Machine Learning-based Pipeline Research Articles

Advancing forest carbon stocks’ mapping using a hierarchical approach with machine learning and satellite imagery

Remote sensing of forests is a powerful tool for monitoring the biodiversity of ecosystems, maintaining general planning, and accounting for resources. Various sensors bring together heterogeneous data, and advanced machine learning methods enable their automatic handling in wide territories. Key forest properties usually under consideration in environmental studies include dominant species, tree age, height, basal area and timber stock. Being proxies of stand productivity, they can be utilized for forest carbon stock estimation to analyze forests’ status and proper climate change mitigation measures on a global scale. In this study, we aim to develop an effective machine learning-based pipeline for automatic carbon stock estimation using solely freely available and regularly updated satellite observations. We employed multispectral Sentinel-2 remote sensing data to predict forest structure characteristics and produce their detailed spatial maps. Using the Extreme Gradient Boosting (XGBoost) algorithm in classification and regression settings and management-level inventory data as reference measurements, we achieved quality of predictions of species equal to 0.75 according to the F1-score, and for stand age, height, and basal area, we achieved an accuracy of 0.75, 0.58 and 0.56, respectively, according to the R2. We focused on the growing stock volume as the main proxy to estimate forest carbon stocks on the example of the stem pool. We explored two approaches: a direct approach and a hierarchical approach. The direct approach leverages the remote sensing data to create the target maps, and the hierarchical approach calculates the target forest properties using predicted inventory characteristics and conversion equations. We estimated stem carbon stock based on the same approach: from Earth observation imagery directly and using biomass and conversion factors developed for the northern regions. Thus, our study proposes an end-to-end solution for carbon stock estimations based on the complexation of inventory data at the forest stand level, Earth observation imagery, machine learning predictions and conversion equations for the region. The presented approach enables more robust and accurate large-scale assessments using limited annotated datasets.

A machine learning-based pipeline for multi-organ/tissue patient-specific radiation dosimetry in CT.

To develop a machine learning-based pipeline for multi-organ/tissue personalized radiation dosimetry in CT. For the study, 95 chest CT scans and 85 abdominal CT scans were collected retrospectively. For each CT scan, a personalized Monte Carlo (MC) simulation was carried out. The produced 3D dose distributions and the respective CT examinations were utilized for the development of organ/tissue-specific dose prediction deep neural networks (DNNs). A pipeline that integrates a robust open-source organ segmentation tool with the dose prediction DNNs was developed for the automatic estimation of radiation doses for 30 organs/tissues including sub-volumes of the heart and lungs. The accuracy and time efficiency of the presented methodology was assessed. Statistical analysis (t-tests) was conducted to determine if the differences between the ground truth organ/tissue radiation dose estimates and the respective dose predictions were significant. The lowest median percentage differences between MC-derived organ/tissue doses and DNN dose predictions were observed for the lung vessels (4.3%), small bowel (4.7%), pulmonary artery (4.7%), and colon (5.2%), while the highest differences were observed for the right lung's upper lobe (13.3%), spleen (13.1%), pancreas (12.1%), and stomach (11.6%). Statistical analysis showed that the differences were not significant (p-value > 0.18). Furthermore, the mean inference time, regarding the validation cohort, of the developed methodology was 77.0 ± 11.0 s. The proposed workflow enables fast and accurate organ/tissue radiation dose estimations. The developed algorithms and dose prediction DNNs are publicly available ( https://github.com/eltzanis/multi-structure-CT-dosimetry ). The accuracy and time efficiency of the developed pipeline compose a useful tool for personalized dosimetry in CT. By adopting the proposed workflow, institutions can utilize an automated pipeline for patient-specific dosimetry in CT. Personalized dosimetry is ideal, but is time-consuming. The proposed pipeline composes a tool for facilitating patient-specific CT dosimetry in routine clinical practice. The developed workflow integrates a robust open-source segmentation tool with organ/tissue-specific dose prediction neural networks.

