Please review below our invited speakers who will be presenting as part of the Industry Track at ICHI 2018. We think this remarkable set of speakers will inspire you in your respective work!
Title: Challenges of Deploying AI-Enabled Solutions in Healthcare
Abstract: The concept of building intelligent algorithms to assist in healthcare decision and workflow tasks has been around for over 50 years. Since the late 1960’s medical expert systems that used artificial intelligence (although primitive by today’s standards), emerged to aid in clinical decision making. However, these systems largely remained in the research domain, as integration into the clinical environment has been fraught with complexity. The slow adoption of medical expert systems/AI has largely been driven by the uncertainty of impact will have in clinical decision making, lack of data to build robust algorithms, perplexity of interpretability, and a conservative approach towards regulatory approval. Remarkably, these challenges have not hindered research productivity, where medical AI-algorithms have experienced exponential growth over the past 50 years. We have learned to build sophisticated diagnostic AI algorithms using radiology, genomics, free text, and more; We have leveraged these algorithms to learn from big data, making them more generalizable than their ancestors; We have the computational power to build AI algorithms in minutes instead of years. However, we still face similar challenges with deployment, regulatory approval and acceptance of AI-algorithms. The focus of this talk is to discuss strategies to integrate AI in healthcare, focusing on bridging the gap between scientists, business and regulatory entities; the importance of engaging with clinical partners early and across all levels of the hospital system; the importance of balancing vision with claims to facilitate regulatory approval of meaningful AI-algorithms.
Shawn Stapleton, PhD
Senior Scientist, Artificial Intelligence Lab, Philips Research North America
Dr. Shawn Stapleton is a senior data scientist at Philips Research North America, where he focuses on developing and deploying intelligent algorithms for integrated diagnostics. Before joining Philips Dr. Stapleton was a senior data scientist at OptumLabs, where he was a faculty at Data Science University and also lead a team of data scientists to apply deep learning to over 50M patient’s worth of claims and electronic health care data. In his former academic career, Dr. Stapleton was an NSERC postdoctoral research fellow in Radiology at Massachusetts General Hospital and Harvard Medical School focusing on computational/quantitative image-based methods to assess and model the pharmacokinetics and pharmacodynamics of nanotherapeutics across multiple biological scales. He received his B.Sc. in Physics and Computer Science from the University of Victoria and his M.Sc. and Ph.D. in Medical Biophysics from the University of Toronto. Dr. Stapleton has over 10 years of experience in radiology, image-guided therapy, and machine learning. He has received 11 national and international awards related to his work research in healthcare, and is a technical advisor for several data science training initiatives in the Boston area.
Title: Interpretable Machine learning for Surgical Decision Support Models
Abstract: Do we understand the model and the answers that machine learning algorithms are throwing out? Unfortunately, the complexity that bestows the predictive abilities on these algorithms also make it hard to understand the results produced. In the United States alone, surgeons perform approximately 700,000 arthroscopic procedures every year. Patient outcomes after arthroscopic procedures compared to other non-invasive options revealed no benefit for patients at 12 months follow-up. Surgical Decision support models leads to high percentages of appropriate treatment. Identifying the target audience for intervention is vital to avoid unnecessary procedures. Although Machine learning capabilities solves the complexity about identifying the right patients, interpretation becomes the key to success. The speaker will provide a quick overview of surgical decision support algorithm used to predict patients with high risk for surgical procedures and demonstrate the different techniques available for interpretation of this algorithm. Analysis performed for a small employer group resulted in 34.5 % of surgeries avoided across 4 major conditions with Low back pain having the highest number of avoided procedures. The program resulted in high patient satisfaction and savings worth $1.4M annually and has been running over a period of 3 years. The presentation will eventually help listeners to apply these techniques for interpretation of their in-house machine learning algorithms.
Lead Data Scientist, Aetna
Harini is responsible for the successful delivery of machine learning and analytic solutions at Aetna addressing Provider Fraud/Waste and Abuse. She is responsible for the integration of analytics and data insights into visual products to address ongoing concerns related to Provider Fraud. She identifies potential opportunities and act as the analytics team lead, and is dedicated to developing actionable solutions. Before Aetna, Harini led a team of data scientists and Analysts at SCIO Health Analytics to develop solutions that address surgical decision support models, Medication Adherence, technical consultation for statistical inference to assess care management and Risk Assessment models. Harini has over 8 years of experience with healthcare analytics and predictive modeling. She holds a Bachelor’s Degree in Civil Engineering from Anna University.
