What question did this study set out to answer?

To develop machine learning models that predict 30-day hospital readmission by integrating clinical and social determinants of health data.

April 26, 2026SHILAP Revista de lepidopterologíaOpen Access

Leveraging a machine learning model to predict hospital readmission risk: integrating clinical and social determinants of health data

Q: What are the key findings of this study?

An XGBoost machine learning model integrating clinical and social determinants of health data significantly improved 30-day hospital readmission prediction compared to a clinical-only baseline model (ROC-AUC 0.79 vs 0.68).

Q: What is the clinical evidence from this study?

Study design: Cohort. Population: Hospital readmission (n=3018). Intervention: XGBoost machine learning model integrating clinical and SDOH data vs. Clinical-only Logistic Regression baseline model. Primary outcome: 30-day all-cause unplanned hospital readmission (evaluated by ROC-AUC) (95% CI 0.75-0.82, p=<0.001).

View Full Paper

Ask AI

Bookmark

Key result

XGBoost model integrating social determinants of health improves 30-day readmission prediction over a clinical-only baseline.

95% CI 0.75-0.82
P<0.001
n=3,018

Why the study?

Most readmission prediction models rely on clinical data, and the use of machine learning to combine clinical and social determinants of health data remains limited.

Does integrating clinical and Social Determinants of Health (SDOH) data into machine learning models improve the prediction of 30-day all-cause hospital readmission in discharged adult patients?

Population

3,018 adult patients discharged from a large academic medical center

Comparison

Six machine learning models integrating clinical and SDOH data vs each other

Design

Retrospective cohort study

Follow-up

30 days

Authors

TZTianyu ZhangBeijing Technology and Business University

Discussion

Member takes

Implication

May improve readmission risk stratification via SDOH-ML; extends conventional models but leaves open prospective validation before adoption.

Key Points

To develop machine learning models that predict 30-day hospital readmission by integrating clinical and social determinants of health data.
Retrospective cohort study of 3,018 adult patients discharged from a medical center from January 2022 to December 2023.
Integrated clinical variables from electronic health records with area-level social determinants from census data.
Trained and evaluated six machine learning models including XGBoost and assessed using ROC-AUC and F1-score.
XGBoost achieved the best model performance with ROC-AUC of 0.79 (95% CI 0.75–0.82) and PR-AUC of 0.71.
Integrated SDOH variables like neighborhood socioeconomic status were important predictors alongside clinical variables.
Ensemble models outperformed traditional Logistic Regression.

Study Design

Type

Cohort (n=3,018)

Multicenter

Structured PICO

Does integrating clinical and Social Determinants of Health (SDOH) data into machine learning models improve the prediction of 30-day all-cause hospital readmission in discharged adult patients?

Population

3,018 adult patients discharged alive from a large academic medical center between January 2022 and December 2023. Included patients had one of five conditions: acute myocardial infarction (18.3%), heart failure (26.7%), pneumonia (22.1%), chronic obstructive pulmonary disease (20.5%), or elective total hip/knee arthroplasty (12.4%). Excluded: in-hospital mortality, transfer to other acute care facilities, discharge to hospice care, and missing essential clinical or demographic data.

Intervention

Machine learning models (XGBoost, Random Forest, LightGBM, CatBoost, Support Vector Machine) integrating clinical data from electronic health records with area-level Social Determinants of Health (SDOH) data (neighborhood deprivation, median income, educational attainment, etc.) derived from the American Community Survey.

Comparator

Clinical-only baseline model (Logistic Regression relying predominantly on clinical and administrative data).

Outcome

30-day all-cause unplanned hospital readmission after discharge from the index hospitalization.

Main Result

Absolute Event Rate: 0.79% vs 0.68%

p-value: p=<0.001

Integrating social determinants of health with clinical data in machine learning models significantly improves the prediction of 30-day hospital readmissions compared to clinical-only models.

