Collaborative Data Science Research Articles

A global effort to benchmark predictive models and reveal mechanistic diversity in long-term stroke outcomes.

Stroke remains a leading cause of mortality and long-term disability worldwide, with variable recovery trajectories posing substantial challenges in anticipating post-event care and rehabilitation planning. In response, we established the NeuralCup consortium to address these challenges by benchmarking predictive models of stroke outcome through a collaborative, data-driven approach. This study presents the findings of 15 participating teams worldwide who used a comprehensive dataset including clinical and imaging data, to identify and compare predictors of motor, cognitive, and emotional outcomes one-year post-stroke. Analyses integrated traditional and novel approaches, including machine learning algorithms to discover 'optimal recipes' for predicting each domain. The differences in these 'optimal recipes' reflect distinct brain mechanisms in response to different tasks. Key predictors across all domains included lesion characteristics, T1-weighted MRI sequences, and demographic factors. Additionally, the integration of FLAIR imaging and white matter tract analysis significantly improved the prediction of cognitive and motor outcomes, respectively. These findings support a multifaceted approach to stroke outcome prediction, underscoring the potential of collaborative data science to develop personalized care strategies that enhance recovery and quality of life for stroke survivors.

Increasing the collaboration of data science stakeholders with a knowledge management system

PurposeThis paper presents the design and implementation of collaborative data science framework (CoDS), a knowledge management system for consolidating data science activities in an enterprise.Design/methodology/approachThe development of the CoDS framework is grounded on the design science research methodology for information systems research. In our case study, we first designed the initial framework for CoDS based on a systematic literature review. Then, we collected the expert opinions of eight data scientists to validate the need for generic content for such a knowledge management system. In the second iteration, a portfolio prototype is developed by the same data scientists as a part of our technical action research. Finally, a survey is conducted with 57 data analyst candidates in the last iteration.FindingsUsing the CoDS portfolio strengthened the communication among data scientists and stakeholders to improve development and scaling activities. It eased the reuse or modification of existing analytical solutions in other company processes.Practical implicationsThe CoDS presents a platform on which business details, data-related knowledge, modeling procedures and deployment steps are shared for (1) mediating and scaling ongoing projects, (2) enriching knowledge transfer among stakeholders, (3) facilitating ideation of new products and (4) supporting the onboarding of new employees and developers.Originality/valueThis study proposes a novel structure and a roadmap for creating a data science knowledge management system for the collaboration of all stakeholders in an enterprise.

Collective Intelligence and Collaborative Data Science

Collective Intelligence and Collaborative Data Science

Application of the Delphi method to the development of common data elements for social drivers of health: A systematic scoping review.

Collaborative data science requires standardized, harmonized, interoperable, and ethically sourced data. Developing an agreed-upon set of elements requires capturing different perspectives on the importance and feasibility of the data elements through a consensus development approach. This study reports on the systematic scoping review of literature that examined the inclusion of diverse stakeholder groups and sources of social drivers of health variables in consensus-based common data element (CDE) sets. This systematic scoping review included sources from PubMed, Embase, CINAHL, WoS MEDLINE, and PsycINFO databases. Extracted data included the stakeholder groups engaged in the Delphi process, sources of CDE sets, and inclusion of social drivers data across 11 individual and 6 social domains. Of the 384 studies matching the search string, 22 were included in the final review. All studies involved experts with healthcare expertise directly relevant to the developed CDE set, and only six (27%) studies engaged health consumers. Literature reviews and expert input were the most frequent sources of CDE sets. Seven studies (32%) did not report the inclusion of any demographic variables in the CDE sets, and each demographic SDoH domain was included in at least one study with age and sex assigned at birth included in all studies, and social driver domains included only in four studies (18%). The Delphi technique engages diverse expert groups around the development of SDoH data elements. Future studies can benefit by involving health consumers as experts.