Title: Bridging Mental and Physical Healthcare
Abstract: Untreated mental health issues lead to adverse effects on patients, primary care doctors and specialists. In this session, Quartet leadership will share how the company uses data science to bridge mental and physical healthcare for better patient health outcomes. We'll provide an overview of how our algorithms work together to identify behavioral health care needs in patients, leveraging machine learning capabilities to power recommendations for improved care pathways.
Our data science team utilizes expert clinical input, claims, and point of care data to develop models to detect underlying mental health needs and generate care recommendations for providers and patients. The discussion will showcase how this comes together in the real world as Quartet partners with various healthcare constituents to reduce the overall burden of mental health.
Mamta Parakh, MS
Product Director - Algorithms, Quartet Health
Leads strategy and development of Quartet's machine learning platform and recommendation services. She heads a team of talented data scientists and machine learning engineers dedicated to bridging the gap between physical and mental health care. She is responsible for integration of analytics and data insights with our collaborative care platform for providers and patients. Before Quartet, Mamta led product development for a sales intelligence and analytics platform at Collective[i]. With a background in life sciences management consulting, Mamta has over 10 years of experience with healthcare analytics, product development, and user research. She holds an M.S. in Management Science and Engineering from Stanford University.
David Wennberg, MD, MPH
Chief Science Officer, Quartet Health
Leads the Data Science and Business Development functions at Quartet, a New York-based technology company transforming the way mental health is delivered, by making it more accessible and integrated into primary care. David previously served as the Chief Executive Officer of the Northern New England Accountable Care Collaborative (NNEACC), and as the Chief Executive Officer of the High Value Health Collaborative at The Dartmouth Institute. A co-founder of Health Dialog Analytic Solutions, the analytic division of Health Dialog, David served as Health Dialog’s Chief Science Officer. David received his MD from McGill University and MPH from the Harvard School of Public Health, and is a member of the Dartmouth Institute for Health Policy and Clinical Practice faculty.
Title: Graph-based Analytics for Historical, Unstructured Medical Data
Abstract: Graph Analytics has long been used in financial and social engineering for insight discoveries and specific solution recommendations. Example use cases include anti-money laundering, cybersecurity monitoring and insurance recommendation. Graph algorithms such as Page Rank, Collaborative Filtering and Community Detection have been proven effective in discovering correlations, causations and solutions in industry-specific problems. The medical industry has gathered tremendous data over the years, with different formats such as time series data, written texts, to high-definition images. While the analysis of these data could lead to useful insights, the lack of interoperability presents a challenge to traditional information processing. In this work, we demonstrate the Graphen Ardi Platform, which combines a scalable graph database, a high-performance graph analytic engine and a cognitive user interface, to uncover interesting relationships in medical data that led to effective solutions for the industry.
Jie Lu, PhD
Dr. Jie Lu is the Chief Technology Officer of Graphen, Inc. Until June 2017, she was the Project and Technical Manager of IBM System G Graph Tools, which include the Graph Database, Graph Analytics, and Graph Visualization. Jie Lu's work focuses on building smart applications that utilize graph and analytical technologies to model and reason about users and context to enable better system performance and user experience.
She was also the project manager of the DARPA Social Media in Strategic Communications (SMISC) project between September 2014 and June 2015. Before that, she led several projects focusing on developing user-centric tools to support contextual information retrieval, interactive visual analytics, and personalized learning in the enterprise.
Dr. Lu received her Ph.D. from Carnegie Mellon University in 2007, M.S. from Yale University, and B.S. from Tsinghua University. Dr. Lu led the next generation Artificial Intelligence and Big Data-based Anti-Money-Laundering System project that processes billion transactions for Bank of America and led Non-Performing-Loan prediction for ICBC.
