#himss22

The Department of Veterans Affairs (VA) talked about its “AI Tech Sprints” with outside collaborators. Since 2019, teams “are provided access to large, unique data sources (e.g., synthetic and anonymized patient data, clinical trials, patents, experimental therapeutics); Veteran and Expert perspectives; technical AI/ML, demo and user-based feedback; and longer term partnership and/or funding opportunities.”

The Connected Health Initiative (CHI) is proposing “good machine learning practices” to provide a framework that the Food and Drug Administration (and other governmental and non-governmental entities) can “leverage to responsibly deploy AI while protecting patient safety”. Great presentation overall.

Read Also

HIMSS22: Day 1 highlights

HIMSS22: Day 2 highlights

Members of the Office of the National Coordinator for Health Information Technology (ONC) provided an interactive session on ongoing work to advance the health IT agenda in the United States. The panelists emphasized the importance of data standardization.

I met with representatives of H2O.ai and Olive, two vendors that believe there should be more emphasis on nursing informatics, notably through the use of AI-centric frameworks.

Read Also

HIMSS22: Day 1 highlights

HIMSS22: Day 3 highlights

There are still several challenges with AI applied to healthcare: algorithmic bias, lack of data representativeness, non generalizability, inequities in deployment, amplification of discrimination, dual use and other unintended consequences to consider. Dark skinned tones need to be better integrated into AI frameworks.

80% of the data collected in healthcare organizations is either unstructured or conversational yet meaningful insights are hard to come by.

Three key areas of GeoAI: clustering, prediction and image classification.

Ricki Fairley gave a touching presentation on the state of Black breast cancer. Treatment and survival needs to be understood.

Read Also

HIMSS22: Day 2 highlights

HIMSS22: Day 3 highlights