping models adhere to a given set of instructions or guidelines and/or reason based on them. However, current technology still falls short when it comes to domains like healthcare, and there is a growing need for safe, reliable, accurate and transparent AI systems that are based on validated clinical guidelines. This is the problem & opportunity that Datum is addressing. Join Ravi Bajracharya, CTO of Datum, as he shares a method that enables safe and transparent use of Generative AI (GenAI) based on validated guidelines. This method is applicable in any domain that has strict linear guidelines, protocols, and operating procedures that are a set of instructions to the user. Their approach uses LLMs to convert guidelines into flowcharts, represented as Directed Acyclic Graphs or DAGs that are in turn linked to broader knowledge bases in the domain represented as a domain knowledge graph which can in turn provide more context to the LLM. If the clinical question provided has enough information to traverse the DAG, a response is generated. If not, the LLM asks further clarifying questions, prompting an interactive conversation with the clinical user rather than attempting to provide fabricated or incomplete answers. Datum leverages a biomedical knowledge graph to enhance context retrieval, linking relevant mappings to support question answering, such as gene allele mappings to their genotype or phenotype for example. Datum sees the biggest applicability of this in personalized medicine, and have therefore built out a use case within pharmacogenomics, developing a drug-gene recommendation checker that is able to accurately provide recommendations based on Clinical Pharmacogenetics Implementation Consortium (CPIC) and US Food & Drug Administration (FDA) guidelines, which are clinically validated and recognized sources of pharmacogenomics recommendations. Datum aim to expand this beyond pharmacogenomics to incorporate all health and biomedical guidelines to build out a robust chat application that is able to answer clinical questions accurately and reliably. Having a tool such as this that clinicians could rely on could be a game-changer for health and biomedicine and similar mission-critical domain. Register here to attend: https://lnkd.in/eNQb62Hw

Leveraging Machine Learning for Large-Scale Anomaly Detection in GIS Systems Large-scale anomaly detection plays a crucial role in modern business processes by identifying unusual patterns or events that deviate from expected behavior. When implemented effectively, AI/ML-enabled anomaly detection helps organizations spot product quality issues, detect operational inefficiencies, and identify emerging market trends. However, flawed anomaly detection can lead to significant problems, such as false positives that waste resources on non-issues, or false negatives that allow real threats to go unnoticed. These shortcomings can result in financial losses, reputational damage, and missed opportunities for innovation. The accuracy and reliability of anomaly detection systems can impact a company's bottom line, operational efficiency, and competitive advantage in rapidly evolving markets. This presentation, featuring Dr. Vijay Boppana, Ph.D., Data Scientist at Milvian Group, shares how Milvian Group and Alatrac partnered to apply Machine Learning to the Alatrac GIS mapping technology which provides location intelligence that connects remote staff, locations, and data. Incorporating ML has been a game-changing development for the platform, enabling Alatrac to analyze copious amounts of collected location, staff activity and inspection data for anomalies and inefficiencies. This function delivers critical validation of collected data and provides project managers with insights that empower them to improve field route optimizations and training for remote staff. To achieve this, Milvian Group and Alatrac data scientists performed the following: - Cost Monitoring & Optimization: Utilized AWS Cost Explorer to track and control costs for large-scale experiments and scheduled machine learning jobs, staying within budget limits. - Scheduling & Automation: Automated ML jobs using ml.t3.medium and ml.m5.xlarge instances to balance performance and cost efficiency. - Workflow Optimization: Focused on scalable, cost-effective ML workflows, improving model performance for business applications. - False Flag Elimination: Obvious or common anomalies like rapid inspections in proximity or incomplete inspections due to access restrictions are identified. Distance in miles is calculated using latitude and longitude values, aiding in anomaly detection. Implemented SQL and data preprocessing techniques to remove false flags due to specific event sequences, enhancing anomaly detection accuracy. - Anomaly Detection: Used Isolation Forest models, tuned for optimal performance, to identify anomalies, including geographic proximity and access-restricted inspections. - Continuous Improvement: Applied a continuous improvement approach for anomaly detection, involving baseline performance establishment, feature extraction, EDA, feature selection, model selection, and hyperparameter tuning. Register here to attend: https://lnkd.in/eNQb62Hw