Successfully reported this slideshow.

k-BOT: Knowledge-driven Chatbot for Health @ CASY2020

0

Share

1 of 9
1 of 9

More Related Content

Similar to k-BOT: Knowledge-driven Chatbot for Health @ CASY2020

Related Books

Free with a 14 day trial from Scribd

See all

Related Audiobooks

Free with a 14 day trial from Scribd

See all

k-BOT: Knowledge-driven Chatbot for Health @ CASY2020

  1. 1. Collaborative Assistants for the Society (CASY 2020) k-BOT: Knowledge-driven Chatbot for Health Hong Yung (Joey) Yip PhD Student @ Artificial Intelligence Institute, UofSC (AIISC) October 16th, 2020 @ University of South Carolina, Columbia, SC, USA
  2. 2. Paradigm Shift in Healthcare ● Episodic to Continuous Monitoring ● Clinician-centric to Patient-centric ● Clinician controlled to Patient-empowered ● Disease Focused to Wellness-focused ● Sparse data to Multimodal Big Data Medical Internet of Things 2 Images from: https://thenounproject.com
  3. 3. Augmented Personalised Health (APH) From Big Data to Smart Data Augmented Personalised Health (APH) is a vision to enhance the healthcare by using AI techniques on semantically integrated Patient-Generated Health Data (PGHD), environmental, clinical, public health & social data. Data Components PGHD, Clinical, Environmental, and Social Data Smart Data Meaningful data after contextualised processing 3 http://wiki.aiisc.ai/index.php/Augmented_Personalized_Health:_How_Smart_Data_with_IoTs_and_AI_is_about_to_Change_Healthcare Images from: https://thenounproject.com
  4. 4. kBOT for kHealth Asthma Highly Diverse Data up to 29 parameters (& collection methods: Active + Passive): Up to 1852 data points/ patient /day kBot with screen interface for conversation Text Speech ● Smarter & engaging agent ● Minimize active sensing (Questions to be asked) ● Ask only informed & intelligent questions ● Relevant & Contextualized conversations ● Personalized & Human-Like 4 https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7432964/
  5. 5. kBOT Demo 5https://youtu.be/oZCaM7rhJAg More on: https://www.slideshare.net/aiisc/kbot-knowledgeenabled-personalized-chatbot-for-selfmanagement-of-asthma-in-pediatric-population
  6. 6. Other Areas of Work & Takeaways WHY ● Obesity is a common lifestyle disease worldwide ● Number of calories contained in food is not always intuitive ● Minimize obesity-induced hospitalization Our approach ● Integrate food knowledge (domain-specific and personalized user- centric) and their nutritional values to augment diet planning for personalized health ● Support multimodal interactions (text, image, speech) For Nutrition Tracking and Diet Monitoring Modeling Social Behavior for Healthcare Utilization in Depression WHY ● Leading cause of disability worldwide ● $40 billion has been spent each year on depression treatment ● Effectiveness of treatment varies between different individuals Our approach ● Integrate MedDRA, Drug Abuse Ontology (DAO), SIDER ● Monitor and keep track of personal well-being and depressive symptoms ● Deliver personalized and efficient behavioral or medical interventions Mental Health (http://wiki.aiisc.ai/index.php/KHealth_Chatbots)Nutrition (http://wiki.aiisc.ai/index.php/Nourich) Conversation systems are making it easy for users (patients) to interact with and use technology. Three most important features are 1. Personalization: It is all about you, it knows your history (not generic) 2. Contextualization: It understands your unique situation (domain knowledge) 3. Secure and private. 6 Acknowledgement: Dipesh Kadariya, Revathy Venkataramanan, Maninder Kalra, Krishnaprasad Thirunarayanan, and Amit Sheth
  7. 7. Contextualization and Personalization kBOT initiates greeting conversation. Understands the patient’s health condition (allergic reaction to high ragweed pollen level) via the personalized patient’s knowledge graph generated from EMR, PGHD, and prior interactions with the kBOT. Generates predictions or recommended course of actions. Inference based on patient’s historical records and background health knowledge graph containing contextualized (domain-specific) knowledge. Figure: Example kBOT conversation which utilizes background health knowledge graph and patient’s knowledge graph to infer and generate recommendation to patients. ★ Conversing only information relevant to the patient Context enabled by relevant healthcare knowledge including clinical protocols. On-going Work (Snapshot) 7
  8. 8. 8 ONE SLIDE TO SHOW HOW PHKG EVOLVES OVER TIME kHealth Ontology API APH kHealth Project (Pediatric Asthma) Reasoning mechanisms Enriching KG Enriching KG In-built rule-based inference engine Machine Learning Updating the KG with more triples Analyzing datasets Executing reasoning Ontology Catalogs: ● BioPortal ● Linked Open Vocabularies (LOV) ● Linked Open Vocabularies for Internet of Things (LOV4IoT) Linked Open Data (LOD): ● UMLS ● SNOMED-CT ● ICD-10 ● Clinical Trials ● Sider Personalized Health Knowledge Graph (PHKG) Personal Sensor Data Electronic Medical Records (EMR) Figure: Personalized Health Knowledge Graph (Amelie et. al, 2019) https://scholarcommons.sc.edu/aii_fac_pub/42/ Personalized Health Knowledge Graph
  9. 9. Sample Ontology 9 Asthma Ontology (In-house)

Editor's Notes

  • Our unique way (sensors)
    Use knowledge to personalize
    Convey we have ontology

    Talk through it (mental health, nutrition, etc)
    Using APH technology to make personalization solutions
    Asthma as instance

    Asthma
    Multifactorial disease
    6.3 million children in USA are affected
    300 million adults & children worldwide [CDC]
    Difficult to diagnose based on episodic visits and clinical records
    Non-adherence to medication makes it one of the poorly controlled disease
    Chatbot could play a pivotal role throughout the unfolding data & knowledge-driven, AI-supported ecosystem for ENHANCED HEALTH
  • TedX script
    Narrate the reasons
    Include evaluations narrative
  • Show critical aspects (develop a prototype, medically relevant, ~ clinical partners evaluated)
  • Time permitting
  • ×