Limitations

Data from a single academic medical center may limit generalizability to other healthcare settings.
Use of area-level SDOH proxies may not fully capture individual-level social circumstances.
Cross-sectional nature of the analysis precludes causal inference regarding SDOH factors and readmission risk.
Clinical implementation requires additional work to integrate models seamlessly into clinical workflows.

Cite This Study

Tianyu Zhang (2026) conducted a cohort in Hospital readmission (n=3,018). XGBoost machine learning model integrating clinical and SDOH data vs. Clinical-only Logistic Regression baseline model was evaluated on 30-day all-cause unplanned hospital readmission (evaluated by ROC-AUC) (95% CI 0.75-0.82, p=<0.001). An XGBoost machine learning model integrating clinical and social determinants of health data significantly improved 30-day hospital readmission prediction compared to a clinical-only baseline model (ROC-AUC 0.79 vs 0.68).

synapsesocial.com/papers/69edaafc4a46254e215b340c https://doi.org/10.3389/fpubh.2026.1754585

View Full Paper

Ask AI

Bookmark

April 26, 2026SHILAP Revista de lepidopterologíaOpen Access

Leveraging a machine learning model to predict hospital readmission risk: integrating clinical and social determinants of health data

View Full Paper

Ask AI

Bookmark

Key result

XGBoost model integrating social determinants of health improves 30-day readmission prediction over a clinical-only baseline.

95% CI 0.75-0.82
P<0.001
n=3,018

Why the study?

Most readmission prediction models rely on clinical data, and the use of machine learning to combine clinical and social determinants of health data remains limited.

Does integrating clinical and Social Determinants of Health (SDOH) data into machine learning models improve the prediction of 30-day all-cause hospital readmission in discharged adult patients?

Population

3,018 adult patients discharged from a large academic medical center

Comparison

Six machine learning models integrating clinical and SDOH data vs each other

Design

Retrospective cohort study

Follow-up

30 days

Authors

TZTianyu ZhangBeijing Technology and Business University

Discussion

Member takes

Implication

May improve readmission risk stratification via SDOH-ML; extends conventional models but leaves open prospective validation before adoption.

Key Points

To develop machine learning models that predict 30-day hospital readmission by integrating clinical and social determinants of health data.
Retrospective cohort study of 3,018 adult patients discharged from a medical center from January 2022 to December 2023.
Integrated clinical variables from electronic health records with area-level social determinants from census data.
Trained and evaluated six machine learning models including XGBoost and assessed using ROC-AUC and F1-score.
XGBoost achieved the best model performance with ROC-AUC of 0.79 (95% CI 0.75–0.82) and PR-AUC of 0.71.
Integrated SDOH variables like neighborhood socioeconomic status were important predictors alongside clinical variables.
Ensemble models outperformed traditional Logistic Regression.

Study Design

Type

Cohort (n=3,018)

Multicenter

Structured PICO

Does integrating clinical and Social Determinants of Health (SDOH) data into machine learning models improve the prediction of 30-day all-cause hospital readmission in discharged adult patients?

Population

Intervention

Comparator

Clinical-only baseline model (Logistic Regression relying predominantly on clinical and administrative data).

Outcome

30-day all-cause unplanned hospital readmission after discharge from the index hospitalization.

Main Result

Absolute Event Rate: 0.79% vs 0.68%

p-value: p=<0.001

Integrating social determinants of health with clinical data in machine learning models significantly improves the prediction of 30-day hospital readmissions compared to clinical-only models.

Limitations

Data from a single academic medical center may limit generalizability to other healthcare settings.
Use of area-level SDOH proxies may not fully capture individual-level social circumstances.
Cross-sectional nature of the analysis precludes causal inference regarding SDOH factors and readmission risk.
Clinical implementation requires additional work to integrate models seamlessly into clinical workflows.

Cite This Study

synapsesocial.com/papers/69edaafc4a46254e215b340c https://doi.org/10.3389/fpubh.2026.1754585

Also Consider

Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context:

View Full Paper

Ask AI

Bookmark