Artificial Intelligence Challenges the Status...

43
Artificial Intelligence Challenges the Status Quo NCHICA 23 rd Annual Conference & Exhibition: Durham, NC September 11, 2017 Vivian West, PhD, MBA, RN: Duke University, Durham, NC

Transcript of Artificial Intelligence Challenges the Status...

  • Artificial Intelligence Challenges the Status Quo

    NCHICA 23rd Annual Conference & Exhibition: Durham, NCSeptember 11, 2017

    Vivian West, PhD, MBA, RN: Duke University, Durham, NC

  • Technology…EHRs…Automation

    • What is AI?• How is AI currently being used?• How is AI challenging the way healthcare

    is provided?• Looking ahead to the future of health

    care…

  • What is Artificial Intelligence (AI)?• Technology that takes in huge amounts of

    information from a specific domain • Uses it to make a decision in a specific case

    in the service of a specified goal

    “Systems that perform tasks intelligently”

    • “A spreadsheet on steroids trained on big data.”

  • Artificial Intelligence

    The theory and development of computer systems able to perform tasks that normally require human intelligence, such as visual perception, speech recognition, decision-making, and translation between languages.

    Google Dictionary

    A branch of computer science dealing with the simulation of intelligent behavior in computers. Merriam-Webster Dictionary

  • Alan Turing• Mathematical code-breaker WWII• Cracked the enigma code• Theorized the first computer• Widely held to be father of

    computer science and AI

    Imitation Game 2014

    Philosophy of AI

  • • "Computing Machinery and Intelligence”• "I propose to consider the question, 'Can

    machines think?'"• "Are there imaginable digital computers which

    would do well in the imitation game?"

    Turing, Alan (October 1950), "Computing Machinery and Intelligence", Mind,LIX (236): 433–

    460,

    Turing Test

  • Turing Test• Exhibit intelligent behavior indistinguishable from

    a human– The branch of computer science: technologies allow

    a computer to pass (or surpass) the Turing test.• “So when a clinical decision support tool says it

    “uses artificial intelligence” to power its analytics, be aware that “using principles of computer science associated with the development of AI” is not really the same thing as offering a fully independent and rational diagnosis-bot.” Bresnick J. Machine Learning in Health Care:Health IT Analytics. https://healthitanalytics.com/news/machine-

    learning-in-healthcare-defining-the-most-common-terms

  • Keyword and PubMed

    July 11, 2017: “artificial intelligence”= 71,828• Machine Learning• Neural networks• Deep learning • Neural networks • Natural language processing• Text mining

  • TutorialMachine learning (ML): uses data and algorithms with statistical analysis to predict outcomes. Supervised (human input) & unsupervised (iterative)Neural Networks (NN): computer systems/software “taught” to recognize similarities and cluster like groups: pattern recognitionDeep Learning: a branch of AI- like neural networks, “learns” pattern recognition in digital representationsNatural Language Processing (NLP): a branch of AI that processes human (natural) language

    Text Mining: uses NLP and ML techniques to derive insight from text

  • Deep Learning

    • Subset of machine learning • Deals with artificial neural networks (algorithms

    that mimic biological brains with neurons and synapses).

    • Pattern recognition, graphic modeling

  • Natural Language Processing• System that processes human language• Examples:

    – Spam classifier (looks at title, date, return address, etc. to determine if it is spam)

    – Google requests• Key component of NLP is semantic computing

    – Uses related data to develop conclusions about meaning, content, and structure of data sets.

    – Uses key elements of both computer science and linguistics.

    Used for text mining, machine translation, automated question answering

  • How is AI being used?

