Data Science Experts Research Articles

Critical Data for Critical Care: A Primer on Leveraging Electronic Health Record Data for Research From Society of Critical Care Medicine's Panel on Data Sharing and Harmonization.

A growing body of critical care research draws on real-world data from electronic health records (EHRs). The bedside clinician has myriad data sources to aid in clinical decision-making, but the lack of data sharing and harmonization standards leaves much of this data out of reach for multi-institution critical care research. The Society of Critical Care Medicine (SCCM) Discovery Data Science Campaign convened a panel of critical care and data science experts to explore and document unique advantages and opportunities for leveraging EHR data in critical care research. This article reviews and illustrates six organizing topics (data domains and common data elements; data harmonization; data quality; data interoperability and digital infrastructure; data access, sharing, and governance; and ethics and equity) as a data science primer for critical care researchers, laying a foundation for future publications from the SCCM Discovery Data Harmonization and Sharing Guiding Principles Panel.

Charting a way forward for the use of data science in competition enforcement and platform regulation

ABSTRACT Competition authorities and other regulators already rely on data science to monitor markets and check compliance with applicable rules. The Digital Markets Act and the Digital Services Act have made the use of data science by regulators even more relevant, given the scale and complexity of the oversight required. The existing use of data science shows that regulators are able to adapt their organizations and processes to realize the potential of using technology tools in their activities. At the same time, the use of data science poses challenges in terms of data reliability and individual privacy. Beyond involvement in specific investigations, data science also has the potential to reform the work of regulators by moving towards a more proactive form of enforcement. The sharing of data science expertise and tools between regulators deserves to be further facilitated to increase collaboration and share resources in monitoring the platform economy.

Accelerating Museum AI Research and Application at the UK Natural History Museum: The NHM AI Lab Programme

The United Kingdom's Natural History Museum (NHM) AI Lab Programme represents a pioneering initiative aimed at harnessing the power of artificial intelligence (AI) to bridge the gap between the museum's extensive collection and cutting-edge AI technologies. Despite its immense potential, the application of AI in museum research remains nascent (e.g., He et al. 2024), with some individual research groups pursuing independent projects without cohesive collaboration with AI experts who know or have experience in similar endeavours. Moreover, differing standards in utilising AI among researchers add complexity to the field. The NHM AI Lab Programme addresses these challenges by co-creating AI pilot projects that bring together the NHM's collection, academic researchers, and AI experts. The NHM AI Lab Programme serves as a nexus for interdisciplinary collaboration, offering expertise in AI, machine learning, data science, and software engineering to support NHM researchers. Through one-to-one consultations and collaborative research projects, the NHM AI Lab Programme facilitates the integration of innovative AI-driven technologies into streamlining digitisation workflows and enhancing Earth and Life Science research at the NHM. In less than a year since its inception, our Programme has achieved several milestones, hosting around 20 diverse projects. These include research projects such as the application of AI for the automatic detection and identification of nannofossils in chalk, the classification of ancient shark and dinosaur teeth, the prediction of mammal disease outbreaks, and the extraction of data from historical bird egg records. Additional projects focus on the automation of mineral analysis and the detection of secondary impact craters on planetary surfaces using AI. Some led to journal publications (e.g., He et al. 2024), while others streamlined NHM researchers' workflows, enhancing their processes of research and digitisation. Moreover, several initiatives have paved the way for new funding streams and collaborative ventures, as well as promising commercial prospects. Certain projects have pioneered the creation or transformation of datasets to meet AI-ready standards, such as data quality, consistency, accessibility, usability, and data governance protocols, helping to embed AI practices into NHM research. This AI Lab Programme can act as a model for other institutions addressing a similar challenge of bridging the gap between AI and their research and collections. This presentation provides insights into the establishment and operation of the NHM AI Lab Programme, shares experiences, highlights successful collaborations, discusses challenges encountered, and outlines future directions.