Aqueous humor proteomics analyzed by bioinformatics and machine learning in PDR cases versus controls

BackgroundTo comprehend the complexities of pathophysiological mechanisms and molecular events that contribute to proliferative diabetic retinopathy (PDR) and evaluate the diagnostic value of aqueous humor (AH) in monitoring the onset of PDR.MethodsA cohort containing 16 PDR and 10 cataract patients and another validation cohort containing 8 PDR and 4 cataract patients were studied. AH was collected and subjected to proteomics analyses. Bioinformatics analysis and a machine learning-based pipeline called inference of biomolecular combinations with minimal bias were used to explore the functional relevance, hub proteins, and biomarkers.ResultsDeep profiling of AH proteomes revealed several insights. First, the combination of SIAE, SEMA7A, GNS, and IGKV3D-15 and the combination of ATP6AP1, SPARCL1, and SERPINA7 could serve as surrogate protein biomarkers for monitoring PDR progression. Second, ALB, FN1, ACTB, SERPINA1, C3, and VTN acted as hub proteins in the profiling of AH proteomes. SERPINA1 was the protein with the highest correlation coefficient not only for BCVA but also for the duration of diabetes. Third, “Complement and coagulation cascades” was an important pathway for PDR development.ConclusionsAH proteomics provides stable and accurate biomarkers for early warning and diagnosis of PDR. This study provides a deep understanding of the molecular mechanisms of PDR and a rich resource for optimizing PDR management.

Abstract 3520: A scalable single cell RNA-seq pipeline leveraging machine learning and high-quality references for cell-type prediction

Abstract Single cell RNA-seq (scRNA-seq) technology transformed our understanding of biology at the cell level. Inferring cell types provide insights into the relative abundance of, and genomic differences between, different cell types. Current methods leverage known cell type markers and genomic similarity measures to attribute cell types to groups of cells. We present a scalable machine learning-based pipeline that can leverage high quality reference annotation data to infer cell types quickly for multiple scRNA-seq experimental samples. This pipeline leverages two machine learning technologies: variational autoencoders (VAE) to create a low-dimensional representation of the reference transcriptome; and, supervised learning methods that use this representation to learn cell types from the reference data. This pipeline can quickly score new experimental data using GPU-enabled computing platforms and provide probabilities of a cell being each cell type present in the reference data. This novel pipeline is evaluated using 5 public reference data sets. We evaluate whether VAE-based representations benefit supervised learning, compared to PCA and t-SNE. We next evaluate different supervised learning methods for predictive performance. Finally, we compare the pipeline’s performance with commonly-used cell type prediction algorithms (Seurat, scANVI). We find that using a VAE is generally better than both PCA and t-SNE for feature generation to predict cell type. Using the VAE representation, there is no significant difference in accuracy between logistic regression and other supervised learning algorithms. Finally, using logistic regression as the learning machine, this pipeline is as accurate as Seurat and better than scANVI, while running about 2-5x faster than Seurat. Table 1 Balanced accuracy (%) from leave-one-subject-out cross-validation across 5 public data sets. Intervals are the range of the metric across the cross-validation folds. Algorithm PBMC (10X) HLCA Eraslan snRNA Blueprint Breast Blueprint Lung Our pipeline 92 (87-95) 92 (88-93) 93 (90-97) 89 (80-96) 97 (95-99) Seurat 96 (95-97) 90 (94-98) 94 (93-97) 96 (87-98) 98 (94-99) scANVI 91 (84-95) 72 (78-86) 92 (88-96) 94 (80-97) 88 (86-100) Citation Format: Abhijit Dasgupta, Grant Duclos, Ricardo Miragaia, Brychan Manry, Etai Jacob, Natasha Markuzon, Asaf Rotem. A scalable single cell RNA-seq pipeline leveraging machine learning and high-quality references for cell-type prediction [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 3520.

LoRa localisation using single mobile gateway

Effective use of GPS and mobile networks for localisation in rangeland areas is constrained by their high power consumption and high deployment costs. Long-range (LoRa), a low-power wide area network (LPWAN) technology, can be employed to mitigate these challenges. In contrast to prior research where the prevalent approaches entail multiple gateways. This work proposes a valuable methodology focused on a single mobile LoRa gateway for localisation. A particle filtering and machine learning-based pipeline is employed to map the distance between a target node and the gateway from the received signal strength indicator (RSSI). Particle filtering is used to reduce the impact of noise on the RSSI values. Then, several machine learning techniques, such as support vector machines, random forest, and k-nearest neighbour, are used on the RSSI values to estimate the distance. The estimated distance is then used for tracking using a centroid pseudo-trilateration method. The proposed method was tested in a real-world semi-line-of-sight setting, using three datasets generated by LoRaWAN-specified hardware components and a server. Two forms of experiments were performed: active searching and passive monitoring. We propose an iterative estimation process to address the dilution of precision caused by the initial positions of the gateway required for active searching applications. The results show that active searching typically requires 2 to 3 hops to reach a target node. The accuracy of passive monitoring depends on the proximity of the gateway, which varies from 20 m to 170 m. This proposed approach has the potential to open the way for localising, tracking, or monitoring target objects within sparsely populated rangeland areas, even when resources are severely constrained.

