ÎÞÓÇ´«Ã½

Global public health crises are redefining what national security entails. The COVID-19 pandemic has had vast domestic and global security impacts spanning health infrastructure, pandemic preparedness, and economic resilience. In addition, COVID-19 cases, hospitalizations, and deaths continue to .

These threats underscore the requirement for new solutions to maintain and improve public health in ways that strengthen U.S. national security. Where will this innovation come from? As the makes clear, artificial intelligence (AI) can provide many of the advanced technology tools public health organizations now need.

Greater availability of data has accelerated the application of AI during the COVID-19 pandemic. AI has enabled organizations to:

• Predict the spread and escalation of outbreaks
• Optimize allocation of healthcare resources
• Accelerate novel therapeutic research
• Automate administrative functions

In addition, the international public health community has collaborated across borders to make viral genomic sequencing widely available to power AI. Domestic consortiums are being led by the Centers for Disease Control and Prevention (e.g., ) and the National Institutes of Health (e.g., and other National Center for Biotechnology Information databases). Global consortiums include the (GISAID) and .

Despite this remarkable progress in leveraging AI for public health, several operational challenges remain. These include disparate procedures for data collection, standardization, and curation; limited integration and linkage of genomic, clinical, and epidemiological data; and a lack of diverse technical skillsets in the public health community. The COVID-19 pandemic continues to highlight the critical importance of well-curated data, cross-trained experts, and accessible technological infrastructure for a robust .

In the context of global public health and national security, the cost of inaction is steep. For example, the COVID-19 pandemic has caused an , challenging the efficacy of and trust in public health authorities. We propose the following recommendations to enable the broader application of AI to strengthen public health efforts:

1.ÌýDevelop Flexible AI Knowledge Assessment and Training. Personalized data science training, which emphasizes analytics skills and knowledge needed for specific roles, will ensure cross-pollination of traditional public health with cutting-edge AI technologies. This approach, in turn, will expand the utility and application of novel analytics in public health. Events such as cross-cutting symposia, lunch seminars, online training/forums, and hackathons can provide employees with opportunities for continuous training and de-silo experts within the workforce.

• ÎÞÓÇ´«Ã½ has collaborated with General Assembly and NVIDIA to offer our people a wide breadth of training, from the fundamentals of data science in our Data Science 5K program to more advanced training in deep learning. We also partnered with a federal health agency to build a more analytics-savvy workforce to support data-driven research and health.

2.ÌýIntegrate Public and Private, Civilian and Defense Data Sources. The integration and linkage of various data types (e.g., clinical, epidemiological, genomic, environmental) are crucial for improving outbreak forecasting and response, as well as answering many open scientific questions about emerging pathogens such as SARS-CoV-2. By creating secure, easily accessible repositories with fully transparent and programmatic access to data and metadata, public health analysts can more rapidly extract actionable knowledge from heterogeneous data.

• ÎÞÓÇ´«Ã½ has partnered with the U.S. Department of Defense (DOD) to use EpiMaps, an interactive analytics platform. The platform pulls multiple sources of public data and can accommodate proprietary data to produce county-level COVID-19 prediction models. DOD uses these models to inform its pandemic response.

3. Equip AI Teams with Wide-Ranging ÎÞÓÇ´«Ã½.ÌýTo ensure that AI solutions target specific public health applications, experts from public health, epidemiology, and clinical communities must be engaged. An AI Operations (AIOps) framework can help establish AI solutions that are scalable, secure, reproducible, testable, transparent, and monitorable. AI solutions require an implementation life cycle that can serve as a framework to .

• ÎÞÓÇ´«Ã½ has teamed with government partners to deploy AIOps for mission-critical efforts. We have also brought together mathematical modelers, public health experts, and AI experts in the fight against COVID-19 to produce a Safe Return Simulator. This tool helps clients and partners use an evidence-based strategy to normalize operations.

4. Automate and Integrate AI into Existing Workflows. Increased access to shareable pipelines and workflows that support reproducible data ingestion and processing can help decrease time-to-insight on large amounts of surveillance data. In addition, integrated tools for data governance and provenance allow an organization to standardize data ingestion, processing, and storage, improving transparency and consistency.

Together, these recommendations can help military and civilian public health authorities collaborate to increase data transparency, trust in AI, and scientific reproducibility. These and other improvements will significantly increase the nation’s pandemic preparedness and readiness to combat emerging pathogens, strengthening both public health and national security.