Creation of a Robust and Generalizable Machine Learning Classifier for Patient Ventilator Asynchrony

BACKGROUND

As healthcare increasingly digitizes, streaming waveform data is being made available from an variety of sources, but there still remains a paucity of performant clinical decision support systems. For example, in the intensive care unit (ICU) existing automated alarm systems typically rely on simple thresholding that result in frequent false positives. Recurrent false positive alerts create distrust of alarm mechanisms that can be directly detrimental to patient health. To improve patient care in the ICU, we need alert systems that are both pervasive, and accurate so as to be informative and trusted by providers.

OBJECTIVE

We aimed to develop a machine learning-based classifier to detect abnormal waveform events using the use case of mechanical ventilation waveform analysis, and the detection of harmful forms of ventilation delivery to patients. We specifically focused on detecting injurious subtypes of patient-ventilator asynchrony (PVA).

METHODS

Using a dataset of breaths recorded from 35 different patients, we used machine learning to create computational models to automatically detect, and classify two types of injurious PVA, double trigger asynchrony (DTA), breath stacking asynchrony (BSA). We examined the use of synthetic minority over-sampling technique (SMOTE) to overcome class imbalance problems, varied methods for feature selection, and use of ensemble methods to optimize the performance of our model.

RESULTS

We created an ensemble classifier that is able to accurately detect DTA at a sensitivity/specificity of 0.960/0.975, BSA at sensitivity/specificity of 0.944/0.987, and non-PVA events at sensitivity/specificity of .967/.980.

CONCLUSIONS

Our results suggest that it is possible to create a high-performing machine learning-based model for detecting PVA in mechanical ventilator waveform data in spite of both intra-patient, and inter-patient variability in waveform patterns, and the presence of clinical artifacts like cough and suction procedures. Our work highlights the importance of addressing class imbalance in clinical data sets, and the combined use of statistical methods and expert knowledge in feature selection.

Leveraging IoTs and Machine Learning for Patient Diagnosis and Ventilation Management in the Intensive Care Unit

Future healthcare systems will rely heavily on clinical decision support systems (CDSS) to improve the decision-making processes of clinicians. To explore the design of future CDSS, we developed a research-focused CDSS for the management of patients in the intensive care unit that leverages Internet of Things (IoT) devices capable of collecting streaming physiologic data from ventilators and other medical devices. We then created machine learning (ML) models that could analyze the collected physiologic data to determine if the ventilator was delivering potentially harmful therapy and if a deadly respiratory condition, acute respiratory distress syndrome (ARDS), was present. We also present work to aggregate these models into a mobile application that can provide responsive, real-time alerts of changes in ventilation to providers. As illustrated in the recent COVID-19 pandemic, being able to accurately predict ARDS in newly infected patients can assist in prioritizing care. We show that CDSS may be used to analyze physiologic data for clinical event recognition and automated diagnosis, and we also highlight future research avenues for hospital CDSS.

Predicting ASD Diagnosis in Children with Synthetic and Image-based Eye Gaze Data

As early intervention is highly effective for young children with autism spectrum disorder (ASD), it is imperative to make accurate diagnosis as early as possible. ASD has often been associated with atypical visual attention and eye gaze data can be collected at a very early age. An automatic screening tool based on eye gaze data that could identify ASD risk offers the opportunity for intervention before the full set of symptoms is present. In this paper, we propose two machine learning methods, synthetic saccade approach and image based approach, to automatically classify ASD given children's eye gaze data collected from free-viewing tasks of natural images. The first approach uses a generative model of synthetic saccade patterns to represent the baseline scan-path from a typical non-ASD individual and combines it with the real scan-path as well as other auxiliary data as inputs to a deep learning classifier. The second approach adopts a more holistic image-based approach by feeding the input image and a sequence of fixation maps into a convolutional or recurrent neural network. Using a publicly-accessible collection of children's gaze data, our experiments indicate that the ASD prediction accuracy reaches 67.23% accuracy on the validation dataset and 62.13% accuracy on the test dataset.