  • "Artificial intelligence is one of many leading-edge technologies that China seeks and that has potential military applications“ (Sen. John Cornyn aide)

    MQ-9 Reaper remotely piloted drone over Creech Air Force Base, Nevada, 6/25/15. U.S. Air Force/Senior Airman Cory D. Payne/via REUTERS

  • HAL 9000 (Heuristically programmed Algorithmic computer

    Arthur Clarke & Stanley Kubrick: Space Odyssey

    https://en.wikipedia.org/wiki/HAL_9000

    • Facial recognition• Lip reading• Natural language

    processing• Speech• Speech recognition• Playing chess• Interpreting

    emotional behavior• Art appreciation• Reasoning

  • Amazon Forbes, May 10, 2017

    https://www.forbes.com/sites/howardhyu/2017/05/10/with-echo-show-amazon-seizes-the-lead-in-the-artificial-intelligence-race/#6c3299571e37

    “Launched a smart Wi-Fi speaker that not only hears what you say, but also shows you what it sees. This 7-inch touchscreen-equipped digital assistant — Echo Show —is the latest addition to the Alexa family, which began as a simple voice-controlled alarm clock and is fast becoming the central hub of digitally connected homes. All your home appliances and accessories, including your vehicle, will soon be connected to the Internet and voice controlled. Someday in the future, when the milk is running low, your fridge will text you a shopping list while routing your Volkswagen to the closest Walmart on the way home.”

    https://www.amazon.com/Amazon-MW46WB-Introducing-Echo-Show/dp/B01J24C0TI/ref=sr_1_10?ie=UTF8&qid=1494337253&sr=8-10&keywords=echo+showhttps://www.engadget.com/2017/01/07/volkwagen-is-adding-amazon-alexa-to-its-cars/

  • Personal AssistantsAmazon: Alexa (Echo)

    • Ford Motor Co.• Sears’ Kenmore refrigerators• Link to Microsoft Outlook

    Samsung: Bixby• Other Samsung products

    Alphabet: Google Assistant (Cortana)• Sony• LG Electronics (dryers and vacuum cleaners)• Walmart

    Apple: Siri

  • Google Brain Research“Make machines intelligent. Improve people’s lives.”

    • Machine learning– Deep Learning– Neural networks

    • Natural Language understanding (for machines to excel at tasks humans are good at)– Word2vec maps words to vectors (open sourced

    2013)—Neural networks• Perception (ability to hear and see: deep learning

    systems)

  • Google Brain Research• Music and art (neural nets for artistic purposes)

    • Robotics: Self-driving cars

    T. Vanderbilt. Let the Robot Drive: The Autonomous Car of the Future is Here. Wired: 1.20.12. Photo by Spencer Higgins

  • Google’s self driving car drove 80 million miles before it ever touched a road

    https://www.healthcatalyst.com/webinar/md-anderson-ibm-watson-announcement-mean-machine-learning-healthcare/

  • Hanson Robotics: Meet Sophia

    • Natural facial expressions

    • Cameras in eyes. Can make eye contact

    • Understands speech and remembers

    https://youtu.be/wimUaNqEJyw?t=697

    https://youtu.be/wimUaNqEJyw?t=697

  • Things described as using AI• Online shopping• Games• Smart cars• Credit card fraud• Online Support (chat bots)• Wordsmith Journalist (Automated Insights, Vista)• Security Surveillance• Entertainment: Pandora (thumbs up/down),

    Netflix, Spotify

  • 8 more things you didn't know AI could do

    Protect the world from errant asteroidsEncourage rape and pedophilia (4 US companies)Get marijuana recommendations (PotBot)Make roti (Rotimatic)Read body language (Carnegie Mellon University Robotics)Optimize nuclear power stations (EDF Energy, UK,

    sensors)Help particle physicists discover new subatomic particlesExplain the birds and bees to your kids (SophiBot)