A Machine Learning-Based Pipeline for the Extraction of Insights from Customer Reviews

The efficiency of natural language processing has improved dramatically with the advent of machine learning models, particularly neural network-based solutions. However, some tasks are still challenging, especially when considering specific domains. This paper presents a model that can extract insights from customer reviews using machine learning methods integrated into a pipeline. For topic modeling, our composite model uses transformer-based neural networks designed for natural language processing, vector-embedding-based keyword extraction, and clustering. The elements of our model have been integrated and tailored to better meet the requirements of efficient information extraction and topic modeling of the extracted information for opinion mining. Our approach was validated and compared with other state-of-the-art methods using publicly available benchmark datasets. The results show that our system performs better than existing topic modeling and keyword extraction methods in this task.

Cecilia: a machine learning-based pipeline for measuring metal abundances of helium-rich polluted white dwarfs

ABSTRACT Over the past several decades, conventional spectral analysis techniques of polluted white dwarfs have become powerful tools to learn about the geology and chemistry of extrasolar bodies. Despite their proven capabilities and extensive legacy of scientific discoveries, these techniques are, however, still limited by their manual, time-intensive, and iterative nature. As a result, they are susceptible to human errors and are difficult to scale up to population-wide studies of metal pollution. This paper seeks to address this problem by presenting cecilia, the first machine learning (ML)-powered spectral modelling code designed to measure the metal abundances of intermediate-temperature (10 000 ≤ Teff ≤ 20 000 K), Helium-rich polluted white dwarfs. Trained with more than 22 000 randomly drawn atmosphere models and stellar parameters, our pipeline aims to overcome the limitations of classical methods by replacing the generation of synthetic spectra from computationally expensive codes and uniformly spaced model grids, with a fast, automated, and efficient neural-network-based interpolator. More specifically, cecilia combines state-of-the-art atmosphere models, powerful artificial intelligence tools, and robust statistical techniques to rapidly generate synthetic spectra of polluted white dwarfs in high-dimensional space, and enable accurate (≲0.1 dex) and simultaneous measurements of 14 stellar parameters – including 11 elemental abundances – from real spectroscopic observations. As massively multiplexed astronomical surveys begin scientific operations, cecilia’s performance has the potential to unlock large-scale studies of extrasolar geochemistry and propel the field of white dwarf science into the era of Big Data. In doing so, we aspire to uncover new statistical insights that were previously impractical with traditional white dwarf characterization techniques.

Machine learning-based morphological quantification of replicative senescence in human fibroblasts.

Although aging has been investigated extensively at the organismal and cellular level, the morphological changes that individual cells undergo along their replicative lifespan have not been precisely quantified. Here, we present the results of a readily accessible machine learning-based pipeline that uses standard fluorescence microscope and open access software to quantify the minute morphological changes that human fibroblasts undergo during their replicative lifespan in culture. Applying this pipeline in a widely used fibroblast cell line (IMR-90), we find that advanced replicative age robustly increases (+28-79%) cell surface area, perimeter, number and total length of pseudopodia, and nuclear surface area, while decreasing cell circularity, with phenotypic changes largely occurring as replicative senescence is reached. These senescence-related morphological changes are recapitulated, albeit to a variable extent, in primary dermal fibroblasts derived from human donors of different ancestry, age, and sex groups. By performing integrative analysis of single-cell morphology, our pipeline further classifies senescent-like cells and quantifies how their numbers increase with replicative senescence in IMR-90 cells and in dermal fibroblasts across all tested donors. These findings provide quantitative insights into replicative senescence, while demonstrating applicability of a readily accessible computational pipeline for high-throughput cell phenotyping in aging research.

