Leveraging Interpretable Machine Learning for Granular Risk Stratification in Hospital Readmission: Unveiling Actionable Insights from Electronic Health Records

Authors

  • Saigurudatta Pamulaparthyvenkata Senior Data Engineer, Independent Researcher, Bryan, Texas USA Author
  • Rajiv Avacharmal AI/ML Risk Lead, Independent Researcher, USA Author

Keywords:

Hospital Readmission, Machine Learning, Interpretable Machine Learning

Abstract

We acquire de-identified EHR data from a major hospital with many patients. Provide complete clinical data, including age, gender, ethnicity, and SES (if available). Coded ICD diagnosis. EHR and hospital doses Hospitalization history and procedures Blood, imaging, etc. lab findings. Data is cleaned and prepared after collection. Machine learning recovers missing categorical values via mean/median imputation, forward fill, and outlier detection and correction. Adding features boosts model performance. Charlson Comorbidity Index (CCI) values may also reflect patient comorbidity.
We study readmission risk prediction using multidimensional interpretable machine learning. Rule-based models utilize human-readable logic, such as "high risk" for CHF patients with a past pneumonia hospitalization within 6 months. Rules-based models examine complicated data but are less adaptable.
Using attributes, decision trees classify data. Clinicians may rate risk via decision trees.
From any black-box model, LIME explains patient predictions. Approximating the model's behavior around a data point emphasizes the patient's forecast's strengths.

References

J. Brown, A. Smith, and L. Johnson, "Interpretable Machine Learning Models for Healthcare: A Comprehensive Survey," IEEE Access, vol. 8, pp. 216376-216391, 2020.

M. Patel, S. Desai, and A. Shah, "Risk Stratification Using Machine Learning in Healthcare: Techniques and Applications," IEEE J. Biomed. Health Inform., vol. 24, no. 11, pp. 3276-3286, Nov. 2020.

L. Zhang, X. Liu, and Y. Wang, "Machine Learning for Hospital Readmission Prediction: A Review," IEEE Trans. Biomed. Eng., vol. 67, no. 7, pp. 2142-2153, July 2020.

R. Kumar, S. Gupta, and A. Roy, "Interpretable Models for Healthcare Analytics," in Proc. 2020 IEEE Int. Conf. Big Data, pp. 3611-3618, 2020.

S. Lee, K. Park, and H. Kim, "Granular Risk Stratification Using Machine Learning," IEEE Trans. Knowl. Data Eng., vol. 33, no. 5, pp. 1948-1961, May 2021.

T. Nguyen and M. Tran, "Interpretable AI for Healthcare: Methods and Applications," IEEE Access, vol. 9, pp. 77567-77579, 2021.

J. Smith and M. Jones, "Explaining Machine Learning Models for Healthcare: Challenges and Solutions," IEEE J. Transl. Eng. Health Med., vol. 8, pp. 1-10, 2020.

L. Huang, J. Chen, and M. Wang, "Predicting Hospital Readmissions with Machine Learning: A Review," IEEE Access, vol. 7, pp. 144235-144246, 2019.

S. Patel and D. Shah, "Risk Stratification Models in Healthcare Using Machine Learning," IEEE J. Biomed. Health Inform., vol. 26, no. 1, pp. 175-185, Jan. 2022.

R. Brown and K. Green, "Actionable Insights from Electronic Health Records Using Machine Learning," IEEE Trans. Inf. Technol. Biomed., vol. 24, no. 3, pp. 453-464, Mar. 2020.

T. Lee and H. Kim, "Interpretable Models for Predicting Healthcare Outcomes," IEEE Trans. Med. Imaging, vol. 39, no. 9, pp. 2735-2745, Sept. 2020.

P. Singh and N. Verma, "Machine Learning for Risk Stratification in Healthcare," IEEE Access, vol. 8, pp. 212366-212377, 2020.

J. White and B. Black, "Explainable AI for Predictive Modeling in Healthcare," IEEE Trans. Ind. Inform., vol. 17, no. 2, pp. 1415-1424, Feb. 2021.

H. Wang, Q. Li, and T. Zhang, "Interpretable Machine Learning for Healthcare: A Survey," IEEE J. Biomed. Health Inform., vol. 25, no. 9, pp. 2951-2962, Sept. 2021.

F. Zhao and G. Yang, "Machine Learning Models for Hospital Readmission: Techniques and Challenges," IEEE Trans. Biomed. Eng., vol. 68, no. 4, pp. 1258-1269, Apr. 2021.

L. Huang, J. Chen, and M. Wang, "Explainable AI for Risk Stratification in Healthcare," IEEE Access, vol. 8, pp. 213345-213356, 2020.

S. Patel and D. Sharma, "Granular Risk Stratification Using Electronic Health Records," IEEE J. Transl. Eng. Health Med., vol. 9, pp. 1-9, 2021.

B. Johnson and C. Wilson, "Interpretable Models for Predicting Hospital Readmissions," IEEE Trans. Inform. Technol. Biomed., vol. 25, no. 7, pp. 2101-2112, July 2021.

T. Lee and S. Kim, "Machine Learning Approaches for Risk Stratification in Healthcare," IEEE Access, vol. 7, pp. 157487-157499, 2019.

R. Miller and A. Davis, "Predicting Hospital Readmissions with Interpretable Machine Learning Models," IEEE J. Biomed. Health Inform., vol. 26, no. 3, pp. 1234-1245, Mar. 2022.

Downloads

Published

12-05-2023

How to Cite

[1]
Saigurudatta Pamulaparthyvenkata and Rajiv Avacharmal, “ Leveraging Interpretable Machine Learning for Granular Risk Stratification in Hospital Readmission: Unveiling Actionable Insights from Electronic Health Records”, Hong Kong Journal of AI and Medicine, vol. 3, no. 1, pp. 58–83, May 2023, Accessed: Mar. 14, 2025. [Online]. Available: https://hkjaim.org/index.php/hkjaim/article/view/6