Data science education in undergraduate physics: Lessons learned from a community of practice

It is becoming increasingly important that physics educators equip their students with the skills to work with data effectively. However, many educators may lack the necessary training and expertise in data science to teach these skills. To address this gap, we created the Data Science Education Community of Practice (DSECOP), bringing together graduate students and physics educators from different institutions and backgrounds to share best practices and lessons learned from integrating data science into undergraduate physics education. In this article, we present insight and experiences from this community of practice, highlighting key strategies and challenges in incorporating data science into the introductory physics curriculum. Our goal is to provide guidance and inspiration to educators who seek to integrate data science into their teaching, helping to prepare the next generation of physicists for a data-driven world.

Decoding career success: A personality-based analysis of data science Professional based on ANFIS modeling

Career selection is one of the most important decisions every person faces in their life. Finding the right career path can be a complicated task, particularly in choosing careers with similarly required proficiencies. One of the critical factors affecting a person's career success is their personality, and taking account of this factor is of paramount importance. This study uses the NEO-FFI questionnaire to find personality patterns of software engineering and data science experts based on the Big Five personality traits: Neuroticism, extraversion, openness to experience, agreeableness, and conscientiousness. Afterward, an ANFIS (Adaptive Network-Based Inference System) is conducted using the experts' personality data to match the participants of these fields with their corresponding choices. This study demonstrated that data scientists and software engineers score higher in conscientiousness and agreeableness, respectively. Also, data experts have higher scores in all traits overall. In the end, the ANFIS is tested with another similar dataset and the prediction accuracy of the model is measured.

Data Scientist Knowledge and Skills Evaluation towards a Data-Driven Research Methodology

The modern business world increasingly requires a higher level of data science expertise as well as abilities in problem solving and data analytics. Data science is a broad and fast-moving field of methods, processes, algorithms and systems to extract insights from data. The University of Winchester has followed the guidance of the Institute for Apprenticeships and Technical Education (IfATE) and worked in partnership with a number of international, national and regional employers in the design and development of its Data Scientist (integrated degree) programme, which leads to a Bachelor’s degree in Data Science. This degree apprenticeship supports students in gaining the knowledge and skills that are in demand by employers today and into the future, where working in multi-disciplinary teams alongside domain experts will often be the norm. IfATE specifies an End-point Assessment (EPA) plan to enable the apprenticeship to be completed in accordance with its Data Scientist degree apprenticeship standard. This paper considers professional practice and competence in data science and links the processes used in completing the EPA with domain-based knowledge and expertise. It reviews representative solution methodologies before demonstrating the applicability of a data-driven research methodology to discover insights and achieve organisational goals.

Clinical nSights: A software platform to accelerate real world oncology analyses.

e23316 Background: Real world data analyses are critical to assess drug safety and effectiveness, including in patient populations underrepresented in clinical trials. Significant barriers to performing such analyses remain, including: (1) incomplete access to electronic health record (EHR) data due to patient privacy concerns, (2) lack of data harmonization within and across EHRs, (3) requirements for time- and labor-intensive data curation efforts, and (4) necessity of coding or data science expertise given the absence of user-friendly platforms. We present Clinical nSights, a software platform that addresses these challenges to enable researchers and clinicians to perform rapid, robust, and scalable real world oncology analyses. Methods: With input EHR data from multiple health systems, we developed intuitive user interfaces and a scalable researcher friendly programming environment for real world clinical analytics. We used (1) nCognito, a previously described HIPAA-compliant de-identification algorithm, to de-identify EHRs across data modalities; (2) the combination of a de novo knowledge graph, vector and distribution similarity metrics, and manual review to harmonize data types including but not limited to laboratory tests, disease codes, and drugs; and (3) a series of Bidirectional Encoder Representations from Transformers (BERT) models to perform augmented curation of variables of interest (e.g., cancer staging, tumor biomarkers) from unstructured clinical records. We developed a user interface for users to create and analyze defined patient cohorts. Results: Clinical nSights facilitates intuitive cohort creation via flexible specification of inclusion and exclusion criteria, including options to perform temporal linking between criteria and propensity score matching. The ability to specify criteria documented in de-identified clinical notes expands the repertoire of available features for cohort creation beyond those captured in structured EHR fields, including but not limited to cancer stage and tumor biomarker status (e.g., HER2 status, microsatellite instability). Data harmonization enables scalable longitudinal analyses of laboratory test data across cancer types, including changes in measurements in the years leading up to and following initial cancer diagnosis. Augmented curation improves the coverage of phenotypes including immune-related toxicities of checkpoint inhibitors and symptoms experienced by patients prior to a cancer diagnosis. Using the no-code interface, users can easily quantify and visualize outcomes such as overall survival and hospitalization rates. Conclusions: Clinical nSights addresses important challenges in the collection, curation, and analysis of real world oncology data. By enabling a broad range of users to rapidly generate and test hypotheses, this platform has the potential to significantly improve care delivery and patient outcomes.

