CHOC PEDIATRICS COMMUNITY OF PRACTICE SERVICES DEVELOPMENT PROPOSAL
CLOUD COMPUTING & BIG DATA
Author: Shirley Tseng
Status: Project Concept
Funding Acknowledgment: None but author labor is available to work on the project.
Proposal to build a Pediatrics Community of Practice KnowledgeBase (CoP KB) around CHOC M1/M2 and AI-Med content using open source tools/platform.
Use open source tools to publish content from monthly CHOC Innovation presentations and annual AI-Med content to build persistent knowledge base that allows CoP KB Members to find/discover resource, links to additional information, annotate with personal comments, as well as share to their friends.
Monthly CHOC M1/M2 Powerpoints would be available in Cop KB, but the information on each slide will also be dis-aggregated and compiled into respective individual categories to enable community members to learn more about and internalize the material. For example, ‘book of the month’ shared each month would be added to a ‘book of the month’ page on the CoP KB with the all of the books shared to date. The ‘book of the month’ page would have information shared that month as well as provide links to web resources – Amazon to buy book, links to published reviews, and allow CHOC CoP members to make their own review/annotations about the book. CHOC CoP members can share their comments about Nudge as they read the book or lessons learned as they practice the Nudge recommendations.
Companies and new technologies shared at CHOC meetings or AI-Med conference can be extracted by the content ingest tools and appropriate CoP KB data elements created.
The goal is to build Cop KB with simple interfaces for CoP members to use, find/discover information, write/publish their comments/information, and share with their friends. Underlying infrastructure with rich tools would hide the details from user.
The CoP KB will be built using open source big data and cloud-based tools and with UCI/CHOC students/interns. Ingest tools can be integrated into the system to read the Powerpoint text into CoP KB data elements with use of pre-defined AI pediatric-based data classifier and ontologies that would properly read in monthly Powerpoints and ‘classify’ each information nuggets in the Powerpoint pages into cumulative CoP KB data elements, labels, and links to content. Data presentation tools can be integrated to enable dynamic content creation using labels or links. CoP DB will have relevant labels, ontology graphs or schemas that would enable dynamic data assembly with user queries or queries via rich ontologies.
If video or audio data is available, open source ingest tools to convert video/audio data to text can be integrated to add those context, with the proper labeling to enable fine-point retrieval of key segments of the data.
The CoP KB information can be organized under multiple views points to enable multiple ways to retrieve the information. For example, question viewpoint would extract the relevant ‘question’ the content ask or answers. AI-based algorithms can be used to detect where the ‘questions’ are asked in the content (Powerpoint, paper, video, or audio). The question ontology CoP KB metadata layer could have pre-define Pediatric questions and new questions detected by the algorithm could be added to the metadata layer for rich semantic layer powering the platform.