Predicting Quality Measure Completion Among 14 Million Low‐Income Patients Enrolled in Medicaid
Open accessAbstract
Low-income populations have disproportionately low completion of recommended healthcare services, from missed vaccinations to cancer screenings. While machine learning models help identify high-risk patients for targeted treatment, they have rarely been evaluated for quality measure gap completion-or among low-income populations underrepresented in typical datasets. Analyzing 14.2 million Medicaid recipients-including those excluded from electronic health records and without prior utilization-we developed models to predict gaps in nine nationally adopted quality measures, including preventive care and chronic disease management. Using clinical data to prioritize outreach, the clinical-only model improved accuracy by 32.5 percentage points (pp) over non-predictive methods such as alphabetical calling or birthday reminders (AUROC: 0.88, F1-score: 0.69). Incorporating social determinants of health data further improved performance by 2.0pp in accuracy (to 84.5%) and increased F1-score by 5.0pp (to 0.74), with no change in AUROC (area under the receiver operating characteristic curve). Compared to the clinical-only model, the SDoH model also reduced pre-existing Black-White disparities in prediction accuracy. Model performance was especially sensitive to SDoH factors like healthcare workforce and facility availability.
The claims and statistics of this paper haven’t been checked yet.
Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.
Integrity signals
Registry
Cites
Statistics
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.
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?
ML-guided outreach improved quality gap closure efficiency in Medicaid populations, outperforming non-predictive outreach strategies by 32.5 percentage points.
“outperforming non‐predictive outreach strategies by 32.5 percentage points”
From Artificial Intelligence and Machine Learning for Identifying Social Determinants of Health in Low‐Income Populations Within United States Health Systems: A Scoping Review (2026) · cites this paper as [22]
The Results (Machine Learning Model Performance; page unavailable) report that the clinical-only model improved prediction accuracy by 32.5 percentage points over a baseline simulating non-predictive outreach. That figure is not a measured improvement in quality-gap closure or outreach efficiency: the paper reports those outcomes separately as simulations and says actual intervention effects require validation. (LLM, conf=0.88)
Experimental — accuracy not yet measuredThe two studies achieving low PROBAST risk of bias, Patel et al. [22] and Gregory et al. [27], both incorporated multi-site or national data and applied external validation procedures.
“The two studies achieving low PROBAST risk of bias, Patel et al. [22] and Gregory et al. [27], both incorporated multi‐site or national data and applied external validation procedures.”
From Artificial Intelligence and Machine Learning for Identifying Social Determinants of Health in Low‐Income Populations Within United States Health Systems: A Scoping Review (2026) · cites this paper as [22]
Methods > Data Sources and Study Oversight reports data from 25 states and Washington, D.C., supporting the multi-site component. Methods > Model Development and Comparison describes a single training/validation/test split, with the test set reserved for final evaluation, rather than external validation. The low-PROBAST rating is the citing authors’ assessment, not a rating established by this source. (LLM, conf=0.88)
Experimental — accuracy not yet measuredPredictive ML modeling was the most common approach, used in 10 studies.
“Predictive ML modeling was the most common approach, used in 10 studies [21, 22, 23, 25, 26, 27, 30, 31, 32, 34].”
SDOH data further improved model performance.
“AUROC 0.88, F1 0.69; ML outperformed non‐predictive outreach by 32.5 percentage points; SDOH data further improved performance”
The ML model predicting quality measure completion among 14 million Medicaid patients had an AUROC of 0.88 and F1 of 0.69.
“AUROC 0.88, F1 0.69; ML outperformed non‐predictive outreach by 32.5 percentage points; SDOH data further improved performance”
ML outperformed non-predictive outreach by 32.5 percentage points.
“AUROC 0.88, F1 0.69; ML outperformed non‐predictive outreach by 32.5 percentage points; SDOH data further improved performance”
Patel et al. [22] reported an AUROC of 0.88 and F1 score of 0.69 across nine quality measure gaps in 14.2 million Medicaid recipients.
“Patel et al. [22] reported the highest performance, with an AUROC of 0.88 and F1 score of 0.69 across nine quality measure gaps in 14.2 million Medicaid recipients”
Two studies achieved low overall PROBAST risk of bias: Patel et al. [22] and Gregory et al. [27].
“Two studies achieved low overall PROBAST risk of bias: Patel et al. [22] and Gregory et al. [27].”
