Exploring bias and prediction metrics to characterise the fairness of machine learning for equity-centered public health decision-making: a narrative review
Open accessAbstract
The rapid advancement of Machine Learning (ML) represents novel opportunities to enhance public health research, surveillance, and decision-making. However, there is a lack of comprehensive understanding of algorithmic bias — systematic errors in predicted population health outcomes — resulting from the public health application of ML. The objective of this narrative review is to explore the types of bias generated by ML and quantitative metrics to assess these biases. We performed search on PubMed, MEDLINE, IEEE (Institute of Electrical and Electronics Engineers), ACM (Association for Computing Machinery) Digital Library, Science Direct, and Springer Nature. We used keywords to identify studies describing types of bias and metrics to measure these in the domain of ML and public and population health published in English between 2008 and 2023, inclusive. A total of 72 articles met the inclusion criteria. Our review identified the commonly described types of bias and quantitative metrics to assess these biases from an equity perspective. The review will help formalize the evaluation framework for ML on public health from an equity perspective.
Integrity signals
Registry
Cites
Statistics
Claims the paper makes
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
Without diverse representative and contextually relevant training data, agentic AI may generate inaccurate, inequitable, or unsafe recommendations for marginalized populations.
“while agentic AI may generate inaccurate, inequitable, or unsafe recommendations for marginalized populations.5,21,22”
Algorithmic bias and inequitable system behavior are issues in AI-enabled care.
“Equally concerning is the issue of algorithmic bias and inequitable system behavior.”
Without diverse representative and contextually relevant training data, empathic AI systems may misinterpret emotional cues across different cultural, linguistic or socioeconomic groups.
“Without diverse representative and contextually relevant training data, empathic AI systems may misinterpret emotional cues across different cultural, linguistic or socioeconomic groups”
Cites
Dissecting racial bias in an algorithm used to manage the health of populations
Obermeyer, Z., Powers, B., Vogeli, C., & Mullainathan, S. (2019) · Science · doi:10.1126/science.aax2342 · Cited by 7,137 · cited 3× · see citing works
“These biases can result in systematic errors in data processing and decision-making, which affects the model fairness and effectiveness across various scenarios and populations [51], [74], [65].”
Dwivedi, Y. K., Hughes, L., Ismagilova, E., Aarts, G., Coombs, C., Crick, T., Duan, Y., Dwivedi, R., Edwards, J., Eirug, A., et al. (2021) · International Journal of Information Management · doi:10.1016/j.ijinfomgt.2019.08.002 · Cited by 4,611 · cited 2× · see citing works
“This lack of cultural sensitivity can lead to inappropriate, ineffective, or even harmful health recommendations and diagnoses [28], [87].”
The Balanced Accuracy and Its Posterior Distribution
Brodersen, K. H., Ong, C. S., Stephan, K. E., & Buhmann, J. M. (2010) · 2010 20th International Conference on Pattern Recognition · doi:10.1109/icpr.2010.764 · Cited by 1,816 · cited 1× · see citing works
“…identify true negatives) in models, especially when dealing with imbalanced datasets or when the cost of different types of errors is similar [13]. an equilibrium between the rates of false positives (incorrectly predicting a positive outcome when it is negative) and false negatives (failing…”
Health disparities and health equity: Concepts and measurement
Braveman, P. (2006) · Annual Review of Public Health · doi:10.1146/annurev.publhealth.27.021405.102103 · Cited by 1,570 · see citing works
Data preprocessing techniques for classification without discrimination
Kamiran, F., & Calders, T. (2012) · Knowledge and Information Systems · doi:10.1007/s10115-011-0463-8 · Cited by 1,305 · see citing works
Fairness definitions explained
Verma, S., & Rubin, J. (2018) · Proceedings of the International Workshop on Software Fairness · doi:10.1145/3194770.3194776 · Cited by 1,124 · see citing works
Disparities in Health Care and the Digital Divide
Saeed, S. A., & Masters, R. M. (2021) · Current Psychiatry Reports · doi:10.1007/s11920-021-01274-4 · Cited by 578 · cited 2× · see citing works
“This divide can be due to various factors, including but not limited to socio-economic status, geographical location, age, and education level [93], [15].”
Drawbacks of artificial intelligence and their potential solutions in the healthcare sector.
khan, B., Fatima, H., Qureshi, A., Kumar, S., Hanan, A., Hussain, J., & Abdullah, S. (2023) · Biomedical Materials & Devices · doi:10.1007/s44174-023-00063-2 · Cited by 461 · see citing works
Selection bias and information bias in clinical research
Tripepi, G., Jager, K. J., Dekker, F. W., & Zoccali, C. (2010) · Nephron Clinical Practice · doi:10.1159/000312871 · Cited by 425 · see citing works
Advancing health equity with artificial intelligence
Thomasian, N. M., Eickhoff, C., & Adashi, E. Y. (2021) · Journal of Public Health Policy · doi:10.1057/s41271-021-00319-5 · Cited by 117 · see citing works
Straw, I. (2020) · Artificial Intelligence in Medicine · doi:10.1016/j.artmed.2020.101965 · Cited by 86 · see citing works
Flagged first, then by citations
Cited by
Sorted by citations
Girdwood, T., Kheirinejad, S., Kheirkhah, P., White, B., Davis, R., Schwartz, D., & Shaban-Nejad, A. (2026) · ESMO Real World Data and Digital Oncology · doi:10.1016/j.esmorw.2026.100694 · Cited by 3 · see citing works
Self-citation
- Authors
- Raza, S., Shaban-Nejad, A., Dolatabadi, E., & Mamiya, H.
- Affiliations
- Vector Institute (CA)
- Department of Pediatrics, College of Medicine, The University of Tennessee Health Science Center, Memphis, TN, USA
- York University (CA)
- McGill University (CA)
- Venue
- IEEE Access
- Published
- 2024
- Volume
- 12
- Pages
- 180815-180829
- ISSN
- 2169-3536 (ISSN-L)
- OpenAlex
- W4404848720
- Cited by
- 25
- References
- 11
- Citations by year
- 2026: 18 · 2025: 7
- FWCI
- 1.91
- Type
- journal article
- Language
- en
- Open access
- gold
- doi.org
- Locations
- IEEE Access (publishedVersion, cc-by)
- DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
- Funders
- Canadian Institutes of Health Research’s Institute of Health Services and Policy Research
- Equitable AI and Public Health Cohort, and Public Health Ontario
- Topics
- Social Sciences › Safety Research › Ethics and Social Impacts of AI
- Health Sciences › Medicine › Health Informatics › Artificial Intelligence in Healthcare and Education
- Health Sciences › Health Professions › Health Information Management › Artificial Intelligence in Healthcare
- Keywords
- algorithmic bias, algorithmic fairness, fairness metrics, health equity, public health decision-making, population health outcomes, health disparities
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
