This document discusses the role of natural language processing (NLP), conversational AI, and virtual voice assistants in pediatrics. It begins with an introduction to NLP and how it allows computers to understand spoken and written human language. It then discusses several use cases for clinical NLP, including automation of workflows, analytics, prediction, and conversational agents. Examples of chatbots and virtual assistants currently used in healthcare are provided. The document outlines the current state of conversational AI and envisions future directions such as multimodal data fusion to create contextual AI, integration of CAI into physician workflows, and use of hybrid technologies combining CAI with augmented reality and robotics. It concludes that NLP can unlock insights from unstructured data, CAI provides
VIP Call Girls Noida Sia 9711199171 High Class Call Girl Near Me
Role of NLP, Conversational AI & Virtual Assistants in Pediatrics
1. Role of NLP, Conversational AI
&
Virtual Voice Assistants in Pediatrics
JAI NAHAR, MD, MBA
Associate Professor of Pediatrics
George Washington University School of Medicine
Attending, Division of Cardiology
Children’s National Hospital, Washington DC
AIMed Pediatrics
Nov 11th, 2020
4. Natural Language Processing (NLP)
• ThisAI methodology allows the
computer to understand spoken as
well as written human language
• NLP = NL Understanding (NLU)+ NL
Generation (NLG)
Intelligence-Based Medicine
1st Edition. August 2020
Artificial Intelligence and Human Cognition in Clinical Medicine and Healthcare
Author: Anthony Chang
Chang A. Analytics and Algorithms, Big Data, Cognitive Computing, and Deep
Learning in Medicine and Health Care. AI Med Ebook; 2017
Language
and
Speech:
Broca’s
and
Wernicke’s
areas
6. Clinical NLP: Use cases
MD Work Flow
Augmentation
Care Delivery
Revenue
Optimization
Research
Patient Portal
UX
Conversational
AI
• Automatic Speech recognition
• EHR documentation
• Chart Review
• Data Mining
• Predictive analytics
• Phenotypic Classification
• Risk Scores/stratification
• Clinical decision support
• Targeted Intervention
• Adverse event prevention
• Quality Improvement
• Computer assisted coding
• Automatic preauthorization
• No show prediction
• Clinical trial matching
• Automated Registry reporting
• Analytics
Understanding Portal’s
Medical Terminology
• Conversational agents
(Virtual Assistants, Chabots)
• Ambient Clinical Intelligence
• Voice Biomarker analysis
7. Clinical NLP and Care Delivery : Data Mining
Input Data
Unstructured
Data
EMR/Other
sources
Use of NLP
Structured
Data (Machine
Interpretable)
ML
Analytics
Targeted Early Intervention
Quality Improvement
Data Discovery Advanced Analytics
Phenotypic
Classification
Risk stratification
Clinical
decision
support
9. Voice,Voice Intelligence, Conversational AI
WhyTalk aboutVoice?
History: Human Computer Interaction
GUI
WEB
www
Mobile
VUI
1980s 1990s 2000s Present
Voice:
• Natural mode of
communication
• Convenient
• Scalable
• Rich Content
• Biomarkers
• Metadata
10. Conversational AI
Technology which allows Human Machine interaction through the use of
Natural conversation, utilizing voice user interface and Machine Intelligence
Human Machine
Conversational
AI
Voice
Technology
NLP
ML,
Deep
Learning
Synergistic
Convergence
Conversation
VUI
11. Current State
Conversational AI Ecosystem
Smart
Speakers
Smart
Displays
Smart voice
enabled
devices
Chatbots
Virtual
Assistants
Ambient
Clinical
Intelligence
Conversational
AI
Voice
VUI
IOT
Ambient
Sensors
12. DigitalTriage: Chatbots
Espinoza J, Crown K, Kulkarni O
A Guide to Chatbots for COVID-19 Screening at Pediatric Health Care Facilities
JMIR Public Health Surveill 2020;6(2):e18808
http://publichealth.jmir.org/2020/2/e18808/
Sample chatbot process map. *: Institutional discretion, follow public health
agency guidelines.