Title: Dr. Adam: An Application of Biomedical Knowledge Graph and Artificial Intelligence
Abstract: Due to the rapid advances in biomedical research and related technology, the biomedical corpus is being used at an ever-increasing rate. As a consequence, without the implementation of automated algorithms, it is very difficult for researchers and medical providers to keep up with new biomedical knowledge and development as document classification and semantic role labeling are core challenges. However, training models based on vectors (created from stemmed and/or stopped document word counts) have proven to be a basic and typically successful approach to solving this issue. In this direction we propose Dr. Adam, an artificial intelligence powered humanoid robot capable of answering questions in the biomedical domain by means of Knowledge Graph, Machine Learning and Natural Language Processing.
José A. Alvarado-Guzmán
Senior Data Scientist, Graphen
José is a Biostatistician and Data Scientist currently working at Graphen as a Senior Data Scientist and team lead of the robotic/health unit. He has over 15 years of experience working in different sectors within the Health industry. Before joining Graphen he was a Data Scientist at Columbia University Medical Center in which he leaded an ETL, reporting and analytics team. He is highly interested in the application of Data Science's and Bio-Informatics techniques and algorithms to help improve the population health and health systems by gaining a better understating of health process. His research interest lies in Question Answering using Neural SPARQL Machines, Knowledge Graph and Open Linked Data.
Title: Using AI to synthesize information from the data chaos in the healthcare
Abstract: Founded in the Netherlands over 180 years ago, Wolters Kluwer is a global leader in information services and expert solutions leveraging AI-based Content Enrichment Services (CES) to improve access to intelligence locked in the unstructured text. Doing so, empowers patient, provider, pharmaceutical, health educator and payer across the board to have a holistic impact along the Institute for Healthcare Improvement’s Triple Aim. The CES uses AI and NLP to extract, enrich, organize, and deliver information such as UMLS concepts, prescribed drugs, health outcomes, smoking status, vital signs, and family history from the electronic health records, clinical trials, scientific papers, and human-expert content. To handle the information overload encountered by clinicians, the CES identifies, extracts and aids curation of information in a consumable format that helps doctors, and nurses with the accurate, contextual, just-in-time information to make the right decision. Currently, such recommendations require a considerable manual effort, that is expensive and often inconsistent. Our vision uses AI to augment doctors, nurses, and others to navigate intelligently through the growing volume, velocity and variety of the healthcare information they encounter daily.
Syed Waseem Haider, PhD
Director of Center of Excellence for Cognitive Computing and AI, Wolters Kluwer
Dr. Syed Waseem Haider is a Leader and an accomplished Scientist with over 10 years of research experience in Healthcare, Life Sciences, and Behavioral Science. He is an expert in Data Mining, Machine Learning, and Deep Learning. He has built data insights solutions, and processes. He leads and guides Advanced Analytics platform and operations for Health and Behavior outcomes.
In his current role, Dr. Haider leads and directs the Center of Excellence for Cognitive Computing and AI at Wolters Kluwer. Here, he guides and architects Advanced Technology solutions for Insights, Optimization, Prediction, and Automation for Health, Legal, Regulatory, and Accounting with Scientists and Engineers. He leads and manages a team of Data Scientists, Architects, Engineers, and Business Analysts. He provides thought-leadership, and consults C-level executives in identifying high value market-opportunities, partners in strategic solutions, and creates Intellectual Property with innovative ideas.
In his past role, Dr. Haider led the Department of Data Science in Health and Wellness Solutions at Johnson & Johnson to build Advanced Analytics platform to improve health behaviors. Dr. Haider has led mHealth at Medidata Solutions. He has been a Member of Research Staff at Philips Research, North America. Dr. Haider has developed algorithms ranging from ICU mortality prediction, prediction of Depth of Anesthesia (Surgery), Bayesian model to infer risk of Ketoacidosis for intubated patients, to Clinical Data Standardization using Natural Language Processing and Ontology from UMLS.
He has penned article in leading business communications, and coauthored in top peer-reviewed journals and conferences. He has filed patents and inventions. He has excellent communications, and visually appealing power point presentation skills. He holds a Liaison position between Society of Behavior Medicine (SBM) and American Medical Informatics Association (AMIA).