102 CUBE: A Collaborative Undergraduate Biostatistics Experience to Bring Diversity and Awareness to the Field of Collaborative Biostatistics

OBJECTIVES/GOALS: Despite a steady rise of graduate degrees in biostatistics earned in the US, the percent from minorities remains low. This poster will describe the Collaborative Undergraduate Biostatistics Experience (CUBE), an 8-week program aimed to diversify and bring awareness to the field of collaborative biostatistics, from recruitment through evaluation. METHODS/STUDY POPULATION: The CUBE program is funded jointly by the NIH’s NIDA/NIAAA (award number: 1R25DA058482-01) and is designed to give underrepresented minority (URM) undergraduate students in STEM the opportunity to engage in a collaborative biostatistics and health data science experience, along with related professional development activities. The program is built on four pillars: 1) training in introductory biostatistics, 2) training in R programming, 3) professional development, and 4) a collaborative research project addressing research questions in various disciplines. The CUBE program was delivered in the summer of 2022 as a pilot to four URM students at Virginia Tech (VT) and the University of Virginia (UVA), with two at each site. In summer 2023, the program was offered to 5 students (3 VT, 2 UVA). RESULTS/ANTICIPATED RESULTS: This poster will provide strategies learned over two summers with respect to recruitment and enrollment, along with details on the program content, timeline, and short- and long-term program evaluation metrics (both quantitative and qualitative). The CUBE program was well-received by students participating in summers 2022 and 2023, where improved attitudes towards statistics were demonstrated, and 7 of the total 9 participants (78%) over the past two summers expressed interest in pursuing a graduate degree in biostatistics or a career in quantitative research. Of these 7 students, 1 is currently enrolled in a biostatistics graduate program in the United States. DISCUSSION/SIGNIFICANCE: Results can be used to offer recommendations to leaders in the field on how to establish similar programs seeking to provide a pipeline for equity and diversity in the practice of collaborative biostatistics and health data science.

Traceable Use of Emerging Technologies in Smart Systems

This volume presents a selection of invited papers from the 3rd Workshop on Collaborative Technologies and Data Science in Smart City Applications (CODASSCA 2022): From Data to Information and Knowledge, held in Yerevan, Armenia, August, 23-25, and further articles from a free call for papers JUCS-CODASSCA-2023 published by Easychair. The workshop continues the cooperation between the University of Duisburg‐Essen (UDE) and the American University of Armenia (AUA) funded by the German Academic Exchange Service (DAAD) and the German Research Foundation (DFG). The workshop took place together with a one-week summer school on the topic Enhancements of Deep Learning for Intelligent Applications and the Connected Society. 

Where Are We Going with Statistical Computing? From Mathematical Statistics to Collaborative Data Science

The field of statistical computing is rapidly developing and evolving. Shifting away from the formerly siloed landscape of mathematics, statistics, and computer science, recent advancements in statistical computing are largely characterized by a fusing of these worlds; namely, programming, software development, and applied statistics are merging in new and exciting ways. There are numerous drivers behind this advancement, including open movement (encompassing development, science, and access), the advent of data science as a field, and collaborative problem-solving, as well as practice-altering advances in subfields such as artificial intelligence, machine learning, and Bayesian estimation. In this paper, we trace this shift in how modern statistical computing is performed, and that which has recently emerged from it. This discussion points to a future of boundless potential for the field.

A comparison of distributed machine learning methods for the support of "many labs" collaborations in computational modeling of decision making.

Deep learning models are powerful tools for representing the complex learning processes and decision-making strategies used by humans. Such neural network models make fewer assumptions about the underlying mechanisms thus providing experimental flexibility in terms of applicability. However, this comes at the cost of involving a larger number of parameters requiring significantly more data for effective learning. This presents practical challenges given that most cognitive experiments involve relatively small numbers of subjects. Laboratory collaborations are a natural way to increase overall dataset size. However, data sharing barriers between laboratories as necessitated by data protection regulations encourage the search for alternative methods to enable collaborative data science. Distributed learning, especially federated learning (FL), which supports the preservation of data privacy, is a promising method for addressing this issue. To verify the reliability and feasibility of applying FL to train neural networks models used in the characterization of decision making, we conducted experiments on a real-world, many-labs data pool including experiment data-sets from ten independent studies. The performance of single models trained on single laboratory data-sets was poor. This unsurprising finding supports the need for laboratory collaboration to train more reliable models. To that end we evaluated four collaborative approaches. The first approach represents conventional centralized learning (CL-based) and is the optimal approach but requires complete sharing of data which we wish to avoid. The results however establish a benchmark for the other three approaches, federated learning (FL-based), incremental learning (IL-based), and cyclic incremental learning (CIL-based). We evaluate these approaches in terms of prediction accuracy and capacity to characterize human decision-making strategies. The FL-based model achieves performance most comparable to that of the CL-based model. This indicates that FL has value in scaling data science methods to data collected in computational modeling contexts when data sharing is not convenient, practical or permissible.

