RESUMEN

Resting-state functional MRI (rs-fMRI) is increasingly employed in multi-site research to analyze neurological disorders, but there exists cross-site/domain data heterogeneity caused by site effects such as differences in scanners/protocols. Existing domain adaptation methods that reduce fMRI heterogeneity generally require accessing source domain data, which is challenging due to privacy concerns and/or data storage burdens. To this end, we propose a source-free collaborative domain adaptation (SCDA) framework using only a pretrained source model and unlabeled target data. Specifically, a multi-perspective feature enrichment method (MFE) is developed to dynamically exploit target fMRIs from multiple views. To facilitate efficient source-to-target knowledge transfer without accessing source data, we initialize MFE using parameters of a pretrained source model. We also introduce an unsupervised pretraining strategy using 3,806 unlabeled fMRIs from three large-scale auxiliary databases. Experimental results on three public and one private datasets show the efficacy of our method in cross-scanner and cross-study prediction.

RESUMEN

BACKGROUND: The American Association of Colleges of Nursing (AACN) and National Organization of Nurse Practitioner Faculties (NONPF) emphasize the role of clinical reasoning in nurse practitioner (NP) competencies. Evidence-based clinical reasoning is vital to patient safety. Collaborative technology tools can aid in assessing progress towards achieving clinical reasoning competency. PURPOSE/AIMS: The purpose of this article is to describe an electronic, collaborative learning framework to teach and assess second year NP students in systematically selecting and eliminating diagnoses and forming treatment plans. DESIGN/METHODS: Post gap analysis, the collaborative learning framework was created. This visual, collaborative resource was scaffolded across two sequential advanced NP second year clinical synthesis courses and embedded with evolving case studies. Students identified pertinent positives and negatives from the history, physical, and diagnostic findings. Each student developed a unique differential diagnosis and plan of care and critiqued their peers. RESULT/FINDINGS: The tool exceeded expectations. Faculty were able to visualize data, provide clarification on interpretation of data and pharmacology, and grade in small groups. CONCLUSION: The collaborative learning framework provided real-time visualization of students' work in clinical reasoning. It was easy to use and integrate into second year NP courses to meet learning objectives and assess clinical reasoning competency.

Asunto(s)

RESUMEN

BACKGROUND: Interprofessional teamwork improves patient care quality, safety, and health outcomes. Interprofessional education (IPE) is crucial in today's medical education to prepare students for the workforce as integral members of a collaborative team. The diversity of IPE learners indicates the importance of exploring the relationship between learning styles and attitudes toward IPE. The purpose of this study was to investigate the relationship between learning styles and attitudes toward IPE. METHODS: A cross-sectional study was conducted between August 2023 and September 2023 in 49 colleges located in the south-eastern region of China. A convenience sampling approach was employed, selecting 500 students majoring in Clinical Medicine and Nursing. The students completed an online questionnaire, which included sociodemographic characteristics, educational characteristics, interprofessional educational characteristics, learning styles, and the readiness for interprofessional learning scale, and Kolb's learning style inventory. Descriptive statistics, Spearman's correlation, and multiple linear regression analysis were used to analyze the data. RESULTS: The most learners are diverger (93.2%), followed by assimilator (3.4%), accommodator (2.6%), and Converger (0.8%). The total score on the RIPLS was 69.70 (7.42), ranging from 48 to 88. A statistical relationship could be established between learning styles and attitudes toward IPE. CONCLUSION: Abstract conceptualization and active experimentation learning modes and convergers were closely linked with positive attitudes toward IPE. Gender, age, and study stress can affect attitudes toward IPE. This study highlights the need for medical education curricula to integrate innovative teaching methods such as PBL, role-playing, scenario simulation and clinical early exposure to strengthen professional identity, and improve abilities related to interprofessional learning.

Asunto(s)

RESUMEN

Assigning students to work in permanent teams is a design principle in Team-based learning (TBL). It has been assumed that a stable team composition supports the emergence of collaborative problem-solving and learning: when students became more familiar with each other, they shared more information and resolved discrepancies together, which in turn stimulated knowledge acquisition and comprehension. However, this assumption had not been probed by a randomized controlled trial with performance assessment as an outcome. In an online course for second term medical students, 50% of the students were reassigned to new teams for each of the 24 problems to be solved during four classes, thus precluding familiarity. The learning outcome was assessed shortly after the third of four classes by a domain knowledge test. Whether TBL teams were permanent or temporary did not affect the score of a domain knowledge test. As expected, participation in online TBL improved the domain knowledge test results. Overall, the permanent team seems to be less important for cognitive learning outcomes than previously assumed, but this may depend on the specific educational setting. However, team familiarity may still be important for team decision-making. As clinical reasoning in the medical workplace often involves collaborating in changing teams, future research on TBL should focus on how to utilize this format to prepare medical students for decision-making and optimal learning outcomes under these conditions.

