1
|
Keats K, Deng S, Chen X, Zhang T, Devlin JW, Murphy DJ, Smith SE, Murray B, Kamaleswaran R, Sikora A. Unsupervised machine learning analysis to identify patterns of ICU medication use for fluid overload prediction. MEDRXIV : THE PREPRINT SERVER FOR HEALTH SCIENCES 2024:2024.03.21.24304663. [PMID: 38562806 PMCID: PMC10984037 DOI: 10.1101/2024.03.21.24304663] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [Key Words] [Grants] [Track Full Text] [Subscribe] [Scholar Register] [Indexed: 04/04/2024]
Abstract
INTRODUCTION Intravenous (IV) medications are a fundamental cause of fluid overload (FO) in the intensive care unit (ICU); however, the association between IV medication use (including volume), administration timing, and FO occurrence remains unclear. METHODS This retrospective cohort study included consecutive adults admitted to an ICU ≥72 hours with available fluid balance data. FO was defined as a positive fluid balance ≥7% of admission body weight within 72 hours of ICU admission. After reviewing medication administration record (MAR) data in three-hour periods, IV medication exposure was categorized into clusters using principal component analysis (PCA) and Restricted Boltzmann Machine (RBM). Medication regimens of patients with and without FO were compared within clusters to assess for temporal clusters associated with FO using the Wilcoxon rank sum test. Exploratory analyses of the medication cluster most associated with FO for medications frequently appearing and used in the first 24 hours was conducted. RESULTS FO occurred in 127/927 (13.7%) of the patients enrolled. Patients received a median (IQR) of 31 (13-65) discrete IV medication administrations over the 72-hour period. Across all 47,803 IV medication administrations, ten unique IV medication clusters were identified with 121-130 medications in each cluster. Among the ten clusters, cluster 7 had the greatest association with FO; the mean number of cluster 7 medications received was significantly greater in patients in the FO cohort compared to patients who did not experience FO (25.6 vs.10.9. p<0.0001). 51 of the 127 medications in cluster 7 (40.2%) appeared in > 5 separate 3-hour periods during the 72-hour study window. The most common cluster 7 medications included continuous infusions, antibiotics, and sedatives/analgesics. Addition of cluster 7 medications to a prediction model with APACHE II score and receipt of diuretics improved the ability for the model to predict fluid overload (AUROC 5.65, p =0.0004). CONCLUSIONS Using ML approaches, a unique IV medication cluster was strongly associated with FO. Incorporation of this cluster improved the ability to predict development of fluid overload in ICU patients compared with traditional prediction models. This method may be further developed into real-time clinical applications to improve early detection of adverse outcomes.
Collapse
Affiliation(s)
- Kelli Keats
- Augusta University Medical Center, Department of Pharmacy, Augusta, GA
| | - Shiyuan Deng
- University of Georgia Franklin College of Arts and Sciences, Department of Statistics, Athens, GA, USA
| | - Xianyan Chen
- University of Georgia Franklin College of Arts and Sciences, Department of Statistics, Athens, GA, USA
| | - Tianyi Zhang
- University of Georgia Franklin College of Arts and Sciences, Department of Statistics, Athens, GA, USA
| | - John W Devlin
- Northeastern University School of Pharmacy, Boston, MA
- Brigham and Women's Hospital, Division of Pulmonary and Critical Care Medicine, Boston, MA
| | - David J Murphy
- Emory University, Division of Pulmonary, Allergy, Critical Care and Sleep Medicine, Atlanta, GA, USA
| | - Susan E Smith
- University of Georgia College of Pharmacy, Department of Clinical and Administrative Pharmacy, Athens, GA, USA
| | - Brian Murray
- University of Colorado Skaggs School of Pharmacy, Aurora, CO, USA
| | - Rishikesan Kamaleswaran
- Department of Biomedical Informatics, Emory University School of Medicine, Atlanta, GA, USA
- Department of Biomedical Engineering, Georgia Institute of Technology, Atlanta, GA, USA
| | - Andrea Sikora
- 1120 15th Street, HM-118 Augusta, GA 30912
- University of Georgia College of Pharmacy, Department of Clinical and Administrative Pharmacy, Augusta, GA, USA
| |
Collapse
|
2
|
Rafiei A, Ghiasi Rad M, Sikora A, Kamaleswaran R. Improving mixed-integer temporal modeling by generating synthetic data using conditional generative adversarial networks: A case study of fluid overload prediction in the intensive care unit. Comput Biol Med 2024; 168:107749. [PMID: 38011778 DOI: 10.1016/j.compbiomed.2023.107749] [Citation(s) in RCA: 1] [Impact Index Per Article: 1.0] [Reference Citation Analysis] [Abstract] [Key Words] [MESH Headings] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 07/16/2023] [Revised: 10/29/2023] [Accepted: 11/20/2023] [Indexed: 11/29/2023]
Abstract