Eliciting Best Practices for Collaboration with Computational Notebooks

Despite the widespread adoption of computational notebooks, little is known about best practices for their usage in collaborative contexts. In this paper, we fill this gap by eliciting a catalog of best practices for collaborative data science with computational notebooks. With this aim, we first look for best practices through a multivocal literature review. Then, we conduct interviews with professional data scientists to assess their awareness of these best practices. Finally, we assess the adoption of best practices through the analysis of 1,380 Jupyter notebooks retrieved from the Kaggle platform. Findings reveal that experts are mostly aware of the best practices and tend to adopt them in their daily work. Nonetheless, they do not consistently follow all the recommendations as, depending on specific contexts, some are deemed unfeasible or counterproductive due to the lack of proper tool support. As such, we envision the design of notebook solutions that allow data scientists not to have to prioritize exploration and rapid prototyping over writing code of quality.

J.UCS special issue on Challenges for Smart Environments – Human-Centered Computing, Data Science, and Ambient Intelligence. Smart Human-Centered Computing (volume 2)

Based on a successful funded collaboration between the American University of Armenia, the University of Duisburg-Essen and the University of Chile, in previous years a network was built, and in September 2020 a group of researchers gathered (although virtually) for the 2nd CODASSCA workshop on “Collaborative Technologies and Data Science in Smart City Applications”. This event has attracted 25 paper submissions which deal with the problems and challenges mentioned above. The studies are in specialized areas and disclose novel solutions and approaches based on existing theories suitably applied. The authors of the best papers published in the conference proceedings on Collaborative Technologies and Data Science in Artificial Intelligence Applications by Logos edition Berlin were invited to submit significantly extended and improved versions of their contributions to be considered for a journal special issue of J.UCS. There was also a J.UCS open call so that any author could submit papers on the highlighted subject. For this volume, we selected those devoted mainly to human-computer interaction problematics, which were rigorously reviewed in three rounds and 6 papers nominated to be published.

Editorial introduction to J.UCS special issue Challenges for Smart Environments – Human-Centered Computing, Data Science, and Ambient Intelligence I

Editorial introduction to J.UCS special issue Challenges for Smart Environments – Human-Centered Computing, Data Science, and Ambient Intelligence I

Enabling Collaborative Data Science Development with the Ballet Framework

While the open-source software development model has led to successful large-scale collaborations in building software systems, data science projects are frequently developed by individuals or small teams. We describe challenges to scaling data science collaborations and present a conceptual framework and ML programming model to address them. We instantiate these ideas in Ballet, the first lightweight framework for collaborative, open-source data science through a focus on feature engineering, and an accompanying cloud-based development environment. Using our framework, collaborators incrementally propose feature definitions to a repository which are each subjected to software and ML performance validation and can be automatically merged into an executable feature engineering pipeline. We leverage Ballet to conduct a case study analysis of an income prediction problem with 27 collaborators, and discuss implications for future designers of collaborative projects.

Collaborative Technologies and Data Science in Smart City Applications

Collaborative Technologies and Data Science in Smart City Applications

An overview of two open interactive computing environments useful for data science education.

