Using Artificial Intelligence Technology for Social Determinants and Risk Factors Surveillance
Bibliographic record
Abstract
Session topic areaData science methods: machine learning in risk factor surveillance Overall objectives or goalBackgroundDecades of research have shown that factors such as living conditions, and not just medical treatments and lifestyles, are strongly associated with the health of individuals and populations. These distal factors (social, economic, cultural and environmental) are collectively called the social determinants of health (SDOH), and affect health inequities (i.e. differences in health outcomes that are avoidable, unfair and unjust). Gathering data on both risk factors (biomedical/clinical) and SDOH is of the utmost importance to quantify their contribution in disease causation at individual and population levels. Social determinants of health and biomedical/clinical risk factors surveillance (collectively termed as “risk factor surveillance”) refers to the monitoring of distal and proximal factors that impact the health of individuals and populations and health equity. It offers the opportunity to “forecast” population health, potential disease incidence, and guide intervention programs to prevent disease manifestation. However, current risk factor surveillance data is limited in geographical representation, completion, and content and time. Identifying novel methods of collecting risk factors and SDOH data can allow for opportunities for population health and disease forecasting using high quality, nationally-representative, real-time data. Recent breakthroughs in artificial intelligence (AI), such as speech and image recognition, offers new opportunities to develop novel methods to collect risk factor information at individual levels. Meanwhile, we can use intelligent computer systems to process vast amount of data and turn those data into actionable information and knowledge for improving population health. Collaborative Session ObjectiveThrough a CIHR-funded project, we are assembling a team of national and international experts including stakeholders, public health officers/physicians, and researchers, who will identify key gaps in risk factor surveillance and data collection technologies. Resulting projects will focus on using AI for risk factor surveillance, for the ultimate purpose of monitoring population health, guiding intervention programs, and preventing disease. Our projects will focus on discovering and refining innovative methods in data collection, management, as well as assessment of data quality (i.e. selection bias). We will engage scientists and knowledge users from the inception of the ideas to ensure the relevancy of the final projects. This project aims to link medical records, clinical information, and SDOH data, to alter the way we conduct surveillance and work with big data. Facilitators involved; home institutions Dr. Vineet Saini, University of Calgary; Alberta Health Services Dr. Mingkai Peng, University of Calgary Dr. Hude Quan, University of Calgary; World Health Organization Collaborating Centre for Classification, Measurement and Standardization Intended output or outcome Identify AI technologies for use in risk factor surveillance; innovative methods in data collection, management, as well as assessment of data quality (i.e. selection bias); uses for new data sources in improving health equity. Create partnerships between national and international experts in risk factor surveillance
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.007 | 0.021 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.004 | 0.003 |
| Science and technology studies | 0.001 | 0.001 |
| Scholarly communication | 0.004 | 0.003 |
| Open science | 0.001 | 0.002 |
| Research integrity | 0.002 | 0.002 |
| Insufficient payload (model declined to judge) | 0.004 | 0.001 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".