Multi-omics disease module detection with an explainable Greedy Decision Forest

Abstract

Machine learning methods can detect complex relationships between variables, but usually do not exploit domain knowledge. This is a limitation because in many scientific disciplines, such as systems biology, domain knowledge is available in the form of graphs or networks, and its use can improve model performance. We need network-based algorithms that are versatile and applicable in many research areas. In this work, we demonstrate subnetwork detection based on multi-modal node features using a novel Greedy Decision Forest (GDF) with inherent interpretability. The latter will be a crucial factor to retain experts and gain their trust in such algorithms. To demonstrate a concrete application example, we focus on bioinformatics, systems biology and particularly biomedicine, but the presented methodology is applicable in many other domains as well. Systems biology is a good example of a field in which statistical data-driven machine learning enables the analysis of large amounts of multi-modal biomedical data. This is important to reach the future goal of precision medicine, where the complexity of patients is modeled on a system level to best tailor medical decisions, health practices and therapies to the individual patient. Our proposed explainable approach can help to uncover disease-causing network modules from multi-omics data to better understand complex diseases such as cancer.

FHBF: Federated hybrid boosted forests with dropout rates for supervised learning tasks across highly imbalanced clinical datasets

“…In addition, the hybrid loss function can be used as an alternative to the existing loss functions that are used for binary classification tasks and are reported in the xgboost documentation, 26 such as “logistic,” “logitraw,” and “hinge.” 26 The proposed hybrid loss function can also be used in interpretable ML algorithms toward the detection of complex relationships between variables, such as the greedy decision forest. 34 The selection of the hybrid loss function is highly recommended to avoid overfitting effects that are introduced by the arbitrary definition of the dropouts. The dominance of the FHBF algorithm was demonstrated in two experimental phases involving the development of computationally complex AI models for lymphoma classification across complex data with increased class imbalance.…”

A machine learning and deep learning-based integrated multi-omics technique for leukemia prediction

“…Different from traditional research for detecting and predicting leukemia disease, our proposed research exploits both ML and DL. The proposed research includes the latest techniques to predict leukemia disease using various ML and DL classifiers, including RF, DT, LR, GB, NB, FNN, and RNN [ [70] , [71] , [72] ]. After selecting the most relevant features, we obtained 10,095 samples.…”

Interpretability and Transparency of Machine Learning in File Fragment Analysis with Explainable Artificial Intelligence

“…Gohel et al [11] focus on XAI techniques in multimedia and outline future research directions, while Saeed et al [12] present a meta-survey that identifies challenges and future research directions in XAI. Adding on to that, Colley et al [13], Pfeifer et al [14], and Liao et al [15] explore specific applications and methodological advancements in XAI. Colley et al [13] propose a conceptual framework for tangible XAI, incorporating graphical user interfaces and physical artifacts.…”