An official website of the United States government
Deep Learning (DL) methods have dramatically increased in popularity in recent years. While its initial success was demonstrated in the classification and manipulation of image data, there has been significant growth in the application of DL methods to problems in the biomedical sciences. However, the greater prevalence and complexity of missing data in biomedical datasets present significant challenges for DL methods. Here, we provide a formal treatment of missing data in the context of Variational Autoencoders (VAEs), a popular unsupervised DL architecture commonly used for dimension reduction, imputation, and learning latent representations of complex data. We propose a new VAE architecture, NIMIWAE, that is one of the first to flexibly account for both ignorable and non-ignorable patterns of missingness in input features at training time. Following training, samples can be drawn from the approximate posterior distribution of the missing data can be used for multiple imputation, facilitating downstream analyses on high dimensional incomplete datasets. We demonstrate through statistical simulation that our method outperforms existing approaches for unsupervised learning tasks and imputation accuracy. We conclude with a case study of an EHR dataset pertaining to 12,000 ICU patients containing a large number of diagnostic measurements and clinical outcomes, where many features are only partially observed.
more »
« less
Data Augmentation with Cross-Modal Variational Autoencoders (DACMVA) for Cancer Survival Prediction
The ability to translate Generative Adversarial Networks (GANs) and Variational Autoencoders (VAEs) into different modalities and data types is essential to improve Deep Learning (DL) for predictive medicine. This work presents DACMVA, a novel framework to conduct data augmentation in a cross-modal dataset by translating between modalities and oversampling imputations of missing data. DACMVA was inspired by previous work on the alignment of latent spaces in Autoencoders. DACMVA is a DL data augmentation pipeline that improves the performance in a downstream prediction task. The unique DACMVA framework leverages a cross-modal loss to improve the imputation quality and employs training strategies to enable regularized latent spaces. Oversampling of augmented data is integrated into the prediction training. It is empirically demonstrated that the new DACMVA framework is effective in the often-neglected scenario of DL training on tabular data with continuous labels. Specifically, DACMVA is applied towards cancer survival prediction on tabular gene expression data where there is a portion of missing data in a given modality. DACMVA significantly (p << 0.001, one-sided Wilcoxon signed-rank test) outperformed the non-augmented baseline and competing augmentation methods with varying percentages of missing data (4%, 90%, 95% missing). As such, DACMVA provides significant performance improvements, even in very-low-data regimes, over existing state-of-the-art methods, including TDImpute and oversampling alone.
more »
« less
- Award ID(s):
- 1944247
- PAR ID:
- 10502312
- Publisher / Repository:
- MDPI
- Date Published:
- Journal Name:
- Information
- Volume:
- 15
- Issue:
- 1
- ISSN:
- 2078-2489
- Page Range / eLocation ID:
- 7
- Format(s):
- Medium: X
- Sponsoring Org:
- National Science Foundation
More Like this
-
-
Abstract. We introduce physics-informed multimodal autoencoders (PIMA) - a variational inference framework for discovering shared information in multimodal datasets. Individual modalities are embedded into a shared latent space and fused through a product-of-experts formulation, enabling a Gaussian mixture prior to identify shared features. Sampling from clusters allows cross-modal generative modeling, with a mixture-of-experts decoder that imposes inductive biases from prior scientific knowledge and thereby imparts structured disentanglement of the latent space. This approach enables cross-modal inference and the discovery of features in high-dimensional heterogeneous datasets. Consequently, this approach provides a means to discover fingerprints in multimodal scientific datasets and to avoid traditional bottlenecks related to high-fidelity measurement and characterization of scientificmore » « less
-
Abstract A fundamental challenge in diagnostics is integrating multiple modalities to develop a joint characterization of physiological state. Using the heart as a model system, we develop a cross-modal autoencoder framework for integrating distinct data modalities and constructing a holistic representation of cardiovascular state. In particular, we use our framework to construct such cross-modal representations from cardiac magnetic resonance images (MRIs), containing structural information, and electrocardiograms (ECGs), containing myoelectric information. We leverage the learned cross-modal representation to (1) improve phenotype prediction from a single, accessible phenotype such as ECGs; (2) enable imputation of hard-to-acquire cardiac MRIs from easy-to-acquire ECGs; and (3) develop a framework for performing genome-wide association studies in an unsupervised manner. Our results systematically integrate distinct diagnostic modalities into a common representation that better characterizes physiologic state.more » « less
-
Multi-modal learning by means of leveraging both 2D graph and 3D point cloud information has become a prevalent method to improve model performance in molecular property prediction. However, many recent techniques focus on specific pre-training tasks such as contrastive learning, feature blending, and atom/subgraph masking in order to learn multi-modality even though design of model architecture is also impactful for both pre-training and downstream task performance. Relying on pre-training tasks to align 2D and 3D modalities lacks direct interaction which may be more effective in multimodal learning. In this work, we propose MolInteract, which takes a simple yet effective architecture-focused approach to multimodal molecule learning which addresses these challenges. MolInteract leverages an interaction layer for fusing 2D and 3D information and fostering cross-modal alignment, showing strong results using even the simplest pre-training methods such as predicting features of the 3D point cloud and 2D graph. MolInteract exceeds state-of-the-art multimodal pre-training techniques and architectures on various downstream 2D and 3D molecule property prediction benchmark tasks.more » « less
-
Learning multimodal representations is a fundamentally complex research problem due to the presence of multiple heterogeneous sources of information. Although the presence of multiple modalities provides additional valuable information, there are two key challenges to address when learning from multimodal data: 1) models must learn the complex intra-modal and cross-modal interactions for prediction and 2) models must be robust to unexpected missing or noisy modalities during testing. In this paper, we propose to optimize for a joint generative-discriminative objective across multimodal data and labels. We introduce a model that factorizes representations into two sets of independent factors: multimodal discriminative and modality-specific generative factors. Multimodal discriminative factors are shared across all modalities and contain joint multimodal features required for discriminative tasks such as sentiment prediction. Modality-specific generative factors are unique for each modality and contain the information required for generating data. Experimental results show that our model is able to learn meaningful multimodal representations that achieve state-of-the-art or competitive performance on six multimodal datasets. Our model demonstrates flexible generative capabilities by conditioning on independent factors and can reconstruct missing modalities without significantly impacting performance. Lastly, we interpret our factorized representations to understand the interactions that influence multimodal learning.more » « less
- Free Publicly Accessible Full Text
-
Accepted Manuscript1.0
- Journal Article:
- https://doi.org/10.3390/info15010007