562 AI Translation Advisory Board: Mastering team science to facilitate implementation of AI into clinical practice

OBJECTIVES/GOALS: Healthcare sectors are rushing to develop AI models. Yet, a dearth of coordinated practices leaves many teams struggling to implement models into practice. The Enterprise AI Translation Advisory Board uses across-disciplinary team to facilitate AI translation. METHODS/STUDY POPULATION: The Mayo Clinic Enterprise AI Translation Advisory Board was established to assess AI solutions lever aging cross-disciplinary team science to accelerate AI innovation and translation. The 23-member board reflects expertise in data science, qualitative research, user experience, IT, human factors, informatics, regulatory compliance,ethics, and clinical care, with members spanning thought leadership, decision-making, and clinical practice. Taking an approach of respectful communication, transparency, scientific debate, and open discussion, the Board has consulted onover two dozen projects at various stages of the AI life cycle. RESULTS/ANTICIPATED RESULTS: Common issues identified for projects earlier in the AI life cycle, sometimes fatal but often address able once identified, include a lack of buy-in from potential product users, a lack of planningabout integration into clinical workflow, inadequately labeled data, and attempting to use machine learning when what is desired is really a causal model for intervening. Recommendations for projects later in the AI life cycle include details of a testing plan (silent evaluation, pragmatic clinical trials), advice about clinical integration, both post-hoc and on going auditing for performance disparities, and planning for regulatory clearance. DISCUSSION/SIGNIFICANCE: Advising is more valuable for projects at the ideation phase, when multi disciplinary interrogation can identify weaknesses. But at all phases, projects have gaps related to a lack of specific disciplinary expertise. A multi disciplinary cluster like the AI Translation Advisory Board seeks to address these gaps.

Analysis and Modelling of Structured Data with Automatic Data Analysis Web Application

Abstract: This paper introduces a novel web application designed for automated analysis and modelling of structured data. The application eliminates the need for user coding, allowing users to interact with the data through a user-friendly interface. By uploading a structured dataset, users can leverage the application's functionalities to improve data quality. These functionalities include, but are not limited to, data cleaning and pre-processing. One of the key strengths of this application is its ability to automatically generate machine learning models based on the uploaded data. Furthermore, the application takes the automation a step further by training the generated models without requiring user intervention. This eliminates the need for data science expertise, making the power of machine learning accessible to a wider audience. This paper details the application's architecture and functionalities, along with an evaluation of its effectiveness in data analysis and model building. The paper discusses the application's potential impact on democratizing data science and its limitations for future research directions.

Theory-Guided Feature Selection in Cybercrime Data Science

Cybercrime data science is being significantly hampered by the presence of 'noisy' features within vast and complex datasets. We draw from the theoretical insights of the behavioural sciences to propose a feature selection model to enrich and improve the value and interpretability of cybercrime intelligence datasets. We piloted our theory-guided feature selection approach on a subset of intelligence datafeeds provided by a global fraud and cybercrime tracking firm. The results of the proposed social influence feature selection model show significant improvement in the interpretability of the machine learning-based exploratory analysis and advanced visualization techniques in an experimental setting. The feature selection model yielded rich insights about cybercriminal psychological tactics from social engineering scam data and has potential applicability in the areas of cyberthreat response and cybersecurity awareness training. Our study shows the value of an interdisciplinary theory-guided approach to cybercrime data analytics that integrates scientific knowledge from the behavioural sciences and data science expertise. Our paper concludes by suggesting avenues for future research on theory-guided feature selection seeking to incorporate behavioural science knowledge in cybercrime data science. We intend to refine, automate, evaluate, and scale our model in future research to assess its effectiveness in producing insights about cybercriminal activities and informing decision-making in a naturalistic and real-time setting. In future research efforts, we aim to automate the encoding of features and apply a wider range of machine learning tools and evaluation metrics to extract more meaningful insights into cybercriminal psychological tactics. We also intend to refine our model on larger datasets to enhance its efficiency and responsiveness to real-time cybercrime data. We call on data scientists and cybercrime domain experts to work together to apply theory-guided feature selection to improve processes of knowledge discovery that enhance our cybersecurity capabilities.

