An official website of the United States government
Accurate prediction and monitoring of patient health in the intensive care unit can inform shared decisions regarding appropriateness of care delivery, risk-reduction strategies, and intensive care resource use. Traditionally, algorithmic solutions for patient outcome prediction rely solely on data available from electronic health records (EHR). In this pilot study, we explore the benefits of augmenting existing EHR data with novel measurements from wrist-worn activity sensors as part of a clinical environment known as the Intelligent ICU. We implemented temporal deep learning models based on two distinct sources of patient data: (1) routinely measured vital signs from electronic health records, and (2) activity data collected from wearable sensors. As a proxy for illness severity, our models predicted whether patients leaving the intensive care unit would be successfully or unsuccessfully discharged from the hospital. We overcome the challenge of small sample size in our prospective cohort by applying deep transfer learning using EHR data from a much larger cohort of traditional ICU patients. Our experiments quantify added utility of non-traditional measurements for predicting patient health, especially when applying a transfer learning procedure to small novel Intelligent ICU cohorts of critically ill patients.
more »
« less
Data heterogeneity in federated learning with Electronic Health Records: Case studies of risk prediction for acute kidney injury and sepsis diseases in critical care
With the wider availability of healthcare data such as Electronic Health Records (EHR), more and more data-driven based approaches have been proposed to improve the quality-of-care delivery. Predictive modeling, which aims at building computational models for predicting clinical risk, is a popular research topic in healthcare analytics. However, concerns about privacy of healthcare data may hinder the development of effective predictive models that are generalizable because this often requires rich diverse data from multiple clinical institutions. Recently, federated learning (FL) has demonstrated promise in addressing this concern. However, data heterogeneity from different local participating sites may affect prediction performance of federated models. Due to acute kidney injury (AKI) and sepsis’ high prevalence among patients admitted to intensive care units (ICU), the early prediction of these conditions based on AI is an important topic in critical care medicine. In this study, we take AKI and sepsis onset risk prediction in ICU as two examples to explore the impact of data heterogeneity in the FL framework as well as compare performances across frameworks. We built predictive models based on local, pooled, and FL frameworks using EHR data across multiple hospitals. The local framework only used data from each site itself. The pooled framework combined data from all sites. In the FL framework, each local site did not have access to other sites’ data. A model was updated locally, and its parameters were shared to a central aggregator, which was used to update the federated model’s parameters and then subsequently, shared with each site. We found models built within a FL framework outperformed local counterparts. Then, we analyzed variable importance discrepancies across sites and frameworks. Finally, we explored potential sources of the heterogeneity within the EHR data. The different distributions of demographic profiles, medication use, and site information contributed to data heterogeneity.
more »
« less
- PAR ID:
- 10438310
- Editor(s):
- Frasch, Martin G.
- Date Published:
- Journal Name:
- PLOS Digital Health
- Volume:
- 2
- Issue:
- 3
- ISSN:
- 2767-3170
- Page Range / eLocation ID:
- e0000117
- Format(s):
- Medium: X
- Sponsoring Org:
- National Science Foundation
More Like this
-
-
Deep learning models have demonstrated impressive accuracy in predicting acute kidney injury (AKI), a condition affecting up to 20% of ICU patients, yet their black-box nature prevents clinical adoption in high-stakes critical care settings. While existing interpretability methods like SHAP, LIME, and attention mechanisms can identify important features, they fail to capture the temporal dynamics essential for clinical decision-making, and are unable to communicate when specific risk factors become critical in a patient's trajectory. This limitation is particularly problematic in the ICU, where the timing of interventions can significantly impact patient outcomes. We present a novel interpretable framework that brings temporal awareness to deep learning predictions for AKI. Our approach introduces three key innovations: (1) a latent convolutional concept bottleneck that learns clinically meaningful patterns from ICU time-series without requiring manual concept annotation, leveraging Conv1D layers to capture localized temporal patterns like sudden physiological changes; (2) Temporal Concept Tracing (TCT), a gradient-based method that identifies not only which risk factors matter but precisely when they become critical addressing the fundamental question of temporal relevance missing from current XAI techniques; and (3) integration with MedAlpaca to generate structured, time-aware clinical explanations that translate model insights into actionable bedside guidance. We evaluate our framework on MIMIC-IV data, demonstrating that our approach performs better than existing explainability frameworks, Occlusion and LIME, in terms of the comprehensiveness score, sufficiency score, and processing time. The proposed method also better captures risk factors inflection points for patients timelines compared to conventional concept bottleneck methods, including dense layer and attention mechanism. This work represents the first comprehensive solution for interpretable temporal deep learning in critical care that addresses both the what and when of clinical risk factors. By making AKI predictions transparent and temporally contextualized, our framework bridges the gap between model accuracy and clinical utility, offering a path toward trustworthy AI deployment in time-sensitive healthcare settings.more » « less