OBJECTIVE The challenge of mixed-integer temporal data, which is particularly prominent for medication use in the critically ill, limits the performance of predictive models. The purpose of this evaluation was to pilot test integrating synthetic data within an existing dataset of complex medication data to improve machine learning model prediction of fluid overload. MATERIALS AND METHODS This retrospective cohort study evaluated patients admitted to an ICU ≥ 72 h. Four machine learning algorithms to predict fluid overload after 48-72 h of ICU admission were developed using the original dataset. Then, two distinct synthetic data generation methodologies (synthetic minority over-sampling technique (SMOTE) and conditional tabular generative adversarial network (CTGAN)) were used to create synthetic data. Finally, a stacking ensemble technique designed to train a meta-learner was established. Models underwent training in three scenarios of varying qualities and quantities of datasets. RESULTS Training machine learning algorithms on the combined synthetic and original dataset overall increased the performance of the predictive models compared to training on the original dataset. The highest performing model was the meta-model trained on the combined dataset with 0.83 AUROC while it managed to significantly enhance the sensitivity across different training scenarios. DISCUSSION The integration of synthetically generated data is the first time such methods have been applied to ICU medication data and offers a promising solution to enhance the performance of machine learning models for fluid overload, which may be translated to other ICU outcomes. A meta-learner was able to make a trade-off between different performance metrics and improve the ability to identify the minority class.
Collapse
Affiliation(s)
- Alireza Rafiei
- Department of Computer Science and Informatics, Emory University, Ste. W302, 400 Dowman Dr., Atlanta, GA, 30322, USA.
| | - Milad Ghiasi Rad
- Department of Electrical and Computer Engineering, Georgia Institute of Technology, Atlanta, GA, USA.
| | - Andrea Sikora
- University of Georgia College of Pharmacy, Department of Clinical and Administrative Pharmacy, Augusta, GA, USA.
| | - Rishikesan Kamaleswaran
- Department of Biomedical Informatics, Emory University School of Medicine, Atlanta, GA, USA; Department of Biomedical Engineering, Georgia Institute of Technology, Atlanta, GA, USA.
| |
Collapse
|
3
|
Zhi L, Wang X, Pan X, Han C. Fluid balance in the resorption stage correlates with outcomes of severe burn patients. Burns 2023; 49:1916-1925. [PMID: 37821273 DOI: 10.1016/j.burns.2023.05.004] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [Key Words] [MESH Headings] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 12/06/2022] [Revised: 04/26/2023] [Accepted: 05/02/2023] [Indexed: 10/13/2023]
Abstract
OBJECTIVE The resorption stage is an important period involving early anti-shock treatment for severe burn patients. We aimed to investigate the quantitative variability in fluid balance during the resorption stage in severe burn patients, and to study its effect on patient outcomes. METHODS We conducted a single-centre retrospective study of 100 severe burn patients with involvement of > 50% total body surface area (TBSA). We extracted clinical data on demographics, clinical characteristics and outcomes; calculated the daily net fluid balance (difference between fluid intake and fluid output) and daily fluid intake/output within one week after injury; and analysed the association between fluid balance and functional outcomes and prognosis. The relative volume (ml/kg/TBSA) was used for the determination of daily fluid volume in this study. RESULTS The daily net fluid balance (ml/kg/TBSA) of the deceased patients on the 4th, 5th, 6th, and 7th days after injury was higher than that of the surviving patients, but the opposite trend was found for the daily fluid output (ml/kg/TBSA). The partial correlation test showed that in the resorption stage of severe burn patients, fluid output was negatively correlated with the index levels of renal function and liver function, CRP level, blood lactic acid (LA) level, frequency of ventilator treatment, and capillary leakage index (CLI), but net fluid balance showed a completely opposite correlation. Moreover, fluid intake was negatively correlated with the index level of renal function and LA level, but positively correlated with the frequency of ventilator treatment. Furthermore, the logistic regression analysis showed that the net fluid balance and fluid output on Day 6 post-injury were independent risk factors for prognosis. CONCLUSION This study suggested that greater fluid output in the resorption stage of severe burn patients was closely related to better outcomes, in addition, a gradually decreasing, lower positive net fluid balance may contribute to the improvement of functional outcomes, which will provide useful information for early fluid management and further prospective clinical study of severe burns.