Asunto(s)

RESUMEN

This study aims to introduce collaborative learning as a method in the Mental Health and Diseases Nursing course for actively enrolled fourth-year nursing students. The objective is to assess the impact of collaborative learning on critical thinking and clinical decision-making among nursing students. Collaborative learning involves the use of small groups to help students identify effective ways of working together to enhance their learning outcomes. A quantitative research methodology was employed in this study. To evaluate the effectiveness of the collaborative learning program, a within-subject repeated-measures design was implemented. This study was conducted in the nursing department of a university in Turkey. The study included 96 female students (92.3 %) and 8 male students (7.7 %). Data collection involved the sociodemographic data form, the California Critical Thinking Disposition Inventory (CCTDI), and the Clinical Decision-Making in Nursing Scale (CDMNS). The students' pre-program CDMNS median score was 146.00 (117.00-177.00), and the post-program CDMNS median score was 147.50 (115.00-175.00). While there was an increase in the students' clinical decision-making scores after the program, this increase was not statistically significant (p > 0.05). The pre-program median score for CCTDI was 223.26 (176.87-296.02), and the post-program median score was 227.88 (188.87-359.00). The students' critical thinking disposition scores showed a statistically significant increase after the program (p < 0.05). The study results revealed notable enhancements among students who participated in courses utilizing the collaborative learning method. These enhancements included heightened academic performance, elevated levels of critical thinking, increased self-confidence, and improved clinical decision-making abilities.

RESUMEN

Introducción: Las redes académicas adquieren elevada significación para la gestión colaborativa de la interdisciplinariedad en la educación de posgrado, de ahí que se suscite el análisis, para su optimización, en el proceso pedagógico de posgrado. El objetivo del presente estudio consistió en revelar criterios teórico-metodológicos para el empleo de las redes académicas en función de la interdisciplinariedad en la educación de posgrado. Métodos: Investigación cualitativa, comprendida entre noviembre de 2022 a septiembre de 2023. Incluyó el empleo de métodos teóricos como el análisis y síntesis, el histórico y lógico, la sistematización y la modelación. Entre los métodos empíricos se aplicó la revisión de documentos y la consulta a especialistas. También se empleó el análisis porcentual para procesar datos. Resultados: Se revelan criterios teórico-metodológicos que fundamentan el empleo de redes académicas en el posgrado, al considerar elementos asociados a la interdisciplinariedad, el trabajo colaborativo, las relaciones interprofesionales e intersectoriales y las alianzas interinstitucionales, a partir de las exigencias de ese nivel educacional. Este resultado es ampliamente generalizable al diseño y gestión de programas de posgrado en sus dos vertientes: superación profesional y formación académica. Discusión: Investigaciones precedentes evidencian un consenso sobre las potencialidades de las redes académicas para el desarrollo del aprendizaje colaborativo, la gestión de proyectos y la práctica interdisciplinaria. Los resultados de este estudio optimizan su empleo en el proceso pedagógico de posgrado. Los criterios teórico-metodológicos revelados en el presente trabajo, tienen un enfoque holístico con elevada pertinencia, según criterios valorativos de los especialistas que participaron en el estudio.

Introduction: Academic networks acquire high significance for the collaborative management of interdisciplinarity in postgraduate education, hence the analysis arises, for its optimization, in the postgraduate pedagogical process. The objective of this study was to reveal theoretical-methodological criteria for the use of academic networks based on interdisciplinarity in postgraduate education. Methods: Qualitative research, between november 2022 and september 2023. It included the use of theoretical methods such as analysis and synthesis, historical and logical, systematization and modeling. Among the empirical methods, document review and consultation with specialists were applied. Percentage analysis was also used to process data. Results: Theoretical-methodological criteria are revealed that support the use of academic networks in postgraduate studies, when considering elements associated with interdisciplinarity, collaborative work, interprofessional and intersectoral relationships and inter-institutional alliances, based on the demands of that educational level. This result is widely generalizable to the design and management of postgraduate programs in its two aspects: professional development and academic training. Discussion: Previous research shows a consensus on the potential of academic networks for the development of collaborative learning, project management and interdisciplinary practice. The results of this study optimize its use in the postgraduate pedagogical process. The theoretical-methodological criteria revealed in this work have a holistic approach with high relevance, according to the evaluation criteria of the specialists who participated in the study.