A machine learning-based pipeline for stride speed estimation with a head-worn inertial sensor

A machine learning-based pipeline for stride speed estimation with a head-worn inertial sensor

EffectorO: Motif-Independent Prediction of Effectors in Oomycete Genomes Using Machine Learning and Lineage Specificity.

Oomycete plant pathogens cause a wide variety of diseases, including late blight of potato, sudden oak death, and downy mildews of plants. These pathogens are major contributors to loss in numerous food crops. Oomycetes secrete effector proteins to manipulate their hosts to the advantage of the pathogen. Plants have evolved to recognize effectors, resulting in an evolutionary cycle of defense and counter-defense in plant-microbe interactions. This selective pressure results in highly diverse effector sequences that can be difficult to computationally identify using only sequence similarity. We developed a novel effector prediction tool, EffectorO, that uses two complementary approaches to predict effectors in oomycete pathogen genomes: i) a machine learning-based pipeline that predicts effector probability based on the biochemical properties of the N-terminal amino-acid sequence of a protein and ii) a pipeline based on lineage specificity to find proteins that are unique to one species or genus, a sign of evolutionary divergence due to adaptation to the host. We tested EffectorO on Bremia lactucae, which causes lettuce downy mildew, and Phytophthora infestans, which causes late blight of potato and tomato, and predicted many novel effector candidates while recovering the majority of known effector candidates. EffectorO will be useful for discovering novel families of oomycete effectors without relying on sequence similarity to known effectors. [Formula: see text] Copyright © 2023 The Author(s). This is an open access article distributed under the CC BY-NC-ND 4.0 International license.

Abstract 6580: A machine learning-based pipeline and web server ImmuneMirror for neoantigen prediction

Abstract Background: Gastrointestinal tract (GIT) cancer is a group of cancers that affect the digestive system. It has become a leading cause of cancer mortality worldwide. However, the clinical outcome for advanced-stage patients with cancer metastasis after conventional treatment remains poor. Although cancer immunotherapy has attracted much attention for the treatment of these deadly cancers as a single agent or in combination with chemotherapy, the objective response rate to this treatment remains low and most of the responsive patients only have a partial response. Accumulating evidence suggests that multiple mechanisms are involved in the resistance to cancer immunotherapy. Methods: To facilitate the identification of the cancer patients who most likely respond to immunotherapy, we developed a stand-alone open-source ImmuneMirror pipeline with graphical analysis results for evaluating the selected features including tumor mutation burden, microsatellite instability (MSI) status, HLA typing, predicted neoantigen load and the top-ranked neoantigens with T cell immunogenicity as well as the expression of innate anti-PD1 resistance (IPRES) gene signature. We incorporated a machine-learning model for neoantigen prediction and prioritization in ImmuneMirror and established a web server. The model was trained and tested using the known neopeptides with T cell immunogenicity collected from 19 published studies, and combined neoantigen-relevant features: agretopicity, foreignness, hydrophobicity, binding stability, peptide processing, and transportation scores. Results: The accuracy of neoantigen prediction accuracy of ImmuneMirror model is 0.87. We utilized ImmuneMirror to analyze 805 normal-tumor paired samples in GIT cancers including esophageal squamous cell carcinoma (ESCC), colorectal cancer (CRC) and hepatocellular carcinoma (HCC). Elevated neoantigen load was correlated with good clinical outcomes in ESCC. Interestingly, we identified a subgroup of MSI-H CRC with relatively low neoantigen loads for MHC Class I and Class II, respectively, though the tumor mutation burdens of this subset of patients were comparable with other MSI-H CRC. Our result suggests this subset of patients may not respond well to immunotherapy. In addition, the neopeptide YMCNSSCMGV derived from the TP53 hotpot mutation G245V mutation restricted by HLA-A02 was identified as the actional target in the ESCC patient by ImmuneMirror and confirmed by the experimental validation. Conclusions: These results showed the reliability and effectiveness of ImmuneMirror for evaluating the relevant features and predicting putative neoantigens. Acknowledgments: We acknowledge the support from Health Medical Research Fund (Grant No. 07182016) from the Research Fund Secretariat of Health Bureau in Hong Kong. Citation Format: Wei Dai, Gulam Sarwar Chuwdhury, Yunshan Guo, Zhonghua Liu. A machine learning-based pipeline and web server ImmuneMirror for neoantigen prediction. [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2023; Part 1 (Regular and Invited Abstracts); 2023 Apr 14-19; Orlando, FL. Philadelphia (PA): AACR; Cancer Res 2023;83(7_Suppl):Abstract nr 6580.