Collapse
Affiliation(s)
- Lizhu Zhi
- Department of Burn Surgery, Second Affiliated Hospital of Zhejiang University School of Medicine, Hangzhou, China.
| | - Xingang Wang
- Department of Burn Surgery, Second Affiliated Hospital of Zhejiang University School of Medicine, Hangzhou, China
| | - Xuanliang Pan
- Department of Burn Surgery, Second Affiliated Hospital of Zhejiang University School of Medicine, Hangzhou, China
| | - Chunmao Han
- Department of Burn Surgery, Second Affiliated Hospital of Zhejiang University School of Medicine, Hangzhou, China
| |
Collapse
|
4
|
Sikora A, Zhang T, Murphy DJ, Smith SE, Murray B, Kamaleswaran R, Chen X, Buckley MS, Rowe S, Devlin JW. Machine learning vs. traditional regression analysis for fluid overload prediction in the ICU. Sci Rep 2023; 13:19654. [PMID: 37949982 PMCID: PMC10638304 DOI: 10.1038/s41598-023-46735-3] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [Key Words] [MESH Headings] [Grants] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 06/01/2023] [Accepted: 11/04/2023] [Indexed: 11/12/2023] Open
Abstract
Fluid overload, while common in the ICU and associated with serious sequelae, is hard to predict and may be influenced by ICU medication use. Machine learning (ML) approaches may offer advantages over traditional regression techniques to predict it. We compared the ability of traditional regression techniques and different ML-based modeling approaches to identify clinically meaningful fluid overload predictors. This was a retrospective, observational cohort study of adult patients admitted to an ICU ≥ 72 h between 10/1/2015 and 10/31/2020 with available fluid balance data. Models to predict fluid overload (a positive fluid balance ≥ 10% of the admission body weight) in the 48-72 h after ICU admission were created. Potential patient and medication fluid overload predictor variables (n = 28) were collected at either baseline or 24 h after ICU admission. The optimal traditional logistic regression model was created using backward selection. Supervised, classification-based ML models were trained and optimized, including a meta-modeling approach. Area under the receiver operating characteristic (AUROC), positive predictive value (PPV), and negative predictive value (NPV) were compared between the traditional and ML fluid prediction models. A total of 49 of the 391 (12.5%) patients developed fluid overload. Among the ML models, the XGBoost model had the highest performance (AUROC 0.78, PPV 0.27, NPV 0.94) for fluid overload prediction. The XGBoost model performed similarly to the final traditional logistic regression model (AUROC 0.70; PPV 0.20, NPV 0.94). Feature importance analysis revealed severity of illness scores and medication-related data were the most important predictors of fluid overload. In the context of our study, ML and traditional models appear to perform similarly to predict fluid overload in the ICU. Baseline severity of illness and ICU medication regimen complexity are important predictors of fluid overload.
Collapse
Affiliation(s)
- Andrea Sikora
- Department of Clinical and Administrative Pharmacy, University of Georgia College of Pharmacy, 1120 15th Street, HM-118, Augusta, GA, 30912, USA
| | - Tianyi Zhang
- Department of Statistics, University of Georgia Franklin College of Arts and Sciences, Athens, GA, USA
| | - David J Murphy
- Division of Pulmonary, Allergy, Critical Care and Sleep Medicine, Emory University, Atlanta, GA, USA
| | - Susan E Smith
- Department of Clinical and Administrative Pharmacy, University of Georgia College of Pharmacy, 1120 15th Street, HM-118, Augusta, GA, 30912, USA
| | - Brian Murray
- Department of Pharmacy, University of North Carolina Medical Center, Chapel Hill, NC, USA
| | - Rishikesan Kamaleswaran
- Department of Biomedical Informatics, Emory University School of Medicine, Atlanta, GA, USA
- Department of Biomedical Engineering, Georgia Institute of Technology, Atlanta, GA, USA
| | - Xianyan Chen
- Department of Statistics, University of Georgia Franklin College of Arts and Sciences, Athens, GA, USA
| | | | - Sandra Rowe
- Department of Pharmacy, Oregon Health and Science University, Portland, OR, USA
| | - John W Devlin
- Northeastern University School of Pharmacy, Boston, MA, USA.
- Division of Pulmonary and Critical Care Medicine, Brigham and Women's Hospital, Boston, MA, USA.
