Network analytics and machine learning for predicting length of stay in elderly patients with chronic diseases at point of admission

Abstract

Background An aging population with a burden of chronic diseases puts increasing pressure on health care systems. Early prediction of the hospital length of stay (LOS) can be useful in optimizing the allocation of medical resources, and improving healthcare quality. However, the data available at the point of admission (PoA) are limited, making it difficult to forecast the LOS accurately. Methods In this study, we proposed a novel approach combining network analytics and machine learning to predict the LOS in elderly patients with chronic diseases at the PoA. Two networks, including multimorbidity network (MN) and patient similarity network (PSN), were constructed and novel network features were created. Five machine learning models (eXtreme Gradient Boosting, Gradient Boosting Decision Tree, Random Forest, Linear Support Vector Machine, and Deep Neural Network) with different input feature sets were developed to compare their performance. Results The experimental results indicated that the network features can bring significant improvements to the performances of the prediction models, suggesting that the MN and PSN are useful for LOS predictions. Conclusion Our predictive framework which integrates network science with data mining can forecast the LOS effectively at the PoA and provide decision support for hospital managers, which highlights the potential value of network-based machine learning in healthcare field.

Risk prediction of heart failure in patients with ischemic heart disease using network analytics and stacking ensemble learning

“…Second, our network features were proposed based on the association of disease pairs or the progressive relationship of disease pairs. The prediction accuracy could be further improved by considering the patient-to-patient similarity inherent in the administrative dataset [ 38 , 54 ]. In addition, graph neural networks (GNNs) are increasingly popular for learning network-based tasks [ 55 , 56 ].…”

Applying precision medicine principles to the management of multimorbidity: the utility of comorbidity networks, graph machine learning, and knowledge graphs

“…When predicting LOS in older patients with chronic disease, the K-NN algorithm was applied to a PSN created using the Jaccard similarity score to detect the K = 100 nearest neighbors. For each node (patient), aggregated LOS functions (mean, SD, min, and max) were then calculated based on their neighbors and used with baseline information and features from a DCN to predict LOS (59). The PSN features accounted for 33.1% of the feature importance for LOS prediction using a random forest algorithm that achieved an R 2 = 0.347.…”

“…Diseases with high eigenvector centrality are those conditions related to more influential diseases, which may help in indicating which disease pairs are causally related (34). When predicting hospital LOS from a multimorbidity network (MN) of older patients, an eigenvector centrality (EVC) score for patients obtained by summing the EVC of their disease nodes was an important factor in predicting LOS, improving the R 2 by 18.7% beyond patient clinical and demographic data (59). In a DCN with 120 communities including nine major disease groups, EVC scores improved overall accuracy, sensitivity, and specificity, which were 69.52, 78.81, and 69.02%, respectively, for predicting high-cost patients (32).…”

A comprehensive review of machine learning algorithms and their application in geriatric medicine: present and future

“…The newly generated features were more predictive for incident diabetes than demographic and clinical features. Similarly, patient–patient network and disease–disease network features accounted for 25.7% and 27.3% of feature importance in predicting hospital length of stay, compared with 15.3% and 31.6% for baseline (e.g., age and gender) and historical features (e.g., length of stay for previous admissions) [ 122 ].…”