Utilization of convolutional neural networks for H I source finding

Context. The future deployment of the Square Kilometer Array (SKA) will lead to a massive influx of astronomical data and the automatic detection and characterization of sources will therefore prove crucial in utilizing its full potential. Aims. We examine how existing astronomical knowledge and tools can be utilized in a machine learning-based pipeline to find 3D spectral line sources. Methods. We present a source-finding pipeline designed to detect 21-cm emission from galaxies that provides the second-best submission of SKA Science Data Challenge 2. The first pipeline step was galaxy segmentation, which consisted of a convolutional neural network (CNN) that took an H I cube as input and output a binary mask to separate galaxy and background voxels. The CNN was trained to output a target mask algorithmically constructed from the underlying source catalog of the simulation. For each source in the catalog, its listed properties were used to mask the voxels in its neighborhood that capture plausible signal distributions of the galaxy. To make the training more efficient, regions containing galaxies were oversampled compared to the background regions. In the subsequent source characterization step, the final source catalog was generated by the merging and dilation modules of the existing source-finding software SOFIA, and some complementary calculations, with the CNN-generated mask as input. To cope with the large size of H I cubes while also allowing for deployment on various computational resources, the pipeline was implemented with flexible and configurable memory usage. Results. We show that once the segmentation CNN has been trained, the performance can be fine-tuned by adjusting the parameters involved in producing the catalog from the mask. Using different sets of parameter values offers a trade-off between completeness and reliability.

Knowledge mining of unstructured information: application to cyber domain

Information on cyber-related crimes, incidents, and conflicts is abundantly available in numerous open online sources. However, processing large volumes and streams of data is a challenging task for the analysts and experts, and entails the need for newer methods and techniques. In this article we present and implement a novel knowledge graph and knowledge mining framework for extracting the relevant information from free-form text about incidents in the cyber domain. The computational framework includes a machine learning-based pipeline for generating graphs of organizations, countries, industries, products and attackers with a non-technical cyber-ontology. The extracted knowledge graph is utilized to estimate the incidence of cyberattacks within a given graph configuration. We use publicly available collections of real cyber-incident reports to test the efficacy of our methods. The knowledge extraction is found to be sufficiently accurate, and the graph-based threat estimation demonstrates a level of correlation with the actual records of attacks. In practical use, an analyst utilizing the presented framework can infer additional information from the current cyber-landscape in terms of the risk to various entities and its propagation between industries and countries.

An Automated Pipeline for Differential Cell Counts on Whole-Slide Bone Marrow Aspirate Smears

The pathologic diagnosis of bone marrow disorders relies in part on the microscopic analysis of bone marrow aspirate (BMA) smears and the manual counting of marrow nucleated cells to obtain a differential cell count (DCC). This manual process has significant limitations, including the analysis of only a small subset of optimal slide areas and nucleated cells, as well as interobserver variability due to differences in cell selection and classification. To address these shortcomings, we developed an automated machine learning-based pipeline for obtaining 11-component DCCs on whole-slide BMAs. This pipeline uses a sequential process of identifying optimal BMA regions with high proportions of marrow nucleated cells, detecting individual cells within these optimal areas, and classifying these cells into 1 of 11 DCC components. Convolutional neural network models were trained on 396,048 BMA region, 28,914 cell boundary, and 1,510,976 cell class images from manual annotations. The resulting automated pipeline produced 11-component DCCs that demonstrated a high statistical and diagnostic concordance with manual DCCs among a heterogeneous group of testing BMA slides with varying pathologies and cellularities. Additionally, we demonstrated that an automated analysis can reduce the intraslide variance in DCCs by analyzing the whole slide and marrow nucleated cells within all optimal regions. Finally, the pipeline outputs of region classification, cell detection, and cell classification can be visualized using whole-slide image analysis software. This study demonstrates the feasibility of a fully automated pipeline for generating DCCs on scanned whole-slide BMA images, with the potential for improving the current standard of practice for utilizing BMA smears in the laboratory analysis of hematologic disorders.

