Artificial intelligence is a powerful tool that can increase the speed, efficiency, and effectiveness of global health systems. By analyzing large amounts of data in real time, AI can help improve clinical and nonclinical decision making, reduce medical variability, and optimize staffing. Likewise, AI can reduce the volume of tedious administrative tasks that often lead to burnout among healthcare professionals. This organizational expert panel will share their AI experience and how we are using AI at UHG.

Optum Digital has deployed a 5-layer deep neural network to predict the next best health action to recommend for a given user on a personalized basis. This model, which powers real-time recommendation ranking throughout the digital experience, has driven a 42% increase in the click-through-rate of the top-ranked real time offer.

The Optum Enterprise and Data Analytics (EDA) Graph & Health @ Scale (GHS) team is happy to announce v1.0 of our g2gnn library. In this presentation we will discuss the g2gnn library, how it works, and how to start using it. We will provide example applications from our own work and also share a demonstration website that allows users to interactively experiment with the tooling.

As Natural Language Processing-based solutions continue to be implemented across the enterprise, there is a need to develop an automatic speech recognition (ASR) solution that is highly accurate, flexible, and affordable to enable NLP at scale. Many enterprise teams, as well as the ATC, are already using offline and real-time ASR services, such as Nexidia ASR and Microsoft Azure Cognitive Services ASR. These efforts have proven that the performance of ASR directly impacts the performance of all downstream NLP solutions. Therefore, the availability of customizable ASR solutions that are optimized for cost-performance will be a foundational capability that benefits the ongoing work of the ATC as well as other enterprise teams.

This presentation will cover about the approach and utility for ingesting DICOM images for the PA and how EDA created the capability to allow providers to share the images with ease while creating the consumption (measurement) approach and improving Medex savings.

One of the biggest challenges in AI today is the fact that most machine-learning based systems don’t learn how to generalize well. They can’t take two similar concepts and connect them together to form a named concept that can be reused in many areas. In this talk, we will discuss the current limits of machine learning and how new knowledge representations can be better at generalization. We will discuss the work of Jeff Hawkins, and Francos Chollet on generalization and then describe how we can structure Healthcare Concepts in the form of graphs to build composable rules that can be reused in many context. This presentation will focus on the practical work for doing analysis and surveillance on integrated healthcare data by Mark Megerian and others within the Healthcare Graph team and OptumHealth.

Our development and use of AI should always be carried out responsibly. During this session, we’ll discuss our enterprise principles, guidelines and Program for Responsible Use of AI/ML. Teams will learn about key considerations and specific expectations for their AI work.

The OTU AI Practitioner Learning Path focuses on the data science, machine learning and deep learning skills necessary to develop your competency in artificial intelligence practices. Programs within the learning path are designed to lay a foundational understanding of data science and analysis concepts, reinforced with hands-on experience deploying and optimizing production level ML models, and deep learning frameworks through hands-on labs and projects. Programs within this learning path are designed to build on each other, and you can enter the learning path at any stage depending on your skill level. No matter where you are at in your learning journey there is a program suited to your desired area of development.

Overview of Optum Ventures, our portfolio, and our ongoing collaboration with the UHG enterprise.

A Data Science algorithm that aims to capture and evaluate the change in provider’s claim behavior as compared to their historical claims, and thereby indicates potential FWAE activity.

The increasing availability, accuracy and cost effectiveness of AI technologies have fueled the uses of Artificial Intelligence in more user-facing products and services than ever before. However, with the increased use of these technologies comes more challenges and questions about how best to deliver AI products and services. Successful AI products and services require more than big data, algorithms, and advanced math, they must reflect a reasonable understanding of user needs and behavior and instill an acceptable le(OTU) the Human-Centered Artificial Intelligence (HCAI)vel of explainability and trust. Human-centered artificial intelligence (HCAI) can help us to make sure we are delivering products that people need and want. HCAI is a person-first approach to creating AI products. It incorporates human-centered design principles and design thinking to create products that suit human needs and augment human capabilities.

Using NLP and Deep-learning to identify undiagnosed conditions, improve member outreach and documentation accuracy.

Join Laura Hamilton, VP of Product for Personalization and Machine Learning, for a deep dive into how her team replaced a keywords-based search experience with natural language processing (NLP). The new ML powered approach drove a 3.9X increase in clicks per day, a 70% increase in the program click-through-rate, and a 6.8X increase in program enrollments within the search experience.

Explainable predictions are any prediction where the results can be understood by humans. Explainable AI predictions are critical for clinical settings where we must be able to explain a recommendation to both patients and healthcare providers.

This capability course is designed to help everyone, even nontechnologists, become familiar with Machine Learning Based Natural Language Processing. There are no programming skills required for this course.

Natural Language Processing (NLP) is one of the fastest growing fields of AI, delivering business value every day at UHG. This class will help you understand how NLP can help you solve business problems. We will focus on the newest machine-learning forms NLP based on large language models (BERT and GPT). We will introduce you to the high-level concepts such as embeddings and sematic search. We will describe the key use-cases of these technologies with UHG including analyzing patient notes, summarizing call center transcripts, creating chatbots, looking for trends in sentiment and finding facts in documents. We will also describe some of our recent pilot projects and look at key trend for the future.

AI-enabled creation of call summaries to promote consistency and quality in call notes, reduce manual effort, and improve employee experience.

Combining Image & Genome based ML techniques for better Oncology related analysis Early and accurate detection of cancer is must for ensuring healthier life for our members. Usually, a biopsy is performed to examine tissue to diagnose cancer. While physician decision is the ultimate truth in this area, research has shown enough variability and subjectivity in biopsy analysis which leads to delays in treatment. An AI and ML based standardized mechanism that can objectively analyze the huge biopsy images and can assist the clinicians in decision making can create a big difference and seems to be the one stop solution. However, historically it has been seen that even machine learning solutions have limitations when used alone. We will present here a mechanism that not only enhances the AI&ML approach but also substantiates the evidence based (biopsy image analysis) system with circumstantial markers (genetic markers) using the two pronged approach. Both image and genomic features set are combined and fed into multiple layers of deep learning model, which provides better accuracy in predicting grades of cancer. With the advancements in the computing and deep learning capabilities and the ever- reducing costs of DNA sequencing, this could be a game changer. Come learn more about this fascinating technology!