Reproducible WiSDM: a workflow for reproducible invasive alien species risk maps under climate change scenarios using standardized open data

IntroductionSpecies distribution models (SDMs) are often used to produce risk maps to guide conservation management and decision-making with regard to invasive alien species (IAS). However, gathering and harmonizing the required species occurrence and other spatial data, as well as identifying and coding a robust modeling framework for reproducible SDMs, requires expertise in both ecological data science and statistics.MethodsWe developed WiSDM, a semi-automated workflow to democratize the creation of open, reproducible, transparent, invasive alien species risk maps. To facilitate the production of IAS risk maps using WiSDM, we harmonized and openly published climate and land cover data to a 1 km2 resolution with coverage for Europe. Our workflow mitigates spatial sampling bias, identifies highly correlated predictors, creates ensemble models to predict risk, and quantifies spatial autocorrelation. In addition, we present a novel application for assessing the transferability of the model by quantifying and visualizing the confidence of its predictions. All modeling steps, parameters, evaluation statistics, and other outputs are also automatically generated and are saved in a R markdown notebook file.ResultsOur workflow requires minimal input from the user to generate reproducible maps at 1 km2 resolution for standard Intergovernmental Panel on Climate Change (IPCC) greenhouse gas emission representative concentration pathway (RCP) scenarios. The confidence associated with the predicted risk for each 1km2 pixel is also mapped, enabling the intuitive visualization and understanding of how the confidence of the model varies across space and RCP scenarios.DiscussionOur workflow can readily be applied by end users with a basic knowledge of R, does not require expertise in species distribution modeling, and only requires an understanding of the ecological theory underlying species distributions. The risk maps generated by our repeatable workflow can be used to support IAS risk assessment and surveillance.

Simplifying Sentiment Analysis on Social Media: A Step-by-Step Approach

This tutorial presents a systematic guide to performing sentiment analysis on social media data, designed to be accessible to researchers and marketers with varying levels of data science expertise. We prioritise open science by providing comprehensive resources, including self-collected data, source code and guidelines, facilitating result reproduction. For marketing and business researchers without programming experience, this tutorial offers a robust resource for conducting sentiment analysis. Experienced data scientists can use it as a reference for evaluating cutting-edge approaches and streamlining the sentiment analysis process. Our work stands out in its unique perspective on the challenges and opportunities of sentiment analysis within the social media data domain. We delve into the potential of sentiment analysis for social media marketing, offering practical guidance and best practices for enhancing brand reputation and customer engagement. Notably, this tutorial advances beyond previous studies by comprehensively comparing a wide range of sentiment analysis methods, including state-of-the-art transfer learning approaches, filling a critical gap in the existing literature. Our commitment to transparency underscores our contribution, as we provide all necessary resources for result reproducibility. We make our resources available at the following address: https://tinyurl.com/SentimentTutorial .

Current challenges of the state-of-the-art of AI techniques for diagnosing brain tumor