Machine learning for the automatic assessment of aortic rotational flow and wall shear stress from 4D flow cardiac magnetic resonance imaging.

Three-dimensional (3D) time-resolved phase-contrast cardiac magnetic resonance (4D flow CMR) allows for unparalleled quantification of blood velocity. Despite established potential in aortic diseases, the analysis is time-consuming and requires expert knowledge, hindering clinical application. The present research aimed to develop and test a fully automatic machine learning-based pipeline for aortic 4D flow CMR analysis. Four hundred and four subjects were prospectively included. Ground-truth to train the algorithms was generated by experts. The cohort was divided into training (323 patients) and testing (81) sets and used to train and test a 3D nnU-Net for segmentation and a Deep Q-Network algorithm for landmark detection. In-plane (IRF) and through-plane (SFRR) rotational flow descriptors and axial and circumferential wall shear stress (WSS) were computed at ten planes covering the ascending aorta and arch. Automatic aortic segmentation resulted in a median Dice score (DS) of 0.949 and average symmetric surface distance of 0.839 (0.632-1.071) mm, comparable with the state of the art. Aortic landmarks were located with a precision comparable with experts in the sinotubular junction and first and third supra-aortic vessels (p = 0.513, 0.592 and 0.905, respectively) but with lower precision in the pulmonary bifurcation (p = 0.028), resulting in precise localisation of analysis planes. Automatic flow assessment showed excellent (ICC > 0.9) agreement with manual quantification of SFRR and good-to-excellent agreement (ICC > 0.75) in the measurement of IRF and axial and circumferential WSS. Fully automatic analysis of complex aortic flow dynamics from 4D flow CMR is feasible. Its implementation could foster the clinical use of 4D flow CMR. • 4D flow CMR allows for unparalleled aortic blood flow analysis but requires aortic segmentation and anatomical landmark identification, which are time-consuming, limiting 4D flow CMR widespread use. • A fully automatic machine learning pipeline for aortic 4D flow CMR analysis was trained with data of 323 patients and tested in 81 patients, ensuring a balanced distribution of aneurysm aetiologies. • Automatic assessment of complex flow characteristics such as rotational flow and wall shear stress showed good-to-excellent agreement with manual quantification.

MAPPER: An Open-Source, High-Dimensional Image Analysis Pipeline Unmasks Differential Regulation of Drosophila Wing Features.

Phenomics requires quantification of large volumes of image data, necessitating high throughput image processing approaches. Existing image processing pipelines for Drosophila wings, a powerful genetic model for studying the underlying genetics for a broad range of cellular and developmental processes, are limited in speed, precision, and functional versatility. To expand on the utility of the wing as a phenotypic screening system, we developed MAPPER, an automated machine learning-based pipeline that quantifies high-dimensional phenotypic signatures, with each dimension quantifying a unique morphological feature of the Drosophila wing. MAPPER magnifies the power of Drosophila phenomics by rapidly quantifying subtle phenotypic differences in sample populations. We benchmarked MAPPER’s accuracy and precision in replicating manual measurements to demonstrate its widespread utility. The morphological features extracted using MAPPER reveal variable sexual dimorphism across Drosophila species and unique underlying sex-specific differences in morphogen signaling in male and female wings. Moreover, the length of the proximal-distal axis across the species and sexes shows a conserved scaling relationship with respect to the wing size. In sum, MAPPER is an open-source tool for rapid, high-dimensional analysis of large imaging datasets. These high-content phenomic capabilities enable rigorous and systematic identification of genotype-to-phenotype relationships in a broad range of screening and drug testing applications and amplify the potential power of multimodal genomic approaches.