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Cites
XGBoost: A Scalable Tree Boosting System
Chen, T., & Guestrin, C. (2016) · Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining · doi:10.1145/2939672.2939785 · Cited by 54,569 · see citing works
Collins, G. S., Reitsma, J. B., Altman, D. G., & Moons, K. G. M. (2015) · Annals of Internal Medicine · doi:10.7326/m14-0697 · Cited by 3,267 · see citing works
(2005) · doi:10.1093/acprof:oso/9780198565895.001.0001 · Cited by 2,155 · see citing works
Chicco, D., & Jurman, G. (2023) · BioData Mining · doi:10.1186/s13040-023-00322-4 · Cited by 667 · see citing works
Guy, R., Hocking, J. S., Wand, H. C., Stott, S., Ali, H., & Kaldor, J. M. (2011) · Health Services Research · doi:10.1111/j.1475-6773.2011.01342.x · Cited by 332 · see citing works
Hyperparameter Importance Across Datasets
van Rijn, J. N., & Hutter, F. (2018) · ACM SIGKDD Conference on Knowledge Discovery and Data Mining (KDD) · doi:10.1145/3219819.3220058 · Cited by 272 · see citing works
Deaths Preventable in the U.S. by Improvements in Use of Clinical Preventive Services
Farley, T. A., Dalal, M. R., Mostashari, F., & Frieden, T. R. (2010) · American Journal of Preventive Medicine · doi:10.1016/j.amepre.2010.02.016 · Cited by 251 · see citing works
Evaluation and Mitigation of Racial Bias in Clinical Machine Learning Models: Scoping Review
Huang, J., Galal, G., Etemadi, M., & Vaidyanathan, M. (2022) · JMIR Medical Informatics · doi:10.2196/36388 · Cited by 186 · see citing works
Njoku, A. U., Evans, M. A., Nimo-Sefah, L., & Bailey, J. (2023) · Healthcare · doi:10.3390/healthcare11030438 · Cited by 174 · see citing works
Informative missingness in electronic health record systems: the curse of knowing
Groenwold, R. (2020) · Diagnostic and Prognostic Research · doi:10.1186/s41512-020-00077-0 · Cited by 139 · see citing works
Zhou, F., Jatlaoui, T. C., Leidner, A. J., Carter, R. J., Dong, X., Santoli, J. M., Stokley, S. K., Daskalakis, D., & Peacock, G. (2024) · MMWR Morbidity and Mortality Weekly Report · doi:10.15585/mmwr.mm7331a2 · Cited by 123 · see citing works
Inoue, T., Ichikawa, D., Ueno, T., Cheong, M., Inoue, T., Whetstone, W. D., Endo, T., Niizuma, K., & Tominaga, T. (2020) · Neurotrauma Reports · doi:10.1089/neur.2020.0009 · Cited by 99 · see citing works
A “Patch” to the NYU Emergency Department Visit Algorithm
Johnston, K. J., Allen, L., Melanson, T., & Pitts, S. R. (2017) · Health Services Research · doi:10.1111/1475-6773.12638 · Cited by 82 · see citing works
Irvin, J., Kondrich, A., Ko, M. K.-L., Rajpurkar, P., Haghgoo, B., Landon, B. E., Phillips, R. L., Petterson, S. M., Ng, A. Y., & Basu, S. (2020) · BMC Public Health · doi:10.1186/s12889-020-08735-0 · Cited by 70 · see citing works
Self-citationAdvancing Artificial Intelligence in Health Settings Outside the Hospital and Clinic
Aggarwal, N., Ahmed, M. K., Basu, S., Curtin, J. J., Evans, B. J., Matheny, M. E., Nundy, S., Sendak, M., Shachar, C., Shah, R. U., et al. (2020) · NAM Perspectives · doi:10.31478/202011f · Cited by 63 · see citing works
Self-citationResidential Structural Racism and Prevalence of Chronic Health Conditions
Mohottige, D., Davenport, C. A., Bhavsar, N. A., Schappe, T., Lyn, M., Maxson, P. J., Johnson, F., Planey, A. M., McElroy, L. M., Wang, V., et al. (2023) · JAMA Network Open · doi:10.1001/jamanetworkopen.2023.48914 · Cited by 58 · see citing works
Herrin, J. S., Abraham, N. S., Yao, X., Noseworthy, P. A., Inselman, J., Shah, N. D., & Ngufor, C. G. (2021) · JAMA Network Open · doi:10.1001/jamanetworkopen.2021.10703 · Cited by 48 · see citing works
Philipson, T., Durie, T., Cong, Z., & Fendrick, A. M. (2023) · BMC Health Services Research · doi:10.1186/s12913-023-09738-4 · Cited by 46 · see citing works
Reducing Readmissions by Addressing the Social Determinants of Health
Evans, W. N., Kroeger, S., Munnich, E. L., Ortuzar, G., & Wagner, K. L. (2020) · American Journal of Health Economics · doi:10.1086/711725 · Cited by 29 · see citing works
Coronado, G. D., Nyongesa, D. B., Petrik, A. F., Thompson, J. H., Escaron, A. L., Younger, B., Harbison, S., & Leo, M. C. (2020) · Clinical Gastroenterology and Hepatology · doi:10.1016/j.cgh.2020.07.053 · Cited by 24 · see citing works