Joint Semi-supervised and Active Learning for Segmentation of Gigapixel Pathology Images with Cost-Effective Labeling

The need for manual and detailed annotations limits the applicability of supervised deep learning algorithms in medical image analyses, specifically in the field of pathology. Semi-supervised learning (SSL) provides an effective way for leveraging unlabeled data to relieve the heavy reliance on the amount of labeled samples when training a model. Although SSL has shown good performance, the performance of recent state-of-the-art SSL methods on pathology images is still under study. The problem for selecting the most optimal data to label for SSL is not fully explored. To tackle this challenge, we propose a semi-supervised active learning framework with a region-based selection criterion. This framework iteratively selects regions for annotation query to quickly expand the diversity and volume of the labeled set. We evaluate our framework on a grey-matter/white-matter segmentation problem using gigapixel pathology images from autopsied human brain tissues. With only 0.1% regions labeled, our proposed algorithm can reach a competitive IoU score compared to fully-supervised learning and outperform the current state-of-the-art SSL by more than 10% of IoU score and DICE coefficient.

Preanalytic variable effects on segmentation and quantification machine learning algorithms for amyloid-β analyses on digitized human brain slides

Computational machine learning (ML)-based frameworks could be advantageous for scalable analyses in neuropathology. A recent deep learning (DL) framework has shown promise in automating the processes of visualizing and quantifying different types of amyloid-β deposits as well as segmenting white matter (WM) from gray matter (GM) on digitized immunohistochemically stained slides. However, this framework has only been trained and evaluated on amyloid-β-stained slides with minimal changes in preanalytic variables. In this study, we evaluated select preanalytical variables including magnification, compression rate, and storage format using three digital slides scanners (Zeiss Axioscan Z1, Leica Aperio AT2, and Leica Aperio GT 450), on over 60 whole slide images, in a cohort of 14 cases having a spectrum of amyloid-β deposits. We conducted statistical comparisons of preanalytic variables with repeated measures analysis of variance evaluating the outputs of two DL frameworks for segmentation and object classification tasks. For both WM/GM segmentation and amyloid-β plaque classification tasks, there were statistical differences with respect to scanner types (p < 0.05) and magnifications (p < 0.05). Although small numbers of cases were analyzed, this pilot study highlights the significance of preanalytic variables that may alter the performance of ML algorithms.

Convolution Neural Network for Pain Intensity Assessment from Facial Expression

Pain is an unpleasant feeling that can reflect a patient's health situation. Since measuring pain is subjective, time-consuming, and needs continuous monitoring, automated pain intensity detection from facial expression holds great potential for smart healthcare applications. Convolutional Neural Networks (CNNs) are recently being used to identify features, map and model pain intensity from facial images, delivering great promise in helping practitioners detect disease. Limited research has been conducted to determine pain intensity levels across multiple classes. CNNs with simple learning schemes are limited in their ability to extract feature information from images. In order to develop a highly accurate pain intensity estimation system, this study proposes a Deep CNN (DCNN) model using the transfer learning technique, where a pre-trained DCNN model is adopted by replacing its dense upper layers, and the model is tuned using painful facial. We conducted experiments on the UNBC-McMaster shoulder pain archive database to estimate pain intensity in terms of seven-level thresholds using a given facial expression image. The experiments show our method achieves a promising improvement in terms of accuracy and performance to estimate pain intensity and outperform the-state-of-the-arts models.

Machine Learning Based Autism Spectrum Disorder Detection from Videos

Early diagnosis of Autism Spectrum Disorder (ASD) is crucial for best outcomes to interventions. In this paper, we present a machine learning (ML) approach to ASD diagnosis based on identifying specific behaviors from videos of infants of ages 6 through 36 months. The behaviors of interest include directed gaze towards faces or objects of interest, positive affect, and vocalization. The dataset consists of 2000 videos of 3-minute duration with these behaviors manually coded by expert raters. Moreover, the dataset has statistical features including duration and frequency of the above mentioned behaviors in the video collection as well as independent ASD diagnosis by clinicians. We tackle the ML problem in a two-stage approach. Firstly, we develop deep learning models for automatic identification of clinically relevant behaviors exhibited by infants in a one-on-one interaction setting with parents or expert clinicians. We report baseline results of behavior classification using two methods: (1) image based model (2) facial behavior features based model. We achieve 70% accuracy for smile, 68% accuracy for look face, 67% for look object and 53% accuracy for vocalization. Secondly, we focus on ASD diagnosis prediction by applying a feature selection process to identify the most significant statistical behavioral features and a over and under sampling process to mitigate the class imbalance, followed by developing a baseline ML classifier to achieve an accuracy of 82% for ASD diagnosis.