Background: Brain is the control center of the human body, in recent time, different variety of brain diseases are being discovered. The brain disease diagnosis tools are becoming challenging and still an open area of research, application of AI in brain disease diagnosis has made disease prediction and detection more precise and accurate. Automated technologies for non-invasive analysis of brain images have become necessary, because disease of brain is fatal and are the cause of large number of deaths in developed countries. Brain tumor surgery augmented with AI can result in safer and more effective treatment. The knowledge gap between clinical and data science experts still presents significant challenges. This paper will review literatures related to current challenges of AI technologies for brain tumor diagnosis and suggest new directions of AI technologies for diagnosing brain tumour. A systematic search of major academic databases (such as Science Direct, IEEE explore digital Library, and Google scholar) was conducted to identify relevant studies published between 2015 and 2023. The search term used in this study include “Brain tumor Diagnosis”, “AI challenges in Brain tumor Diagnosis”, ‘AI techniques” and “AI challenges in medicine and future”. Studies were included if they utilized AI techniques for brain tumor diagnosis. The identified studies were evaluated for the key challenges they encountered in their diagnostic approaches. The Present study identified several challenges related to the application of AI techniques in brain tumour diagnosis. These challenges include: Interpretability and explainability, variations in tumour location, shape, and size which make accurate segmentation and classification difficult. Overall, the challenges in explaining brain tumor detection stem from the unique requirements and complexities of the healthcare domain, necessitating specialized techniques and approaches. This study summarizes the new directions for AI as (I) Data Hungry: Large, standardized, annotated data sets and excellent ground truth data are necessary for the development of accurate AI. (II) Radiomics: makes it possible to extract a vast number of quantitative features from intricate clinical imaging arrays and convert them into high-dimensional data that can be further processed to determine their relationship to the histological features of the tumor, which represent underlying genetic mutations and malignancy as well as grade, progression, response to therapy, and even overall survival (OS). (III) Black box: AI, for instance, is capable of predicting the best course of care for a patient, but it is unable to explain its reasoning. A trend toward easing this restriction is interpretable deep learning, (IV) Demonstrating the generalizability of deep learning applications and conducting external validation are two major obstacles. (V) There are knowledge gaps in clinical oncology that need to be filled in order to successfully integrate AI and maximize its effects. (VI) Several national professional bodies have started programs to bridge these knowledge gaps and advance the adoption of AI in oncology in response to these difficulties.

Empowering Home Buyers: A Data-Ddriven Approach to Real Estate Pricing in Ames, Iowa

The demand for both housing and investment real estate has significantly increased due to rising urbanization and growing household savings. To address the need for operable real estate pricing models, this study explores potential variables affecting the sale prices of properties in Ames, Iowa, from 2006 to 2010. Utilizing data mining techniques and regression analysis, this study develops a model incorporating 12 independent variables that can be easily obtained during property visits. By offering an easily appliable tool, this research enables potential home buyers to estimate property sale prices, even without extensive expertise in Data Science, Investment, and Economics. The findings demonstrate that analyzing 12 variables directly related to the property itself such as interior finish of the garage, foundation material, and remodeling date can explain approximately 85.2% of the variance in sale prices. Empowering consumers with this knowledge can help reduce the information gap in the real estate market and promote informed decision-making in property purchases.

TELL: A web‐based app to capture speech and language markers of neurodegeneration

AbstractBackgroundOver the last decade, the search for scalable markers of neurodegeneration has been boosted by automated speech and language analysis (ASLA). Participants are asked to produce natural discourse, generating acoustic and linguistic data that can be digitally extracted and analyzed with statistical or machine learning methods. This approach has revealed subtle patterns that differentiate among disorders, predict core symptom severity, correlate with anatomo‐functional brain alterations, and identify autopsy‐confirmed pathology years before death. Accordingly, multiple studies have described ASLA as a promising resource for patient assessment. However, this translational potential remains unfulfilled, as the underlying technologies are unavailable for immediate clinical use. Aiming to tackle this issue, we developed the Toolkit to Examine Lifelike Language (TELL), a web‐based app for collecting, storing, analyzing, and visualizing speech data.MethodTELL version 1.0 was developed by an interdisciplinary team of language, brain health, and data science experts. We designed a speech elicitation protocol comprised of validated spontaneous, semi‐spontaneous, and non‐spontaneous tasks with different cognitive demands. Each task is analyzed via ASLA algorithms capturing word frequency, affective valence, and emotional magnitude, as well as rhythm‐ and pitch‐related features. On‐the‐fly results are offered through intuitive visualizations for each metric. The measures are complemented with a language profile survey. Audio files, transcriptions, and data matrices can be downloaded for offline analysis. The app runs on multiple devices, requires no installation, and offers multi‐language capabilities.ResultTELL version 1.0 is now being used in 10 sites worldwide, mainly including low‐income countries. Preliminary results from persons with Alzheimer’s and Parkinson’s disease, as well as frontotemporal dementia, show that its metrics are sensitive for disease identification, discrimination, and monitoring. Updates have been implemented drawing from users’ feedback. Funds have been secured to develop TELL version 2.0, with novel metrics, improved speech‐to‐text algorithms, and a streamlined front end.ConclusionTELL turns basic language science methods into clinically usable tools, offering an objective, low‐cost, non‐invasive complement to classical assessments. More generally, it illustrates how digital technologies can foster equitable insights on brain health. Ongoing updates will further increase its capacity to establish scalable markers of diverse neurodegenerative diseases.