Sample Chatbot process map
13. Virtual Assistants
My Children’s Enhanced Recovery after surgery (ERAS)
Boston Children's Hospital
• Patient facing
• Post operative recovery update
• Post Op appointment information
https://www.voice.health/news/my-childrens-enhanced-recovery-after-
surgery-eras-alexa-skill
Task: Information
exchange
Technology: Alexa
Environment:
Patient’s home
14. Embodied Conversational Agents (ECA)
Embodied Conversational Agents
Edited by Justine Cassell, Joseph Sullivan, Scott Prevost and Elizabeth F. Churchill
March 2000
https://mitpress.mit.edu/books/embodied-conversational-agents
Embodied conversational agents are
computer-generated cartoonlike characters
that demonstrate many of the same
properties as humans in face-to-face
conversation, including the ability to
produce and respond to verbal and
nonverbal communication.
Examples of embodied conversational agent embodiments
Embodied Conversational Agents in Clinical Psychology: A Scoping Review
Provoost S, Lau HM, Ruwaard J, Riper H
J Med Internet Res 2017;19(5):e151
URL: https://www.jmir.org/2017/5/e151
15. Future Directions
• Multimodal data fusion Contextual AI
• Smart Physician Friendly EHRs (NLP and CAI)
• Advanced HCI: Children Friendly HumanoidVirtual Assistants
• HybridTechnology: CAI+AR; CAI+Robotics
16. Conclusion: KeyTakeaways
1. NLP: Unlocking Actionable Insights from Unstructured data
2. CAI: Intelligent User Interface
3. UX, Smooth Work Flow integration
4. Data Privacy, Security, is integral to successful implementation
Thanks to Anthony and AIMEd team for the invitation and honor to present
________________________________________________________________
With the advancements in Voice technology and Natural language processing, Conversational AI and Virtual Voice Assistants are gaining increasing attention in health care for developing provider, patient and enterprise facing solutions. This workshop will focus on Conversational AI, Virtual voice assistants, and cover the following key points ( next slide)
My talk today will geared towards application of NLP and CAI in clinical care, including some introductory concepts.
In My Talk today I will go over introductory concepts, then focus on applications of Clinical NLP, Conversational AI & Virtual Voice Assistants
in health care delivery, then wrap up future directions.
So what is NLP?
NLU: Comprehending what is being said
NLG: Generating a response in human like language format
Some of our Natural Language and speech functions which are performed by our
Brain’s Broca’s and Wernicke’s areas (left hemisphere), can be performed by Machine’s Natural Language
Processing.
In clinical world NLP can be used to achieve fgs important functions:
Automate: Data Extraction from EHR, Entry/Feed into Registry
Analyze -- RTAP-> Intelligence/Insights
Predict
Classify, Risk Stratify
Assist
Coach (Virtual Assistant)
Sense (ASR)
Express (NLG)
Building upon the NLP functions from previous slide let us look at the use cases of clinical NLP which are highlighted as discrete categories in this slide.
Other use cases are: --------
Physician Work flow augmentation: Speech recognition and EHR documentation
Care delivery: Clinical decision support, Risk stratification and predictive analytics
Revenue optimization: Computer assisted coding, Automatic preauthorization, No show prediction
Research: Automation of Registry data entry, Analytics, Registry reporting, Clinical trial matching,
Conversational AI: Conversational agents (Virtual agents, chatbots), Voice Biomarker analysis
Patient portal UX: NLP tools linking medical terms in portal documents to simple definitions and explanations for the patient, will help to improve patient’s EHR understanding and portal user experience
Let us look at use of NLP in care delivery.
Big challenge of current health care data is that it is trapped in the form of unstructured format.
First step shown in this slide is data discover: that is application of NLP tools to the unstructured data from various sources, including EHR thus making this data Machine interpretable.
This in turn would facilitate the Second step of advanced analytics using ML to facilitate ----------
Let us look at one real life application of NLP.
Dr Amir Kimia, and his team at BCH have utilized NLP based safety system for mining unstructured data from EHR to increase the capture of safety events.
They did a pilot project on IV infiltration events, which helped to capture new safety events which were not captured previously.
Similar automated and scalable safety surveillance systems NLP can be used to improve quality, outcomes and prompt detection of Adverse safety events.
Over past 3-4 decades our interaction with computers has progressed through the GUI, www revolution, mobile revolution including TUI and now Voice revolution, VUI and conversational AI.
The Era of VUI has great promise
as Voice is Natural mode of communication
Convenient
Scalable
Rich Content
Metadata
1989: World Wide Web invention
Jan 2007: First I phone released, beginning of smart phone era
2011 –present: Era of Conversational AI
Foundation of conversational AI is NLP.
