Space Biology RESEARCH FOR HUMAN EXPLORATION

Transcription

1 Space Biology RESEARCH FOR HUMAN EXPLORATION TRISH Artificial Intelligence Workshop California Institute of Technology, Pasadena July 31, 2018 Elizabeth Keller, Space Biology Science Manager 1

2 Content Analysis of Space Biology Research: Identifying Knowledge Gaps to Inform Strategic Planning

3 SPACE BIOLOGY RESEARCH SUPPORTS HUMAN EXPLORATION Understanding how life responds and adapts to space Develop new technologies Help humans explore further into space Benefits for Humanity on Earth

4 To pursue safe human space exploration Space Biology investigates five elements: Plants Cells and Molecules Animals Microbes Developing Organisms

5

6 Decadal Survey: Research analysis began with the mid-term review and assessment of NASA s progress and accomplishments against the Decadal Survey recommendations

7 Decadal Survey: Assessing Research Progress Determining performance metrics: grants and publications Collecting Space Biology-funded research publications (citation and study metadata) Extracting grant data from the NASA Task Book Reviewing data and assigning one or more Decadal survey (DS) recommendation categories Tallying totals for each DS category providing quantitative metrics for progress in each area

8 Decadal Survey Assessment Limitations Categories are too broad (P2) or too narrow (AH7) Research capabilities are rapidly changing and evolving New discoveries are being made and potential health issues continue to be identified Space Biology needs to align with internal NASA programs, external agencies and institutions Must demonstrate the value of fundamental biological research to advancing space exploration Metadata on progress doesn t answer the question, What have we learned?

9 Strategic Planning: Where We re Going 1.2 Understand responses of physical and biological systems to spaceflight The first stages of progress toward achieving this strategic objective will be clearly measured by the formulation of agreements between the research programs on the one hand and the internal NASA customer (for enabling exploration) or external organizations (for scientific discovery) on the other. Such agreements will specify what research questions will be addressed by the NASA research programs and may include schedules. Subsequent progress will be measured by the accomplishment of intermediate milestones in the research program. Final accomplishment of the research objectives will be measured by showing how the research products address the original agreement s needs.

10 Tactical Planning: How We re Getting There Taking a deeper dive into the scientific accomplishments in space life sciences Conduct a systematic review of scientific knowledge to date Align/reconcile current knowledge with guiding questions and specific aims to identify gaps Ingest and analyze the contents of journal articles You Are Here Prepare content maps, evidence reports, and create a Research Roadmap to plan future research and track ongoing progress

11 Tactical Planning: How We re Getting There

12 Why Artificial Intelligence? Broad scope of Space Biology Structure of journal articles Scanning & indexing scientific papers Pace and volume of scientific research

13 What would Science AI look like? INPUTS: Scientific papers in digital form NCBI GeneLab JGI - Title, keywords, abstract, funding source(s), citation data, methods, procedures (assays used), independent variable(s), dependent variables (measures), statistical method, analytical results, accession number and data repository, references) Data across multiple open access repositories

14 OUTPUTS: What would Science AI look like? Initiate a scientific query in the form of an hypothesis Perform meta-analysis across publications for all sources/journals Reports: - Collapse across samples, assay types, experimental subject types, independent variable types (e.g. treatments, manipulations) - Show the traceability of specific findings and data sources - Provide a score, or set of weighted scores, that would show; - Strength of effect - Number and quality of source studies - Quantitative scores on experimental samples, subjects - Ingest and analyze qualitative data (e.g. changes phenotype/morphology)

15 What would Science AI look like? OUTPUTS: Visualizations to show the relationships amongst the key findings and lines of inquiry (e.g. cell signaling pathways and networks, proteins, transcripts)

16 VOSViewer

17