TELL: A web‐based app to capture speech and language markers of neurodegeneration

AbstractBackgroundOver the last decade, the search for scalable markers of neurodegeneration has been boosted by automated speech and language analysis (ASLA). Participants are asked to produce natural discourse, generating acoustic and linguistic data that can be digitally extracted and analyzed with statistical or machine learning methods. This approach has revealed subtle patterns that differentiate among disorders, predict core symptom severity, correlate with anatomo‐functional brain alterations, and identify autopsy‐confirmed pathology years before death. Accordingly, multiple studies have described ASLA as a promising resource for patient assessment. However, this translational potential remains unfulfilled, as the underlying technologies are unavailable for immediate clinical use. Aiming to tackle this issue, we developed the Toolkit to Examine Lifelike Language (TELL), a web‐based app for collecting, storing, analyzing, and visualizing speech data.MethodTELL version 1.0 was developed by an interdisciplinary team of language, brain health, and data science experts. We designed a speech elicitation protocol comprised of validated spontaneous, semi‐spontaneous, and non‐spontaneous tasks with different cognitive demands. Each task is analyzed via ASLA algorithms capturing word frequency, affective valence, and emotional magnitude, as well as rhythm‐ and pitch‐related features. On‐the‐fly results are offered through intuitive visualizations for each metric. The measures are complemented with a language profile survey. Audio files, transcriptions, and data matrices can be downloaded for offline analysis. The app runs on multiple devices, requires no installation, and offers multi‐language capabilities.ResultsTELL version 1.0 is now being used in 10 sites worldwide, mainly including low‐income countries. Preliminary results from persons with Alzheimer’s and Parkinson’s disease, as well as frontotemporal dementia, show that its metrics are sensitive for disease identification, discrimination, and monitoring. Updates have been implemented drawing from users’ feedback. Funds have been secured to develop TELL version 2.0, with novel metrics, improved speech‐to‐text algorithms, and a streamlined front end.ConclusionTELL turns basic language science methods into clinically usable tools, offering an objective, low‐cost, non‐invasive complement to classical assessments. More generally, it illustrates how digital technologies can foster equitable insights on brain health. Ongoing updates will further increase its capacity to establish scalable markers of diverse neurodegenerative diseases.

Data science training needs in sub-Saharan Africa: Implications for biomedical research and therapeutics capacity

Data use is becoming increasingly valuable worldwide and has positively improved decision-making in various sectors. Data science has revolutionized the study of epidemiology and disease control. Despite the unlimited potential of data science, Africa lags in its innovation and technological advancements. Although sub-Saharan Africa (SSA) has not taken full advantage of data science in healthcare delivery and biomedical research, attempts have been made to harness this opportunity. This article reviews the current level, potentials, and gaps in data science, genomics, and bioinformatics in SSA. We also identified needs associated with developing tools and building capacity in the region, while also acknowledging advances made in these areas to date. We identified opportunities in health data science in SSA, including benefits in combating diseases that burden this region. Applying data science has the potential to create comprehensive healthcare reports and convert them into relevant critical insights that can then be used to provide better care, reduce costs of treatment, predict outbreaks of epidemics, avoid preventable diseases, and improve quality of life in general. In order to achieve this, artificial intelligence and machine learning to build and apply models are some of the common mechanisms applied in all domains of medical research, alongside computational tools and models to simulate systems or expedite biomedical research like bioinformatics and computational biology. Data science, a body of knowledge that uses modern tools and techniques for data management and utilization of big data for innovative interventions and biomedical research is upcoming in Sub-Saharan Africa. Its importance in understanding epidemiology of diseases, public health, and surveillance as well as other domains in allied medical disciplines were highlighted in this review. The dearth in data science skills, expertise, and institutions engaged in training data science needs urgent intervention to optimize the potential uses to improve medicine and healthcare delivery in SSA.