Peeking into the invitation-based adoption process of OSN-based applications

Online social network (OSN) based applications often rely on user interactions to propagate information or to recruit more users, producing a sequence of user actions called adoption process or cascades. This paper presents the first attempt to quantitatively study the adoption process or cascade of such OSN-based applications by analyzing detailed user activity data from a popular Facebook gifting application. In particular, due to the challenge of monitoring user interactions over all possible channels on OSN platforms, we focus on characterizing the adoption process that relies only on user-based invitation (which is applicable to most gifting applications). We characterize the adoptions by tracking the invitations sent by the existing users to their friends through the Facebook gifting application and the events when their friends install the application for the first time. We found that a small number of big cascades carry the adoption of most of the application users. Contrary to common beliefs, we did not observe special influential nodes that are responsible for the viral adoption of the application.

Physiowise: A Physics-aware Approach to Dicrotic Notch Identification

Dicrotic Notch (DN), one of the most significant and indicative features of the arterial blood pressure (ABP) waveform, becomes less pronounced and thus harder to identify as a matter of aging and pathological vascular stiffness. Generalizable and automatic DN identification for such edge cases is even more challenging in the presence of unexpected ABP waveform deformations that happen due to internal and external noise sources or pathological conditions that cause hemodynamic instability. We propose a physics-aware approach, named Physiowise (PW), that first employs a cardiovascular model to augment the original ABP waveform and reduce unexpected deformations, then apply a set of predefined rules on the augmented signal to find DN locations. We have tested the proposed method on in-vivo data gathered from 14 pigs under hemorrhage and sepsis study. Our result indicates 52% overall mean error improvement with 16% higher detection accuracy within the lowest permitted error range of 30ms. An additional hybrid methodology is also proposed to allow combining augmentation with any application-specific user-defined rule set.

Enhanced Critical Congenital Cardiac Disease Screening by Combining Interpretable Machine Learning Algorithms

Critical Congenital Heart Disease (CCHD) screening that only uses oxygen saturation (SpO2), measured by pulse oximetry, fails to detect an estimated 900 US newborns annually. The addition of other pulse oximetry features such as perfusion index (PIx), heart rate, pulse delay and photoplethysmography characteristics may improve detection of CCHD, especially those with systemic blood flow obstruction such as Coarctation of the Aorta (CoA). To comprehensively study the most relevant features associated with CCHD, we investigated interpretable machine learning (ML) algorithms by using Recursive Feature Elimination (RFE) to identify an optimal subset of features. We then incorporated the trained ML models into the current SpO2-alone screening algorithm. Our proposed enhanced CCHD screening system, which adds the ML model, improved sensitivity by approximately 10 percentage points compared to the current standard SpO2-alone method with minimal to no impact on specificity.Clinical relevance- This establishes proof of concept for a ML algorithm that combines pulse oximetry features to improve detection of CCHD with little impact on false positive rate.

A novel system to collect dual pulse oximetry data for critical congenital heart disease screening research

INTRODUCTION

Access to patient medical data is critical to building a real-time data analytic pipeline for improving care providers' ability to detect, diagnose, and prognosticate diseases. Critical congenital heart disease (CCHD) is a common group of neonatal life-threatening defects that must be promptly diagnosed to minimize morbidity and mortality. CCHD can be diagnosed both prenatally and postnatally. However, despite current screening practices involving oxygen saturation analysis, timely diagnosis is missed in approximately 900 infants with CCHD annually in the USA and can benefit from increased data processing capabilities. Adding non-invasive perfusion measurements to oxygen saturation data can improve the timeliness and fidelity of CCHD diagnostics. However, real-time monitoring and interpretation of non-invasive perfusion data are currently limited.

METHODS

To address this challenge, we created a hardware and software architecture utilizing a Pi-top™ for collecting, visualizing, and storing dual oxygen saturation, perfusion indices, and photoplethysmography data. Data aggregation in our system is automated and all data files are coded with unique study identifiers to facilitate research purposes.

RESULTS

Using this system, we have collected data from 190 neonates, 130 presumably without and 60 with congenital heart disease, in total comprising 1665 min of information. From these data, we are able to extract non-invasive perfusion features such as perfusion index, radiofemoral delay, and slope of systolic rise or diastolic fall.

CONCLUSION

This data collection and waveform analysis is relatively inexpensive and can be used to enhance future CCHD screening algorithms.