    7/17/17: O'Reilly Artificial Intelligence Newsletter [email protected]>

    https://urldefense.proofpoint.com/v2/url?u=http-3A__post.oreilly.com_rd_9z1zc4u82l8eupmcch1d4ae8g54qd784gfh0ms8utd0&d=DwMBaQ&c=imBPVzF25OnBgGmVOlcsiEgHoG1i6YHLR0Sj_gZ4adc&r=TgWXq_oC_5KHEfC7vqjBuXZOErnVLG1MnXR3g7p_RAI&m=MZTjkhv7N3TPqEcnLQOvFoMJzW096jXzHe3XrKsF99Q&s=gOEQVhLWfT5ENcX77HL_1_N77AbSp888eJptCInncTM&e=https://urldefense.proofpoint.com/v2/url?u=http-3A__post.oreilly.com_rd_9z1zegfd93eg888nprp7e5rrgglc3pba772kqjmo3v8&d=DwMBaQ&c=imBPVzF25OnBgGmVOlcsiEgHoG1i6YHLR0Sj_gZ4adc&r=TgWXq_oC_5KHEfC7vqjBuXZOErnVLG1MnXR3g7p_RAI&m=MZTjkhv7N3TPqEcnLQOvFoMJzW096jXzHe3XrKsF99Q&s=MSvXzJYBKrhIrzG0NbsefcyL4VW-sqUKVNLRs7TSAgo&e=https://urldefense.proofpoint.com/v2/url?u=http-3A__post.oreilly.com_rd_9z1zdvb3jhbeabd9af39o3rmb61j7vtq5v9mjdnq4a8&d=DwMBaQ&c=imBPVzF25OnBgGmVOlcsiEgHoG1i6YHLR0Sj_gZ4adc&r=TgWXq_oC_5KHEfC7vqjBuXZOErnVLG1MnXR3g7p_RAI&m=MZTjkhv7N3TPqEcnLQOvFoMJzW096jXzHe3XrKsF99Q&s=vWe0W2sGWDqo_J0DS7Ir_DuIWO6ZPYo_GjlaSxjv-z8&e=https://urldefense.proofpoint.com/v2/url?u=http-3A__post.oreilly.com_rd_9z1zcoebtdte9l1lsgslga0d8adbcv4hj1b0e1g3ftg&d=DwMBaQ&c=imBPVzF25OnBgGmVOlcsiEgHoG1i6YHLR0Sj_gZ4adc&r=TgWXq_oC_5KHEfC7vqjBuXZOErnVLG1MnXR3g7p_RAI&m=MZTjkhv7N3TPqEcnLQOvFoMJzW096jXzHe3XrKsF99Q&s=cpjr2RCmgGHoCzDzImfKlXtGVJFrkYXmQ8GgyEOa7C4&e=https://urldefense.proofpoint.com/v2/url?u=http-3A__post.oreilly.com_rd_9z1zdped9qj0tq11ltk0a3fh1772svf7gq6e8jk552o&d=DwMBaQ&c=imBPVzF25OnBgGmVOlcsiEgHoG1i6YHLR0Sj_gZ4adc&r=TgWXq_oC_5KHEfC7vqjBuXZOErnVLG1MnXR3g7p_RAI&m=MZTjkhv7N3TPqEcnLQOvFoMJzW096jXzHe3XrKsF99Q&s=9k-fJONiG-IxBeCVTLiX2noowbKczwDG_tNZXAoUpYs&e=https://urldefense.proofpoint.com/v2/url?u=http-3A__post.oreilly.com_rd_9z1zcclk75sf7kp8otgo8kvuahj78iqhdu11p7146lo&d=DwMBaQ&c=imBPVzF25OnBgGmVOlcsiEgHoG1i6YHLR0Sj_gZ4adc&r=TgWXq_oC_5KHEfC7vqjBuXZOErnVLG1MnXR3g7p_RAI&m=MZTjkhv7N3TPqEcnLQOvFoMJzW096jXzHe3XrKsF99Q&s=hsT_NJiKpF7eEW-P_m6JX4eU_dBwwTHEV6Ubv6deJqg&e=https://urldefense.proofpoint.com/v2/url?u=http-3A__post.oreilly.com_rd_9z1zd6jh88fsgog18mf14p6k29hd44jfm5qdmdn944g&d=DwMBaQ&c=imBPVzF25OnBgGmVOlcsiEgHoG1i6YHLR0Sj_gZ4adc&r=TgWXq_oC_5KHEfC7vqjBuXZOErnVLG1MnXR3g7p_RAI&m=MZTjkhv7N3TPqEcnLQOvFoMJzW096jXzHe3XrKsF99Q&s=i-PWFdTkcRZtoWqtT07sbJUf0LKnzgdu9MMwO6XBUFo&e=https://urldefense.proofpoint.com/v2/url?u=http-3A__post.oreilly.com_rd_9z1zpsutdvruohb423ubdmkelj7ofcligf7hhisnirg&d=DwMBaQ&c=imBPVzF25OnBgGmVOlcsiEgHoG1i6YHLR0Sj_gZ4adc&r=TgWXq_oC_5KHEfC7vqjBuXZOErnVLG1MnXR3g7p_RAI&m=MZTjkhv7N3TPqEcnLQOvFoMJzW096jXzHe3XrKsF99Q&s=yRMZCXdIUhZ1MHRqFu3lU_4H2s40eDb_1ZdQrNWNDqE&e=