ObjectiveTo discuss and illustrate the utility of two open collaborative data science platforms, and how they would benefit data science and informatics education.Methods and MaterialsThe features of two online data science platforms are outlined. Both are useful for new data projects and both are integrated with common programming languages used for data analysis. One platform focuses more on data exploration and the other focuses on containerizing, visualization, and sharing code repositories.ResultsBoth data science platforms are open, free, and allow for collaboration. Both are capable of visual, descriptive, and predictive analyticsDiscussionData science education benefits by having affordable open and collaborative platforms to conduct a variety of data analyses.ConclusionOpen collaborative data science platforms are particularly useful for teaching data science skills to clinical and nonclinical informatics students. Commercial data science platforms exist but are cost-prohibitive and generally limited to specific programming languages.

Dataflow representation of data analyses: Toward a platform for collaborative data science

Data science plays an increasingly important role in solving today's scientific and social challenges. To promote progress toward a cure for multiple sclerosis, the Accelerated Cure Project has created an open repository of biological and survey data on patients with multiple sclerosis. Similar large-scale repositories are being created in other domains. As the open, data-driven model of science proliferates, the research community faces a growing need for a cloud platform for collaborative data science. Such a platform should facilitate collaboration between domain experts and data scientists and possess artificial intelligence capabilities for organizing, recommending, and manipulating data analyses. In this paper, we present some foundational technologies motivated by this vision. Our system automatically extracts a high-level dataflow graph from a data analysis. This graph describes how data flows through an analysis pipeline, including which statistical methods are used and how they fit together. The system requires no special annotations from the data analyst and consumes analyses written in Python using standard tools, such as Scikit-learn and Statsmodels. In this paper, we explain how our system works and how it fits into our larger vision for a collaborative data science platform.

Security and privacy for cloud-based data management in the health network service chain: a microservice approach

New and useful tools that facilitate the harmonization and interconnection of health data have become a requirement for monitoring and preventing illness and also for sharing medical knowledge. Nowadays, cloud-based solutions can support collaborative data science platforms and deliver all types of processing operations through the network service chain. In this article we deal with healthcare-related data management and exchange, and we propose security and privacy requirements together with a novel microservices approach. We investigate how cloud computing can be adopted within healthcare systems. The interoperability of existing technologies will improve the quality of life and the efficiency of healthcare systems by making them more personalized and centered on patients, together with reducing operational costs and medical errors. To have a socially acceptable health network service chain, the security and privacy issues need to be analyzed and addressed. We explore the the security and privacy requirements and implications, and also discuss existing methods, and in the end propose an architecture of a secure manager for cloud-based healthcare-related data management and exchange.

Data-driven Performance Evaluation of Ventilated Photovoltaic Double-skin Facades in the Built Environment

In this paper we present a collaborative data science project bringing together experts from the fields of physical and information sciences to tackle to challenges and opportunities of analysing large and rich datasets obtained from the monitoring of building integrated PV systems operating in the built environment. We present data mining analysis techniques to classify data according to environmental conditions and system performance, to distinguish between nominal and anomalous behaviour, and to identify instrumentation faults. These methods were implemented using data from the RESSOURCES project to construct evaluate the performance of a PV envelope, and to validate a simplified physical model to predict thermal and aerodynamic behaviour.

Principles of Dataset Versioning: Exploring the Recreation/Storage Tradeoff.

The relative ease of collaborative data science and analysis has led to a proliferation of many thousands or millions of versions of the same datasets in many scientific and commercial domains, acquired or constructed at various stages of data analysis across many users, and often over long periods of time. Managing, storing, and recreating these dataset versions is a non-trivial task. The fundamental challenge here is the storage-recreation trade-off: the more storage we use, the faster it is to recreate or retrieve versions, while the less storage we use, the slower it is to recreate or retrieve versions. Despite the fundamental nature of this problem, there has been a surprisingly little amount of work on it. In this paper, we study this trade-off in a principled manner: we formulate six problems under various settings, trading off these quantities in various ways, demonstrate that most of the problems are intractable, and propose a suite of inexpensive heuristics drawing from techniques in delay-constrained scheduling, and spanning tree literature, to solve these problems. We have built a prototype version management system, that aims to serve as a foundation to our DataHub system for facilitating collaborative data science. We demonstrate, via extensive experiments, that our proposed heuristics provide efficient solutions in practical dataset versioning scenarios.