Correcting biases in deep learning models

纠正深度学习模型中的偏差

• 批准号: 10584314
• 负责人: Albert Amos Montillo
• 金额: $ 34.44万
• 依托单位: UT SOUTHWESTERN MEDICAL CENTER
• 依托单位国家: 美国
• 财政年份: 2023
• 资助国家: 美国
• 起止时间: 2023-01-20 至 2027-12-31
• 项目状态: 未结题
• 来源: https://reporter.nih.gov/project-details/10584314
• 关键词: 3-Dimensional; Accounting; Address; Algorithms; Alzheimer' s Disease; Alzheimer' s disease diagnosis; Architecture; Artificial Intelligence; Automation; Biological; Biological Assay; Biological Sciences; Blood; Brain; Cells; Classification; Code; Cognitive; Communities; Complex; Computer Models; Computer software; Data; Data Analyses; Data Correlations; Data Set; Dependence; Dimensions; Future; Generations; Glioma; Goals; Health Sciences; Image; Individual; Institution; Learning; Machine Learning; Magnetic Resonance Imaging; Measurement; Measures; Medical; Medical Imaging; Methods; Methylation; Microscope; Microscopy; Modeling; Modification; Nerve Degeneration; Outcome; Participant; Patients; Performance; Peripheral Blood Mononuclear Cell; Pythons; Recovery; Sampling; Site; TensorFlow; Testing; The Cancer Imaging Archive; Three-Dimensional Image; Tissues; Training; Validation; Visualization; autoencoder; base; biomedical imaging; cancer cell; convolutional neural network; deep learning; deep learning model; feedforward neural network; human disease; improved; innovation; learning community; learning network; live cell imaging; machine learning framework; mild cognitive impairment; multimodality; neural network; neural network architecture; novel; open source; predictive modeling; preservation; resilience; statistics; success; tool; transcriptome sequencing; vector

Project Summary/Abstract Deep learning (DL) has been widely applied across all life sciences to construct predictive models. However, it relies on the assumption that training samples are independent and identically distributed. This is frequently violated in the life sciences, where data is “grouped” by measurements from the same sample (patient, cell, tissue), by the same observer, or at the same site. This leads to clusters of correlated data (random effects), and when the models are fit to such data, the model parameters can be severely biased, leading to type I and II errors. Proper accounting for such dependencies in DL models has gone unsolved. The objective of this proposal is to develop the appropriate DL modifications to separately model global fixed effects and random effects that increase model interpretability and performance for precise unbiased predictions related to human disease. Our proposal is based on a novel, model-agnostic framework to transform conventional DL models into proper mixed effects DL (MEDL) models. This affords capabilities of statistical linear mixed effects models, including the separation of cluster-invariant fixed effects from cluster-specific random effects, while preserving the ability of DL to learn data-driven nonlinear associations. The core premise is that proper MEDL models 1) are more resilient to confounding effects and more attentive to true predictive features, 2) can capture, quantify, and visualize random effects to enhance interpretability, and 3) attain better generalization to new clusters. We propose to incorporate MEDL into three of the most important DL model types including dense feed-forward neural networks (DFNNs), convolutional neural networks (CNNs), and autoencoders. Our preliminary results demonstrate multiple advantages of MEDL over conventional DL in both accuracy and interpretability. MEDL outperforms previous clustered data approaches including: domain adversarial models, meta-learning, and the inclusion of cluster membership as an input covariate. We developed an ME-DFNN to predict conversion from mild cognitive impairment to Alzheimer’s Disease (AD) from tabular data, an ME-CNN to diagnose AD from MRI, and an ME-autoencoder to compress and classify live cell images. Across these test cases, MEDL models were the most discriminative between known confounded and real features; they were able to quantify or visualize the random effects and outperformed other models on clusters both seen and unseen during training. This proposal further develops the methods to handle complex architectures and hierarchical effects, with external validation, through these aims: 1) Develop ME-DFNNs for classification and regression. 2) Develop 3D ME-CNNs and multi- modal 3D ME-CNNs for medical image classification. 3) Develop convolutional and vector ME-autoencoders for image and omics data. We describe the innovative incorporation of an adversarial classifier to constrain the base model to learn fixed effects, a Bayesian random effects subnetwork, and an approach to apply random effects to unseen clusters. All these solutions will be released as open source software that improve existing DL models to ultimately support precision biomedicine for the study and treatment of human disease.

项目总结/摘要 深度学习（DL）已广泛应用于所有生命科学领域来构建预测模型。但 依赖于训练样本是独立且同分布的假设。这经常被 在生命科学中，数据通过来自相同样品（患者，细胞， 组织），由同一观察者，或在同一部位。这会导致相关数据的集群（随机效应）， 当模型与这些数据拟合时，模型参数可能会严重偏差，导致I型和II型 错误.在深度学习模型中对这种依赖性的正确解释还没有解决。本提案的目的 是开发适当的DL修改，以分别建模全局固定效应和随机效应， 提高模型的可解释性和性能，以实现与人类疾病相关的精确无偏预测。 我们的建议是基于一个新的，模型不可知的框架，将传统的DL模型转换为适当的 混合效应DL（MEDL）模型。这提供了统计线性混合效应模型的功能，包括 将集群不变的固定效应与集群特定的随机效应分离，同时保留 学习数据驱动的非线性关联。核心前提是适当的MEDL模型1）更多 对混杂效应有弹性，更关注真正的预测特征，2）可以捕获，量化， 可视化随机效应以增强可解释性，以及3）获得对新聚类的更好的泛化。我们 建议将MEDL纳入三种最重要的DL模型类型，包括密集前馈 神经网络（DFNN）、卷积神经网络（CNN）和自动编码器。我们的初步结果 在准确性和可解释性方面，MEDL比传统DL具有多个优势。MEDL 优于以前的聚类数据方法，包括：域对抗模型，元学习， 包含聚类成员资格作为输入协变量。我们开发了一个ME-DFNN来预测从 从表格数据中轻度认知障碍到阿尔茨海默病（AD），从MRI诊断AD的ME-CNN， 以及ME自动编码器，用于对活细胞图像进行压缩和分类。在这些测试用例中，MEDL模型 已知混淆和真实的特征之间最具区分力;他们能够量化或可视化 随机效应，并且在训练过程中看到和看不到的集群上都优于其他模型。这项建议 进一步发展的方法来处理复杂的架构和层次效应，与外部验证， 通过这些目标：1）开发用于分类和回归的ME-DFNN。2)开发3D ME-CNN和多 用于医学图像分类的模态3D ME-CNN。3)开发卷积和矢量ME自动编码器， 图像和组学数据。我们描述了一个对抗分类器的创新结合，以约束基础 学习固定效应的模型，贝叶斯随机效应子网络，以及应用随机效应的方法 到看不见的星团所有这些解决方案都将作为开源软件发布，以改进现有的DL模型 最终支持用于研究和治疗人类疾病的精确生物医学。