-
The predictive Intensive Care Unit (ICU) scoring system plays an important role in ICU management for its capability of predicting important outcomes, especially mortality. There are many scoring systems that have been developed and used in the ICU. These scoring systems are primarily based on the structured clinical data contained in the electronic health record (EHR), which may suffer the loss of the important clinical information contained in the narratives and images. In this work, we build a deep learning based survival prediction model with multimodality data to predict ICU-mortality. Four sets of features are investigated: (1) physiological measurements of Simplified Acute Physiology Score (SAPS) II, (2) common thorax diseases predefined by radiologists, (3) BERT-based text representations, and (4) chest X-ray image features. We use the Medical Information Mart for Intensive Care IV (MIMIC-IV) dataset to evaluate the proposed model. Our model achieves the average C-index of 0.7847 (95% confidence interval, 0.7625–0.8068), which substantially exceeds that of the baseline with SAPS-II features (0.7477 (0.7238–0.7716)). Ablation studies further demonstrate the contributions of pre-defined labels (2.12%), text features (2.68%), and image features (2.96%). Our model achieves a higher average C-index than the traditional machine learning methods under the same feature fusion setting, which suggests that the deep learning methods can outperform the traditional machine learning methods in ICU-mortality prediction. These results highlight the potential of deep learning models with multimodal information to enhance ICU-mortality prediction. We make our work publicly available at https://github.com/bionlplab/mimic-icu-mortality.more » « less
-
This survey explores the transformative impact of foundation models (FMs) in artificial intelligence, focusing on their integration with federated learning (FL) in biomedical research. Foundation models such as ChatGPT, LLaMa, and CLIP, which are trained on vast datasets through methods including unsupervised pretraining, self-supervised learning, instructed fine-tuning, and reinforcement learning from human feedback, represent significant advancements in machine learning. These models, with their ability to generate coherent text and realistic images, are crucial for biomedical applications that require processing diverse data forms such as clinical reports, diagnostic images, and multimodal patient interactions. The incorporation of FL with these sophisticated models presents a promising strategy to harness their analytical power while safeguarding the privacy of sensitive medical data. This approach not only enhances the capabilities of FMs in medical diagnostics and personalized treatment but also addresses critical concerns about data privacy and security in healthcare. This survey reviews the current applications of FMs in federated settings, underscores the challenges, and identifies future research directions including scaling FMs, managing data diversity, and enhancing communication efficiency within FL frameworks. The objective is to encourage further research into the combined potential of FMs and FL, laying the groundwork for healthcare innovations.more » « less
-
Abstract Background Sepsis is a heterogeneous syndrome, and the identification of clinical subphenotypes is essential. Although organ dysfunction is a defining element of sepsis, subphenotypes of differential trajectory are not well studied. We sought to identify distinct Sequential Organ Failure Assessment (SOFA) score trajectory-based subphenotypes in sepsis. Methods We created 72-h SOFA score trajectories in patients with sepsis from four diverse intensive care unit (ICU) cohorts. We then used dynamic time warping (DTW) to compute heterogeneous SOFA trajectory similarities and hierarchical agglomerative clustering (HAC) to identify trajectory-based subphenotypes. Patient characteristics were compared between subphenotypes and a random forest model was developed to predict subphenotype membership at 6 and 24 h after being admitted to the ICU. The model was tested on three validation cohorts. Sensitivity analyses were performed with alternative clustering methodologies. Results A total of 4678, 3665, 12,282, and 4804 unique sepsis patients were included in development and three validation cohorts, respectively. Four subphenotypes were identified in the development cohort: Rapidly Worsening ( n = 612, 13.1%), Delayed Worsening ( n = 960, 20.5%), Rapidly Improving ( n = 1932, 41.3%), and Delayed Improving ( n = 1174, 25.1%). Baseline characteristics, including the pattern of organ dysfunction, varied between subphenotypes. Rapidly Worsening was defined by a higher comorbidity burden, acidosis, and visceral organ dysfunction. Rapidly Improving was defined by vasopressor use without acidosis. Outcomes differed across the subphenotypes, Rapidly Worsening had the highest in-hospital mortality (28.3%, P -value < 0.001), despite a lower SOFA (mean: 4.5) at ICU admission compared to Rapidly Improving (mortality:5.5%, mean SOFA: 5.5). An overall prediction accuracy of 0.78 (95% CI, [0.77, 0.8]) was obtained at 6 h after ICU admission, which increased to 0.87 (95% CI, [0.86, 0.88]) at 24 h. Similar subphenotypes were replicated in three validation cohorts. The majority of patients with sepsis have an improving phenotype with a lower mortality risk; however, they make up over 20% of all deaths due to their larger numbers. Conclusions Four novel, clinically-defined, trajectory-based sepsis subphenotypes were identified and validated. Identifying trajectory-based subphenotypes has immediate implications for the powering and predictive enrichment of clinical trials. Understanding the pathophysiology of these differential trajectories may reveal unanticipated therapeutic targets and identify more precise populations and endpoints for clinical trials.more » « less
- Free Publicly Accessible Full Text
-
Accepted Manuscript1.0
- Journal Article:
- https://doi.org/10.1371/journal.pdig.0000117