Patel, S. Y., Huskamp, H. A., Barnett, M. L., Zubizarreta, J. R., Zachrison, K. S., Busch, A. B., Wilcock, A. D., & Mehrotra, A. (2021) · Psychiatric Services · doi:10.1176/appi.ps.202100145 · Cited by 16 · see citing works
Self-citationRadley, D., Shah, A., Collins, S. R., Powe, N. R., & Zephyrin, L. (2024) · The Commonwealth Fund Website · doi:10.26099/vw02-fa96 · Cited by 15 · see citing works
Silvestri, D. M., Goutos, D., Lloren, A., Zhou, S., Zhou, G., Farietta, T., Charania, S., Herrin, J. S., Peltz, A., Lin, Z., et al. (2022) · JAMA Health Forum · doi:10.1001/jamahealthforum.2021.4611 · Cited by 13 · see citing works
Prediction of non emergent acute care utilization and cost among patients receiving Medicaid
Patel, S. Y., Baum, A., & Basu, S. (2024) · Scientific Reports · doi:10.1038/s41598-023-51114-z · Cited by 10 · see citing works
Self-citationHogg‐Graham, R., Scott, A. M., Clear, E. R., Riley, E. N., & Waters, T. M. (2024) · BMC Health Services Research · doi:10.1186/s12913-024-10705-w · Cited by 6 · see citing works
Coronado, G. D., Nyongesa, D. B., Petrik, A. F., Thompson, J. H., Escaron, A. L., Pham, T., & Leo, M. C. (2024) · Cancer Epidemiology Biomarkers & Prevention · doi:10.1158/1055-9965.epi-23-0940 · Cited by 6 · see citing works
Addressing Social Needs in Clinical Settings: Early Lessons from Accountable Health Communities
Beidler, L. B., Colvin, J. D., Winterer, C. M., & Fraze, T. (2023) · Population Health Management · doi:10.1089/pop.2023.0119 · Cited by 4 · see citing works
Flagged first, then by citations
Research works printed but not linked: 4
- [6]National Academy of Medicine; The Learning Health System Series, Whicher D, Ahmed M, Israni ST, Matheny M, eds. Artificial Intelligence in Health Care: The Hope, the Hype, the Promise, the Peril. Washington (DC): National Academies Press (US); August 2, 2023. 39146448
- [8]Hall, A. K., Cole-Lewis, H. & Bernhardt, J. M. Using text message reminders in health care services: A narrative literature review. mHealth1, 21 (2015). https://www.sciencedirect.com/science/article/pii/S2214782918300022 Available from.
- [19]Baum, A., Batniji, R., Ratcliffe, H., DeGosztonyi, M. & Basu, S. Supporting rising-risk Medicaid patients through early intervention. NEJM Catal. Innov. Care Deliv. 5, 11 (2024).
- [53]Lundberg, S. M. & Lee, S. I. A unified approach to interpreting model predictions. Adv. Neural Inf. Process. Syst. 30, 4765–4774 (2017).
Cited by
Sorted by citations
Patel, S. Y., & Basu, S. (2026) · BMJ Open · doi:10.1136/bmjopen-2025-111343 · Cited by 1 · see citing works
Self-citationOptimal machine learning models and factors for global infectious diseases: An ecological analysis
Lv, H., Wang, L., Zhang, X., Dang, C., Liu, F., Liu, Z., Xu, Y., & Chen, H. (2026) · Infectious Medicine · doi:10.1016/j.imj.2026.100251 · Cited by 1 · see citing works
Razzano, A. V. (2026) · Health Science Reports · doi:10.1002/hsr2.73083
Basu, S., & Batniji, R. (2026) · Research Square · doi:10.21203/rs.3.rs-9903473/v1
Self-citation
- Authors
- Patel, S. Y., Barnett, M. L., & Basu, S.
- Affiliations
- University of Pennsylvania (US)
- Brigham and Women's Hospital (US)
- University of California, San Francisco (US)
- Corresponding author
- Patel, Sadiq Y.
- Venue
- npj Digital Medicine
- Published
- 2025-07-02
- Volume
- 8
- Issue
- 1
- Pages
- 393-393
- ISSN
- 2398-6352 (ISSN-L)
- PMC
- PMC12223227
- OpenAlex
- W4411918463
- Cited by
- 4
- References
- 27
- Citations by year
- 2026: 4
- FWCI
- 1.60
- Type
- journal article
- Language
- en
- Open access
- gold
- doi.org
- Licence
- Locations
- npj Digital Medicine (publishedVersion, cc-by-nc-nd)
- PubMed Central (publishedVersion, cc-by-nc-nd)
- PubMed (publishedVersion)
- DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
- Topics
- Physical Sciences › Computer Science › Artificial Intelligence › Machine Learning in Healthcare
- Health Sciences › Health Professions › Health Information Management › Artificial Intelligence in Healthcare
- Health Sciences › Medicine › Public Health, Environmental and Occupational Health › Health Promotion and Cardiovascular Prevention
- Keywords
- Medicaid, low-income patients, preventive medicine, chronic disease management, social determinants of health, healthcare workforce, racial disparities, black-white disparities, missed vaccination, cancer screening
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