A Semi-supervised Learning for Segmentation of Gigapixel Histopathology Images from Brain Tissues

Automated segmentation of grey matter (GM) and white matter (WM) in gigapixel histopathology images is advantageous to analyzing distributions of disease pathologies, further aiding in neuropathologic deep phenotyping. Although supervised deep learning methods have shown good performance, its requirement of a large amount of labeled data may not be cost-effective for large scale projects. In the case of GM/WM segmentation, trained experts need to carefully trace the delineation in gigapixel images. To minimize manual labeling, we consider semi-surprised learning (SSL) and deploy one state-of-the-art SSL method (FixMatch) on WSIs. Then we propose a two-stage scheme to further improve the performance of SSL: the first stage is a self-supervised module to train an encoder to learn the visual representations of unlabeled data, subsequently, this well-trained encoder will be an initialization of consistency loss-based SSL in the second stage. We test our method on Amyloid-β stained histopathology images and the results outperform FixMatch with the mean IoU score at around 2% by using 6,000 labeled tiles while over 10% by using only 600 labeled tiles from 2 WSIs.Clinical relevance- this work minimizes the required labeling efforts by trained personnel. An improved GM/WM segmentation method could further aid in the study of brain diseases, such as Alzheimer's disease.

BrainSec: Automated Brain Tissue Segmentation Pipeline for Scalable Neuropathological Analysis

As neurodegenerative disease pathological hallmarks have been reported in both grey matter (GM) and white matter (WM) with different density distributions, automating the segmentation process of GM/WM would be extremely advantageous for aiding in neuropathologic deep phenotyping. Standard segmentation methods typically involve manual annotations, where a trained researcher traces the delineation of GM/WM in ultra-high-resolution Whole Slide Images (WSIs). This method can be time-consuming and subjective, preventing a scalable analysis on pathology images. This paper proposes an automated segmentation pipeline (BrainSec) combining a Convolutional Neural Network (CNN) module for segmenting GM/WM regions and a post-processing module to remove artifacts/residues of tissues. The final output generates XML annotations that can be visualized via Aperio ImageScope. First, we investigate two baseline models for medical image segmentation: FCN, and U-Net. Then we propose a patch-based approach, BrainSec, to classify the GM/WM/background regions. We demonstrate BrainSec is robust and has reliable performance by testing it on over 180 WSIs that incorporate numerous unique cases as well as distinct neuroanatomic brain regions. We also apply gradient-weighted class activation mapping (Grad-CAM) to interpret the segmentation masks and provide relevant explanations and insights. In addition, we have integrated BrainSec with an existing Amyloid-β pathology classification model into a unified framework (without incurring significant computation complexity) to identify pathologies, visualize their distributions, and quantify each type of pathologies in segmented GM/WM regions, respectively.

DCAP

This paper proposes a detection mechanism called DCAP for a network provider to monitor incoming traffic and identify misbehaving flows without having to keep per-flow accounting at any of its routers. Misbehaving flows refer to flows that exceed their stipulated bandwidth limit. Through collaborative aggregate policing at both ingress and egress nodes, DCAP is able to quickly narrow the search to a candidate group that contains the misbehaving flows, and eventually identify the individual culprits. In comparison to per-flow policing, the amount of state maintained at an edge router is reduced from O ( n ) to O (√ n ), where n is the number of admitted flows. Simulation results show that DCAP can successfully detect a majority (64--83%) of the misbehaving flows with almost zero false alarms. Packet losses suffered by innocent flows due to undetected misbehaving activity are insignificant (0.02--0.9%). We also successfully build a prototype that demonstrates how DCAP can be deployed with minimal processing overhead in a soft-QoS architecture.

ProgME

Traffic measurements provide critical input for a wide range of network management applications, including traffic engineering, accounting, and security analysis. Existing measurement tools collect traffic statistics based on some pre-determined, inflexible concept of "flows". They do not have sufficient built-in intelligence to understand the application requirements or adapt to the traffic conditions. Consequently, they have limited scalability with respect to the number of flows and the heterogeneity of monitoring applications.

We present ProgME, a Prog rammable ME asurement architecture based on a novel concept of flowset - arbitrary set of flows defined according to application requirements and/or traffic conditions. Through a simple flowset composition language, ProgME can incorporate application requirements, adapt itself to circumvent the challenges on scalability posed by the large number of flows, and achieve a better application-perceived accuracy. ProgME can analyze and adapt to traffic statistics in real-time. Using sequential hypothesis test, ProgME can achieve fast and scalable heavy hitter identification.