  • PotBot: Personalize your medical marijuana: More Science, Less Guessing

    *Symptoms/conditions*Recommended cannabinoid levels and strains*Dispensary near you

    http://www.potbot.com/?imm_mid=0f44aa&cmp=em-data-na-na-newsltr_ai_20170717

  • AI in Health Care

  • Mycin System (1970s)• Early expert system written in Lisp• Identify bacteria causing severe infections and

    recommend antibiotics, …dosage adjusted for patient's body weight

    • Diagnosis of blood clotting diseases. • Proposed an acceptable therapy (69%), was

    better than infectious disease experts• Never used

  • Wall Street Journal July 8-9, 2017 , C1.

  • Microsoft Seeing AI App

    “Uses Artificial intelligence and camera on iPhone.”

    • Reads documents• Produce identification with barcode• People identification: facial recognition

    • Gender, facial expression, other identifying conditions

    • Surroundings (Beta testing)

    Seeing AI App is Now Available in the iOS app store. July 12, 2017https://blogs.msdn.microsoft.com/accessibility/2017/07/12/seeing-ai-app-is-now-available-in-the-ios-app-store/?imm_mid=0f44aa&cmp=em-data-na-na-newsltr_ai_20170717

  • NLPhttps://www.cultofmac.com/293066/iphone-app-diagnose-jaundice-babies/

  • Opthalmology• 60% Agreement• 65% Self-agreement

    • Used deep learning algorithms • Training set: 410 images of congenital

    cataracts + 476 images disease-free• Diagnosis re: severity and treatment

    suggestions the same (98%)• “Complement the physician”

    • DeepMind and Mooresfield Eye Hospital 2016

  • Google DeepMind

    • University College London Hospital– August 2016– 700 scans of head and neck cancer – identify cancerous tissue versus non-

    cancerous (segmentation)• Can AI identify areas to be treated?• Could reduce time from 4 to 1 hrs.

  • Google, Verily Life Sciences• Alcon: smart lens, glucose monitoring• Debug: mosquitoes…good bugs versus bad bugs• GlaxoSmithKline…Galvani Bioelectronics: clinical

    proof of principle in inflammatory, metabolic and endocrine disorders and developing miniaturized, precise devices

    • Sanofi…Onduo: Focusing on Type 2 diabetes. Using devices, software, medicine, professional care for intelligent disease management

    • Ethicon (J&J)…Verb Surgical, platform using robotics, visualization, machine learning, advanced instrumentation to assist OR team