Conversational AI: Human Machine interaction through the use of conversation, utilizing voice user interface and Machine Intelligence.
This is made possible by synergistic convergence of Voice technology, and Artificial Intelligence technology (Natural Language processing, Machine and Deep Learning).
It includes smart voice enabled devices, chat bots and virtual assistants
Two common applications of Conversational AI are virtual assistants and chatbots, these are also known as virtual or conversational agents.
A virtual assistant (e.g. Apple’s Siri, Google Assistant, and Amazon Alexa) is an AI-inspired software agent that is capable of performing certain tasks or services via text or voice.
A chat bot is a conversational AI application that is capable of providing information to the user. Virtual voice assistant helps the user in performing simple daily tasks
There are many ways in we encounter conversational AI in our daily lives. These are in form of smart speakers,
Smart displays next evolution of smart speakers and include Voice + touch screen Display : Amazon Echo show, Google home hub, Facebook portal
A chat bot is a conversational AI application that is capable of providing information to the user. Virtual voice assistant helps the user in performing simple daily tasks
ACI is a newer application of CAI aimed to decrease physician burnout.
Two common applications of Conversational AI are virtual assistants and chatbots, these are also known as virtual or conversational agents.
A virtual assistant (e.g. Apple’s Siri, Google Assistant, and Amazon Alexa) is an AI-inspired software agent that is capable of performing certain tasks or services via text or voice.
Let us look at one application of Conversational agents that is chat bots.
Chat bots can be used to fill the gaps in access to health care. This can be done by using them for patient facing Digital or E triage solutions.
Chat bots came very handy during COVID -19 pandemic.
Referenced on this slide is an excellent recent article on COVID-19 screening chat bots. This is from Omkar Kulkarni and his team from CHLA . The sample Chabot process map illustration shows how chatbot can be used to triage pt to appropriate type of medical care.
Similar solutions can be customized for other use cases for optimizing Virtual pediatric care delivery.
My Children's Enhanced Recovery After Surgery (ERAS) (by Boston Children's Hospital, a leading children's hospital): Parents and caregivers of children in the ERAS program at Boston Children's Hospital can provide their care teams updates on recovery progress and receive information regarding their post-op appointments.
This skill, developed by Boston Children’s Hospital, is part of the Enhanced Recovery After Surgery (ERAS) program, which is designed to help parents and caregivers of children who have recently undergone heart surgery. In its first version, the skill will support parents and caregivers post-discharge, allowing them to provide quick updates to their care team around recovery progress (including pain and activity level). Additionally, Alexa can provide parents and caregivers information regarding their scheduled post-op appointments.
Embodied CAI agents are special type of conversational agents, which are computer generated carton like characters, and can be designed to communicate with our pediatric patients in more natural way, using speech, facial expressions, and body language (hand gestures, and body stance).
In the figure on bottom-right: is humanoid robot KASPAR.
They constitute a type of (a) multimodal interface where the modalities are those natural to human conversation: speech, facial displays, hand gestures, and body stance;
Examples of embodied conversational agent embodiments: top-left: emotional reinforcement with a smiley face; bottom-left: virtual psychiatric nurse; middle: SPARX’s (Smart, Positive, Active, Realistic, X-factor thoughts) guide character;
top-right: SimSensei Kiosk virtual counselor; bottom-right: humanoid robot KASPAR.
Embodied CAI agents are special type of conversational agents, which are computer generated carton like characters, and can be designed to communicate with our pediatric patients in more natural way, using speech, facial expressions, hand gestures, and body stance.
In the figure on bottom-right: is humanoid robot KASPAR.
In future consideration should be given to
1. Use of NP in multimodal data fusion to develop contextual AI to help in better prediction and prescription.
2. Development of Smart EHR using NLP tools and CAI
3. Development of Advanced HCI solutions such as Children Friendly Humanoid Virtual Assistants
with Cognition & Empathy.
4. Hybrid tech (integration of CAI+AR to form intelligent reality) , CAI +Robotics can be used to design children Friendly applications
In conclusion NLP is an important AI Technology to unlock Actionable Insights from Unstructured data
CAI can be used as Intelligent user interface for children and providers in augmenting care delivery
Good UX enabled by Design thinking, Smooth workflow integration, Data privacy and security are integral to --------