An interpretable RL framework for pre-deployment modeling in ICU hypotension management.

Computational methods from reinforcement learning have shown promise in inferring treatment strategies for hypotension management and other clinical decision-making challenges. Unfortunately, the resulting models are often difficult for clinicians to interpret, making clinical inspection and validation of these computationally derived strategies challenging in advance of deployment. In this work, we develop a general framework for identifying succinct sets of clinical contexts in which clinicians make very different treatment choices, tracing the effects of those choices, and inferring a set of recommendations for those specific contexts. By focusing on these few key decision points, our framework produces succinct, interpretable treatment strategies that can each be easily visualized and verified by clinical experts. This interrogation process allows clinicians to leverage the model's use of historical data in tandem with their own expertise to determine which recommendations are worth investigating further e.g. at the bedside. We demonstrate the value of this approach via application to hypotension management in the ICU, an area with critical implications for patient outcomes that lacks data-driven individualized treatment strategies; that said, our framework has broad implications on how to use computational methods to assist with decision-making challenges on a wide range of clinical domains.

A systematic review of federated learning applications for biomedical data.

OBJECTIVES: Federated learning (FL) allows multiple institutions to collaboratively develop a machine learning algorithm without sharing their data. Organizations instead share model parameters only, allowing them to benefit from a model built with a larger dataset while maintaining the privacy of their own data. We conducted a systematic review to evaluate the current state of FL in healthcare and discuss the limitations and promise of this technology. METHODS: We conducted a literature search using PRISMA guidelines. At least two reviewers assessed each study for eligibility and extracted a predetermined set of data. The quality of each study was determined using the TRIPOD guideline and PROBAST tool. RESULTS: 13 studies were included in the full systematic review. Most were in the field of oncology (6 of 13; 46.1%), followed by radiology (5 of 13; 38.5%). The majority evaluated imaging results, performed a binary classification prediction task via offline learning (n = 12; 92.3%), and used a centralized topology, aggregation server workflow (n = 10; 76.9%). Most studies were compliant with the major reporting requirements of the TRIPOD guidelines. In all, 6 of 13 (46.2%) of studies were judged at high risk of bias using the PROBAST tool and only 5 studies used publicly available data. CONCLUSION: Federated learning is a growing field in machine learning with many promising uses in healthcare. Few studies have been published to date. Our evaluation found that investigators can do more to address the risk of bias and increase transparency by adding steps for data homogeneity or sharing required metadata and code.

Predicting hypoglycemia in critically Ill patients using machine learning and electronic health records.

Hypoglycemia is a common occurrence in critically ill patients and is associated with significant mortality and morbidity. We developed a machine learning model to predict hypoglycemia by using a multicenter intensive care unit (ICU) electronic health record dataset. Machine learning algorithms were trained and tested on patient data from the publicly available eICU Collaborative Research Database. Forty-four features including patient demographics, laboratory test results, medications, and vitals sign recordings were considered. The outcome of interest was the occurrence of a hypoglycemic event (blood glucose < 72 mg/dL) during a patient's ICU stay. Machine learning models used data prior to the second hour of the ICU stay to predict hypoglycemic outcome. Data from 61,575 patients who underwent 82,479 admissions at 199 hospitals were considered in the study. The best-performing predictive model was the eXtreme gradient boosting model (XGBoost), which achieved an area under the received operating curve (AUROC) of 0.85, a sensitivity of 0.76, and a specificity of 0.76. The machine learning model developed has strong discrimination and calibration for the prediction of hypoglycemia in ICU patients. Prospective trials of these models are required to evaluate their clinical utility in averting hypoglycemia within critically ill patient populations.

Artificial intelligence-based prediction of transfusion in the intensive care unit in patients with gastrointestinal bleeding.

OBJECTIVE: Gastrointestinal (GI) bleeding commonly requires intensive care unit (ICU) in cases of potentialhaemodynamiccompromise or likely urgent intervention. However, manypatientsadmitted to the ICU stop bleeding and do not require further intervention, including blood transfusion. The present work proposes an artificial intelligence (AI) solution for the prediction of rebleeding in patients with GI bleeding admitted to ICU. METHODS: A machine learning algorithm was trained and tested using two publicly available ICU databases, the Medical Information Mart for Intensive Care V.1.4 database and eICU Collaborative Research Database using freedom from transfusion as a proxy for patients who potentially did not require ICU-level care. Multiple initial observation time frames were explored using readily available data including labs, demographics and clinical parameters for a total of 20 covariates. RESULTS: The optimal model used a 5-hour observation period to achieve an area under the curve of the receiving operating curve (ROC-AUC) of greater than 0.80. The model was robust when tested against both ICU databases with a similar ROC-AUC for all. CONCLUSIONS: The potential disruptive impact of AI in healthcare innovation is acknowledge, but awareness of AI-related risk on healthcare applications and current limitations should be considered before implementation and deployment. The proposed algorithm is not meant to replace but to inform clinical decision making. Prospective clinical trial validation as a triage tool is warranted.