  • Verily Science Watch• UNC, Harvard…Aurora Study

    5,000 post-traumatic conditions• Duke and Stanford: Project Baseline

    Longitudinal study; transitions between health/disease

    • Netherlands: Personalized Parkinson’s ProjectMolecular analysis, brain imaging, sensorsIn-depth analysisCharacterization of disease

  • Google Brain, Verily Life Sciences• “Deep learning AI research project”• Breast cancer• Neural network architecture with images

    – Applied to digital pathology– Surveying the data on a pixel-by-pixel basis– Produced “heat maps” to predict likelihood of

    tumor cells• 89% accuracy vs. 73% pathologist accuracy

    HealthITAnalytics.com is published by XtelligentMedia, LLC. 3/6/2017https://healthitanalytics.com/ news/googles-machine-learning-imaging-analytics-flag-breast-cancer

  • Motivation

    – 30 hours for 130 slides (4.3 slides/hr.)– 280 million slides / day (11,666,666/hr.)

    50/day = 15,342 years200/day = 3,836 years

    TIME

  • • Technology: includes 64 of IBM’s TrueNorthchips in a standard server shelf – capable of scaling to half a billion artificial

    neurons with its modern architecture• By 2020, will have a “10 billion neuron equal

    system that can fit in a two-liter box and need only 20 watts to operate, would provide a desktop operation with roughly 10% the computing ability of the human brain that could simply run on a smartphone battery”

    IBM and Watson

  • Looking ahead to ………………….. the future of health care

    Outperform human beings at a given task ?

  • The Real Threat of Artificial Intelligence

    • A.I. products that now exist are improving faster than most people realize and promise to radically transform our world….

    • They are only tools, not a competing form of intelligence….will reshape what work means …. it will eliminate many jobs…. radiologists will gradually be replaced by such software.

    KF Lee. The Real Threat of Artificial Intelligence. The New York Times; 6/24/17https://www.nytimes.com/2017/06/24/opinion/sunday/artificial-intelligence-economic-inequality.html?imm_mid=0f403e&cmp=em-data-na-na-newsltr_20170703

  • The Real Advantage of Artificial Intelligence

    • Relieve provider shortage• Offer more healthcare choices• Improve accuracy• Improve patient outcomes• Reduce costs

    KF Lee. The Real Threat of Artificial Intelligence. The New York Times; 6/24/17https://www.nytimes.com/2017/06/24/opinion/sunday/artificial-intelligence-economic-inequality.html?imm_mid=0f403e&cmp=em-data-na-na-newsltr_20170703

  • In its infancy…but potentially contributes immensely

    Think ahead 10, 20, 30 years!

  • Thank you!

    �Artificial Intelligence Challenges the Status Quo�Technology…EHRs…AutomationWhat is Artificial Intelligence (AI)?Artificial IntelligenceImitation Game 2014Turing TestTuring TestKeyword and PubMedTutorialDeep LearningNatural Language ProcessingSlide Number 12"Artificial intelligence is one of many leading-edge technologies that China seeks and that has potential military applications“ (Sen. John Cornyn aide)HAL 9000 (Heuristically programmed Algorithmic computerAmazonPersonal Assistants Google Brain Research�“Make machines intelligent. Improve people’s lives.”Google Brain ResearchGoogle’s self driving car drove 80 million miles before it ever touched a roadSlide Number 20Things described as using AI8 more things you didn't know �AI could do*Symptoms/conditions�*Recommended cannabinoid levels and strains�*Dispensary near you�AI in Health CareMycin System (1970s)Slide Number 26Microsoft Seeing AI AppNLPOpthalmologyGoogle DeepMindGoogle, Verily Life Sciences�Verily Science Watch�Google Brain, Verily Life SciencesMotivationSlide Number 35�Looking ahead to ………………….. � the future of health care�Slide Number 37Slide Number 38Slide Number 39The Real Threat of Artificial IntelligenceThe Real Advantage of Artificial IntelligenceIn its infancy…but potentially contributes immenselySlide Number 43