Machine Learning analysis of high-grade serous ovarian cancer proteomic dataset reveals novel candidate biomarkers

Ovarian cancer is one of the most common gynecological malignancies, ranking third after cervical and uterine cancer. High-grade serous ovarian cancer (HGSOC) is one of the most aggressive subtype, and the late onset of its symptoms leads in most cases to an unfavourable prognosis. Current predictive algorithms used to estimate the risk of having Ovarian Cancer fail to provide sufficient sensitivity and specificity to be used widely in clinical practice. The use of additional biomarkers or parameters such as age or menopausal status to overcome these issues showed only weak improvements. It is necessary to identify novel molecular signatures and the development of new predictive algorithms able to support the diagnosis of HGSOC, and at the same time, deepen the understanding of this elusive disease, with the final goal of improving patient survival. Here, we apply a Machine Learning-based pipeline to an open-source HGSOC Proteomic dataset to develop a decision support system (DSS) that displayed high discerning ability on a dataset of HGSOC biopsies. The proposed DSS consists of a double-step feature selection and a decision tree, with the resulting output consisting of a combination of three highly discriminating proteins: TOP1, PDIA4, and OGN, that could be of interest for further clinical and experimental validation. Furthermore, we took advantage of the ranked list of proteins generated during the feature selection steps to perform a pathway analysis to provide a snapshot of the main deregulated pathways of HGSOC. The datasets used for this study are available in the Clinical Proteomic Tumor Analysis Consortium (CPTAC) data portal (https://cptac-data-portal.georgetown.edu/).

Using a Machine Learning Approach to Monitor COVID-19 Vaccine Adverse Events (VAE) from Twitter Data.

Social media can be used to monitor the adverse effects of vaccines. The goal of this project is to develop a machine learning and natural language processing approach to identify COVID-19 vaccine adverse events (VAE) from Twitter data. Based on COVID-19 vaccine-related tweets (1 December 2020–1 August 2021), we built a machine learning-based pipeline to identify tweets containing personal experiences with COVID-19 vaccinations and to extract and normalize VAE-related entities, including dose(s); vaccine types (Pfizer, Moderna, and Johnson & Johnson); and symptom(s) from tweets. We further analyzed the extracted VAE data based on the location, time, and frequency. We found that the four most populous states (California, Texas, Florida, and New York) in the US witnessed the most VAE discussions on Twitter. The frequency of Twitter discussions of VAE coincided with the progress of the COVID-19 vaccinations. Sore to touch, fatigue, and headache are the three most common adverse effects of all three COVID-19 vaccines in the US. Our findings demonstrate the feasibility of using social media data to monitor VAEs. To the best of our knowledge, this is the first study to identify COVID-19 vaccine adverse event signals from social media. It can be an excellent supplement to the existing vaccine pharmacovigilance systems.

Flow Regime Estimation in Two-Phase Adiabatic Flows: Can Correlations and Maps Be Replaced with a Reproducible and Explainable Machine Learning-Based Pipeline?

Flow Regime Estimation in Two-Phase Adiabatic Flows: Can Correlations and Maps Be Replaced with a Reproducible and Explainable Machine Learning-Based Pipeline?

PassFlow: a multimodal workflow for predicting deep brain stimulation outcomes.

Deep Brain Stimulation (DBS) is a proven therapy for Parkinson's Disease (PD), frequently resulting in an enhancement of motor function. Nonetheless, several undesirable side effects can occur after DBS, which can worsen the quality of life of the patient. Thus, the clinical team has to carefully select patients on whom to perform DBS. Over the past decade, there have been some attempts to relate pre-operative data and DBS clinical outcomes, with most focused on the motor symptomatology. In this paper, we propose a machine learning-based method able to predict a large number of DBS clinical outcomes for PD. We propose a multimodal pipeline, referred to as PassFlow, which predicts 84 clinical post-operative clinical scores. PassFlow is composed of an artificial neural network to compress clinical information, an image processing method from the state-of-the-art to extract morphological biomarkers our of T1 imaging, and an SVM to perform the regressions. We validated PassFlow on 196 PD patients who undergone a DBS. PassFlow showed correlation coefficients as high as 0.71 and were able to significantly predict 63 out of the 84 scores, outperforming a comparative linear method. The number of metrics that are predicted with this pre-operative information was also found to be correlated with the number of patients with this information available, indicating that the PassFlow method is still actively learning. We presented a novel, machine learning-based pipeline to predict a variety of post-operative clinical outcomes of DBS for PD patients. PassFlow took into account various bio-markers, arising from different data modalities, showing high correlation coefficients for some scores from pre-operative data only. It indicates that many clinical outcomes of DBS can be predicted agnostic to the specific simulation parameters, as PassFlow has been validated without such stimulation-related information.