PREDICTING AUTISM DIAGNOSIS USING IMAGE WITH FIXATIONS AND SYNTHETIC SACCADE PATTERNS

Signs of autism spectrum disorder (ASD) emerge in the first year of life in many children, but diagnosis is typically made much later, at an average age of 4 years in the United States. Early intervention is highly effective for young children with ASD, but is typically reserved for children with a formal diagnosis, making accurate identification as early as possible imperative. A screening tool that could identify ASD risk during infancy offers the opportunity for intervention before the full set of symptoms is present. In this paper, we propose two machine learning methods, synthetic saccade approach and image based approach, to automatically classify ASD given the scanpath data from children on free viewing of natural images. The first approach uses a generative model of synthetic saccade patterns to represent the baseline scan-path from a typical non-ASD individual and combines it with the input scanpath as well as other auxiliary data as inputs to a deep learning classifier. The second approach adopts a more holistic image based approach by feeding the input image and a sequence of fixation maps into a state-of-the-art convolutional neural network. Our experiments indicate that we can get 65.41% accuracy on the validation dataset.

Minimizing L 1 over L 2 norms on the gradient

In this paper, we study the L ₁ /L ₂ minimization on the gradient for imaging applications. Several recent works have demonstrated that L ₁ /L ₂ is better than the L ₁ norm when approximating the L ₀ norm to promote sparsity. Consequently, we postulate that applying L ₁ /L ₂ on the gradient is better than the classic total variation (the L ₁ norm on the gradient) to enforce the sparsity of the image gradient. Numerically, we design a specific splitting scheme, under which we can prove subsequential and global convergence for the alternating direction method of multipliers (ADMM) under certain conditions. Experimentally, we demonstrate visible improvements of L ₁ /L ₂ over L ₁ and other nonconvex regularizations for image recovery from low-frequency measurements and two medical applications of MRI and CT reconstruction. Finally, we reveal some empirical evidence on the superiority of L ₁ /L ₂ over L ₁ when recovering piecewise constant signals from low-frequency measurements to shed light on future works.

Differentially Private Generative Adversarial Networks with Model Inversion

To protect sensitive data in training a Generative Adversarial Network (GAN), the standard approach is to use differentially private (DP) stochastic gradient descent method in which controlled noise is added to the gradients. The quality of the output synthetic samples can be adversely affected and the training of the network may not even converge in the presence of these noises. We propose Differentially Private Model Inversion (DPMI) method where the private data is first mapped to the latent space via a public generator, followed by a lower-dimensional DP-GAN with better convergent properties. Experimental results on standard datasets CIFAR10 and SVHN as well as on a facial landmark dataset for Autism screening show that our approach outperforms the standard DP-GAN method based on Inception Score, Frechet Inception Distance, and classification accuracy under the same privacy guarantee.

How cy pres promotes transdisciplinary convergence science: an academic health center for women's cardiovascular and brain health

Cardiovascular disease (CVD) is largely preventable, and the leading cause of death for men and women. Though women have increased life expectancy compared to men, there are marked sex disparities in prevalence and risk of CVD-associated mortality and dementia. Yet, the basis for these and female-male differences is not completely understood. It is increasingly recognized that heart and brain health represent a lifetime of exposures to shared risk factors (including obesity, hyperlipidemia, diabetes, and hypertension) that compromise cerebrovascular health. We describe the process and resources for establishing a new research Center for Women's Cardiovascular and Brain Health at the University of California, Davis as a model for: (1) use of the cy pres principle for funding science to improve health; (2) transdisciplinary collaboration to leapfrog progress in a convergence science approach that acknowledges and addresses social determinants of health; and (3) training the next generation of diverse researchers. This may serve as a blueprint for future Centers in academic health institutions, as the cy pres mechanism for funding research is a unique mechanism to leverage residual legal settlement funds to catalyze the pace of scientific discovery, maximize innovation, and promote health equity in addressing society's most vexing health problems.