Investigating the value of glucodensity analysis of continuous glucose monitoring data in type 1 diabetes: an exploratory analysis.

Continuous glucose monitoring (CGM) devices capture longitudinal data on interstitial glucose levels and are increasingly used to show the dynamics of diabetes metabolism. Given the complexity of CGM data, it is crucial to extract important patterns hidden in these data through efficient visualization and statistical analysis techniques. In this paper, we adopted the concept of glucodensity, and using a subset of data from an ongoing clinical trial in pediatric individuals and young adults with new-onset type 1 diabetes, we performed a cluster analysis of glucodensities. We assessed the differences among the identified clusters using analysis of variance (ANOVA) with respect to residual pancreatic beta-cell function and some standard CGM-derived parameters such as time in range, time above range, and time below range. Distinct CGM data patterns were identified using cluster analysis based on glucodensities. Statistically significant differences were shown among the clusters with respect to baseline levels of pancreatic beta-cell function surrogate (C-peptide) and with respect to time in range and time above range. Our findings provide supportive evidence for the value of glucodensity in the analysis of CGM data. Some challenges in the modeling of CGM data include unbalanced data structure, missing observations, and many known and unknown confounders, which speaks to the importance of--and provides opportunities for--taking an approach integrating clinical, statistical, and data science expertise in the analysis of these data.

Computational approaches to detect experts in distributed online communities: a case study on Reddit

AbstractThe irreplaceable key to the triumph of Question & Answer (Q & A) platforms is their users providing high-quality answers to the challenging questions posted across various topics of interest. From more than a decade, the expert finding problem attracted much attention in information retrieval research. Based on the encountered gaps in the expert identification across several Q & A portals, we inspect the feasibility of identifying data science experts in Reddit. Our method is based on the manual coding results where two data science experts labelled not only expert and non-expert comments, but also out-of-scope comments, which is a novel contribution to the literature, enabling the identification of more groups of comments across web portals. We present a semi-supervised approach which combines 1113 labelled comments with 100,226 unlabelled comments during training. We proved that it is possible to develop models that can identify expert, non-expert and out-of-scope comments peaking the AUC score at 0.93, accuracy at 0.83, MAE at 0.15 degrees and R2 score at 0.69. The proposed model uses the activity behaviour of every user, including Natural Language Processing (NLP), crowdsourced and user feature sets. We conclude that the NLP and user feature sets contribute the most to the better identification of these three classes. It means that this method can generalise well within the domain. Finally, we make a novel contribution by presenting different types of users in Reddit, which opens many future research directions.

OpenMindat: Open and FAIR mineralogy data from the Mindat database

AbstractThe open data movement has brought revolutionary changes to the field of mineralogy. With a growing number of datasets made available through community efforts, researchers are now able to explore new scientific topics such as mineral ecology, mineral evolution and new classification systems. The recent results have shown that the necessary open data coupled with data science skills and expertise in mineralogy will lead to impressive new scientific discoveries. Yet, feedback from researchers also reflects the needs for better FAIRness of open data, that is, findable, accessible, interoperable and reusable for both humans and machines. In this paper, we present our recent work on building the open data service of Mindat, one of the largest mineral databases in the world. In the past years, Mindat has supported numerous scientific studies but a machine interface for data access has never been established. Through the OpenMindat project we have achieved solid progress on two activities: (1) cleanse data and improve data quality, and (2) build a data sharing platform and establish a machine interface for data query and access. We hope OpenMindat will help address the increasing data needs from researchers in mineralogy for an internationally recognized authoritative database that is fully compliant with the FAIR guiding principles and helps accelerate scientific discoveries.