Proceedings of machine learning research最新文献

Obesity is a major public health concern. Multidisciplinary pediatric weight management programs are considered standard treatment for children with obesity who are not able to be successfully managed in the primary care setting. Despite their great potential, high dropout rates (referred to as attrition) are a major hurdle in delivering successful interventions. Predicting attrition patterns can help providers reduce the alarmingly high rates of attrition (up to 80%) by engaging in earlier and more personalized interventions. Previous work has mainly focused on finding static predictors of attrition on smaller datasets and has achieved limited success in effective prediction. In this study, we have collected a five-year comprehensive dataset of 4,550 children from diverse backgrounds receiving treatment at four pediatric weight management programs in the US. We then developed a machine learning pipeline to predict (a) the likelihood of attrition, and (b) the change in body-mass index (BMI) percentile of children, at different time points after joining the weight management program. Our pipeline is greatly customized for this problem using advanced machine learning techniques to process longitudinal data, smaller-size data, and interrelated prediction tasks. The proposed method showed strong prediction performance as measured by AUROC scores (average AUROC of 0.77 for predicting attrition, and 0.78 for predicting weight outcomes).

Protein engineering is currently being revolutionized by deep learning applications, especially through natural language processing (NLP) techniques. It has been shown that state-of-the-art self-supervised language models trained on entire protein databases capture hidden contextual and structural information in amino acid sequences and are capable of improving sequence-to-function predictions. Yet, recent studies have reported that current compound-protein modeling approaches perform poorly on learning interactions between enzymes and substrates of interest within one protein family. We attribute this to low-grade substrate encoding methods and over-compressed sequence representations received by downstream predictive models. In this study, we propose a new substrate-encoding based on Extended Connectivity Fingerprints (ECFPs) and a convolutional-pooling of the sequence embeddings. Through testing on an activity profiling dataset of haloalkanoate dehalogenase superfamily that measures activities of 218 phosphatases against 168 substrates, we show substantial improvements in predictive performances of compound-protein interaction modeling. In addition, we also test the workflow on three other datasets from the halogenase, kinase and aminotransferase families and show that our pipeline achieves good performance on these datasets as well. We further demonstrate the utility of this downstream model architecture by showing that it achieves good performance with six different protein embeddings, including ESM-1b (Rives et al., 2021), TAPE (Rao et al., 2019), ProtBert, ProtAlbert, ProtT5, and ProtXLNet (Elnaggar et al., 2021). This study provides a new workflow for activity prediction on novel substrates that can be used to engineer new enzymes for sustainability applications.

Deep neural networks (DNNs) have advanced our ability to take DNA primary sequence as input and predict a myriad of molecular activities measured via high-throughput functional genomic assays. Post hoc attribution analysis has been employed to provide insights into the importance of features learned by DNNs, often revealing patterns such as sequence motifs. However, attribution maps typically harbor spurious importance scores to an extent that varies from model to model, even for DNNs whose predictions generalize well. Thus, the standard approach for model selection, which relies on performance of a held-out validation set, does not guarantee that a high-performing DNN will provide reliable explanations. Here we introduce two approaches that quantify the consistency of important features across a population of attribution maps; consistency reflects a qualitative property of human interpretable attribution maps. We employ the consistency metrics as part of a multivariate model selection framework to identify models that yield high generalization performance and interpretable attribution analysis. We demonstrate the efficacy of this approach across various DNNs quantitatively with synthetic data and qualitatively with chromatin accessibility data.

Accelerated MRI reconstructs images of clinical anatomies from sparsely sampled signal data to reduce patient scan times. While recent works have leveraged deep learning to accomplish this task, such approaches have often only been explored in simulated environments where there is no signal corruption or resource limitations. In this work, we explore augmentations to neural network MRI image reconstructors to enhance their clinical relevancy. Namely, we propose a ConvNet model for detecting sources of image artifacts that achieves a classifier F ₂ score of 79.1%. We also demonstrate that training reconstructors on MR signal data with variable acceleration factors can improve their average performance during a clinical patient scan by up to 2%. We offer a loss function to overcome catastrophic forgetting when models learn to reconstruct MR images of multiple anatomies and orientations. Finally, we propose a method for using simulated phantom data to pre-train reconstructors in situations with limited clinically acquired datasets and compute capabilities. Our results provide a potential path forward for clinical adaptation of accelerated MRI.

Electronic Health Record modeling is crucial for digital medicine. However, existing models ignore higher-order interactions among medical codes and their causal relations towards downstream clinical predictions. To address such limitations, we propose a novel framework CACHE, to provide effective and insightful clinical predictions based on hypergraph representation learning and counterfactual and factual reasoning techniques. Experiments on two real EHR datasets show the superior performance of CACHE. Case studies with a domain expert illustrate a preferred capability of CACHE in generating clinically meaningful interpretations towards the correct predictions.

Cerebrovascular diseases are among the world's top causes of death and their screening and diagnosis rely on angiographic imaging. We focused on automated anatomical labeling of cerebral arteries that enables their cross-sectional quantification and inter-subject comparisons and thereby identification of geometric risk factors correlated to the cerebrovascular diseases. We used 152 cerebral TOF-MRA angiograms from three publicly available datasets and manually created reference labeling using Slicer3D. We extracted centerlines from nnU-net based segmentations using VesselVio and labeled them according to the reference labeling. Vessel centerline coordinates, in combination with additional vessel connectivity, radius and spatial context features were used for training seven distinct PointNet++ models. Model trained solely on the vessel centerline coordinates resulted in ACC of 0.93 and across-labels average TPR was 0.88. Including vessel radius significantly improved ACC to 0.95, and average TPR to 0.91. Finally, focusing spatial context to the Circle of Willis are resulted in best ACC of 0.96 and best average TPR of 0.93. Hence, using vessel radius and spatial context greatly improved vessel labeling, with the attained perfomance opening the avenue for clinical applications of intracranial vessel labeling.

An increasing amount of research is being devoted to applying machine learning methods to electronic health record (EHR) data for various clinical purposes. This growing area of research has exposed the challenges of the accessibility of EHRs. MIMIC is a popular, public, and free EHR dataset in a raw format that has been used in numerous studies. The absence of standardized preprocessing steps can be, however, a significant barrier to the wider adoption of this rare resource. Additionally, this absence can reduce the reproducibility of the developed tools and limit the ability to compare the results among similar studies. In this work, we provide a greatly customizable pipeline to extract, clean, and preprocess the data available in the fourth version of the MIMIC dataset (MIMIC-IV). The pipeline also presents an end-to-end wizard-like package supporting predictive model creations and evaluations. The pipeline covers a range of clinical prediction tasks which can be broadly classified into four categories - readmission, length of stay, mortality, and phenotype prediction. The tool is publicly available at https://github.com/healthylaife/MIMIC-IV-Data-Pipeline.