Gait Event Detection and Travel Distance Using Waist-Worn Accelerometers across a Range of Speeds: Automated Approach

Estimation of temporospatial clinical features of gait (CFs), such as step count and length, step duration, step frequency, gait speed, and distance traveled, is an important component of community-based mobility evaluation using wearable accelerometers. However, accurate unsupervised computerized measurement of CFs of individuals with Duchenne muscular dystrophy (DMD) who have progressive loss of ambulatory mobility is difficult due to differences in patterns and magnitudes of acceleration across their range of attainable gait velocities. This paper proposes a novel calibration method. It aims to detect steps, estimate stride lengths, and determine travel distance. The approach involves a combination of clinical observation, machine-learning-based step detection, and regression-based stride length prediction. The method demonstrates high accuracy in children with DMD and typically developing controls (TDs) regardless of the participant's level of ability. Fifteen children with DMD and fifteen TDs underwent supervised clinical testing across a range of gait speeds using 10 m or 25 m run/walk (10 MRW, 25 MRW), 100 m run/walk (100 MRW), 6-min walk (6 MWT), and free-walk (FW) evaluations while wearing a mobile-phone-based accelerometer at the waist near the body's center of mass. Following calibration by a trained clinical evaluator, CFs were extracted from the accelerometer data using a multi-step machine-learning-based process and the results were compared to ground-truth observation data. Model predictions vs. observed values for step counts, distance traveled, and step length showed a strong correlation (Pearson's r = -0.9929 to 0.9986, p < 0.0001). The estimates demonstrated a mean (SD) percentage error of 1.49% (7.04%) for step counts, 1.18% (9.91%) for distance traveled, and 0.37% (7.52%) for step length compared to ground-truth observations for the combined 6 MWT, 100 MRW, and FW tasks. Our study findings indicate that a single waist-worn accelerometer calibrated to an individual's stride characteristics using our methods accurately measures CFs and estimates travel distances across a common range of gait speeds in both DMD and TD peers.

A Machine Learning Driven Pipeline for Automated Photoplethysmogram Signal Artifact Detection

Recent advances in Critical Congenital Heart Disease (CCHD) research using Photoplethysmography (PPG) signals have yielded an Internet of Things (IoT) based enhanced screening method that performs CCHD detection comparable to SpO2 screening. The use of PPG signals, however, poses a challenge due to its measurements being prone to artifacts. To comprehensively study the most effective way to remove the artifact segments from PPG waveforms, we performed feature engineering and investigated both Machine Learning (ML) and rule based algorithms to identify the optimal method of artifact detection. Our proposed artifact detection system utilizes a 3-stage ML model that incorporates both Gradient Boosting (GB) and Random Forest (RF). The proposed system achieved 84.01% of Intersection over Union (IoU), which is competitive to state-of-the-art artifact detection methods tested on higher resolution PPG.

Improving the precision of shock resuscitation by predicting fluid responsiveness with machine learning and arterial blood pressure waveform data

Fluid bolus therapy (FBT) is fundamental to the management of circulatory shock in critical care but balancing the benefits and toxicities of FBT has proven challenging in individual patients. Improved predictors of the hemodynamic response to a fluid bolus, commonly referred to as a fluid challenge, are needed to limit non-beneficial fluid administration and to enable automated clinical decision support and patient-specific precision critical care management. In this study we retrospectively analyzed data from 394 fluid boluses from 58 pigs subjected to either hemorrhagic or distributive shock. All animals had continuous blood pressure and cardiac output monitored throughout the study. Using this data, we developed a machine learning (ML) model to predict the hemodynamic response to a fluid challenge using only arterial blood pressure waveform data as the input. A Random Forest binary classifier referred to as the ML fluid responsiveness algorithm (MLFRA) was trained to detect fluid responsiveness (FR), defined as a ≥ 15% change in cardiac stroke volume after a fluid challenge. We then compared its performance to pulse pressure variation, a commonly used metric of FR. Model performance was assessed using the area under the receiver operating characteristic curve (AUROC), confusion matrix metrics, and calibration curves plotting predicted probabilities against observed outcomes. Across multiple train/test splits and feature selection methods designed to assess performance in the setting of small sample size conditions typical of large animal experiments, the MLFRA achieved an average AUROC, recall (sensitivity), specificity, and precision of 0.82, 0.86, 0.62. and 0.76, respectively. In the same datasets, pulse pressure variation had an AUROC, recall, specificity, and precision of 0.73, 0.91, 0.49, and 0.71, respectively. The MLFRA was generally well-calibrated across its range of predicted probabilities and appeared to perform equally well across physiologic conditions. These results suggest that ML, using only inputs from arterial blood pressure monitoring, may substantially improve the accuracy of predicting FR compared to the use of pulse pressure variation. If generalizable, these methods may enable more effective, automated precision management of critically ill patients with circulatory shock.