| |
Collapse
|
5
|
Rafiei A, Rad MG, Sikora A, Kamaleswaran R. Improving irregular temporal modeling by integrating synthetic data to the electronic medical record using conditional GANs: a case study of fluid overload prediction in the intensive care unit. MEDRXIV : THE PREPRINT SERVER FOR HEALTH SCIENCES 2023:2023.06.20.23291680. [PMID: 37425768 PMCID: PMC10327174 DOI: 10.1101/2023.06.20.23291680] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [Track Full Text] [Subscribe] [Scholar Register] [Indexed: 07/11/2023]
Abstract
Objective The challenge of irregular temporal data, which is particularly prominent for medication use in the critically ill, limits the performance of predictive models. The purpose of this evaluation was to pilot test integrating synthetic data within an existing dataset of complex medication data to improve machine learning model prediction of fluid overload. Materials and Methods This retrospective cohort study evaluated patients admitted to an ICU ≥ 72 hours. Four machine learning algorithms to predict fluid overload after 48-72 hours of ICU admission were developed using the original dataset. Then, two distinct synthetic data generation methodologies (synthetic minority over-sampling technique (SMOTE) and conditional tabular generative adversarial network (CT-GAN)) were used to create synthetic data. Finally, a stacking ensemble technique designed to train a meta-learner was established. Models underwent training in three scenarios of varying qualities and quantities of datasets. Results Training machine learning algorithms on the combined synthetic and original dataset overall increased the performance of the predictive models compared to training on the original dataset. The highest performing model was the metamodel trained on the combined dataset with 0.83 AUROC while it managed to significantly enhance the sensitivity across different training scenarios. Discussion The integration of synthetically generated data is the first time such methods have been applied to ICU medication data and offers a promising solution to enhance the performance of machine learning models for fluid overload, which may be translated to other ICU outcomes. A meta-learner was able to make a trade-off between different performance metrics and improve the ability to identify the minority class.
Collapse
|
6
|
Jacobi J. The pathophysiology of sepsis - 2021 update: Part 2, organ dysfunction and assessment. Am J Health Syst Pharm 2021; 79:424-436. [PMID: 34651652 DOI: 10.1093/ajhp/zxab393] [Citation(s) in RCA: 2] [Impact Index Per Article: 0.7] [Reference Citation Analysis] [Abstract] [Key Words] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Indexed: 11/14/2022] Open
Abstract
DISCLAIMER In an effort to expedite the publication of articles, AJHP is posting manuscripts online as soon as possible after acceptance. Accepted manuscripts have been peer-reviewed and copyedited, but are posted online before technical formatting and author proofing. These manuscripts are not the final version of record and will be replaced with the final article (formatted per AJHP style and proofed by the authors) at a later time. PURPOSE This is the second article in a 2-part series discussing the pathophysiology of sepsis. Part 1 of the series reviewed the immunologic response and overlapping pathways of inflammation and coagulation that contribute to the widespread organ dysfunction. In this article (part 2), major organ systems and their dysfunction in sepsis are reviewed, with discussion of scoring systems used to identify patterns and abnormal vital signs and laboratory values associated with sepsis. SUMMARY Sepsis is a dysregulated host response to infection that produces significant morbidity, and patients with shock due to sepsis have circulatory and cellular and metabolic abnormalities that lead to a higher mortality. Cardiovascular dysfunction produces vasodilation, reduced cardiac output and hypotension/shock requiring fluids, vasopressors, and advanced hemodynamic monitoring. Respiratory dysfunction may require mechanical ventilation and attention to volume status. Renal dysfunction is a frequent manifestation of sepsis. Hematologic dysfunction produces low platelets and either elevation or reduction of leucocytes, so consideration of the neutrophil:lymphocyte ratio may be useful. Procoagulant and antifibrinolytic activity leads to coagulation that is stimulated by inflammation. Hepatic dysfunction manifest as elevated bilirubin is often a late finding in sepsis and may cause reductions in production of essential proteins. Neurologic dysfunction may result from local endothelial injury and systemic inflammation through activity of the vagus nerve. CONCLUSION Timely recognition and team response with efficient use of therapies can improve patient outcome, and pharmacists with a complete understanding of the pathophysiologic mechanisms and treatments are valuable members of that team.
Collapse
|
7
|
Abstract
Acute kidney injury (AKI) and chronic kidney disease are common interconnected syndromes that represent a public health problem. Acute kidney disease (AKD) is defined as the post-AKI status of acute or subacute kidney damage/dysfunction manifested by persistence of AKI beyond 7 to 90 days after the initial AKI diagnosis. Limited clinical data exist regarding AKD epidemiology but its incidence is observed in ∼25% of AKI survivors. Useful risk-stratification tools to predict risk of AKD and its prognosis are needed. Interventions on fluid management, nephrotoxic exposure, and follow-up care hold promise to ameliorate the burden of AKD and its complications.
Collapse
Affiliation(s)
- Javier A Neyra
- Department of Internal Medicine, Division of Nephrology, Bone and Mineral Metabolism, University of Kentucky Medical Center, 800 Rose Street, MN668, Lexington, KY 40536, USA.
| | - Lakhmir S Chawla
- Department of Medicine, Veterans Affairs Medical Center, 3350 La Jolla Village Drive, San Diego, CA 92161, USA
| |
Collapse
|