RESUMEN

Collaborative methods for learning such as team-based learning, case-based learning, and problem-based learning have become leading methods for active learning within the field of health professions education. Critical thinking and exposure to diverse perspectives within a group are highlighted as important benefits of team-based learning. However, without consideration of the potential socioemotional, behavioral, and team dynamic challenges associated with this format of learning, the tenets which support a welcoming learning community can begin to deteriorate. In this essay, authors highlight the core concepts of growth mindset, psychological safety, and culturally responsive pedagogy within a framework of 1) what learners should know and 2) what educators can do, to co-create an inclusive collaborative learning experience.

RESUMEN

Introduction: The Pediatric Rheumatology Care and Outcomes Improvement Network (PR-COIN) is a North American learning health network focused on improving outcomes of children with juvenile idiopathic arthritis (JIA). JIA is a chronic autoimmune disease that can lead to morbidity related to persistent joint and ocular inflammation. PR-COIN has a shared patient registry that tracks twenty quality measures including ten outcome measures of which six are related to disease activity. The network's global aim, set in 2021, was to increase the percent of patients with oligoarticular or polyarticular JIA that had an inactive or low disease activity state from 76% to 80% by the end of 2023. Methods: Twenty-three hospitals participate in PR-COIN, with over 7,200 active patients with JIA. The disease activity outcome measures include active joint count, physician global assessment of disease activity, and measures related to validated composite disease activity scoring systems including inactive or low disease activity by the 10-joint clinical Juvenile Arthritis Disease Activity Score (cJADAS10), inactive or low disease activity by cJADAS10 at 6 months post-diagnosis, mean cJADAS10 score, and the American College of Rheumatology (ACR) provisional criteria for clinical inactive disease. Data is collated to measure network performance, which is displayed on run and control charts. Network-wide interventions have included pre-visit planning, shared decision making, self-management support, population health management, and utilizing a Treat to Target approach to care. Results: Five outcome measures related to disease activity have demonstrated significant improvement over time. The percent of patients with inactive or low disease activity by cJADAS10 surpassed our goal with current network performance at 81%. Clinical inactive disease by ACR provisional criteria improved from 46% to 60%. The mean cJADAS10 score decreased from 4.3 to 2.6, and the mean active joint count declined from 1.5 to 0.7. Mean physician global assessment of disease activity significantly improved from 1 to 0.6. Conclusions: PR-COIN has shown significant improvement in disease activity metrics for patients with JIA. The network will continue to work on both site-specific and collaborative efforts to improve outcomes for children with JIA with attention to health equity, severity adjustment, and data quality.

RESUMEN

BACKGROUND: The increasing complexity of the healthcare environment and the necessity of multidisciplinary teamwork have highlighted the importance of interprofessional education (IPE). IPE aims to enhance the quality of patient care through collaborative education involving various healthcare professionals, such as doctors, nurses, and pharmacists. This study sought to analyze how game-based IPE activities influence students' perceptions and reflective thinking. It also aimed to identify the shifts in perception and effectiveness caused by this educational approach. METHODS: The study is based on a game-based IPE program conducted at University A, involving medical and nursing students in structured learning and team-based activities. Data were collected using essays written by the students after they had participated in IPE activities. Text network analysis was conducted by extracting key terms, performing centrality analysis, and visualizing topic modeling to identify changes in students' perceptions and reflective thinking. RESULTS: Keywords such as "patient," "thought," "group," "doctor," "nurse," and "communication" played a crucial role in the network, indicating that students prioritized enhancing their communication and problem-solving skills within the educational environment. The topic modeling results identified three main topics, each demonstrating the positive influence of game-based collaborative activities, interprofessional perspectives, and interdisciplinary educational experiences on students. Topic 3 (interdisciplinary educational experience) acted as a significant mediator connecting Topic 1 (game-based collaborative activity experience) and Topic 2 (interprofessional perspectives). CONCLUSION: This study demonstrates that game-based IPE activities are an effective educational approach for enhancing students' team building skills, particularly communication and interprofessional perspectives. Based on these findings, future IPE programs should focus on creating collaborative learning environments, strengthening communication skills, and promoting interdisciplinary education. The findings provide essential insights for educational designers and medical educators to enhance the effectiveness of IPE programs. Future research should assess the long-term impacts of game-based IPE on clinical practice, patient outcomes, and participants' professional development.

Asunto(s)

RESUMEN

This article reports a systematic literature review that examined past research exploring the effectiveness of collaborative Augmented Reality (AR) enabled instruction, in higher education contexts. To be included, an article should consist of an experimental study investigating the use of collaborative AR for learning in higher education settings. An initial search was conducted on five databases that resulted in a total of 2537 articles, of which 20 were finalized for this review. The main findings suggest that AR-enabled collaborative learning benefits students' overall learning outcomes and provides a positive collaboration experience in higher education settings. Further research is needed to determine the interaction elements, collaboration mechanisms, and information representation through AR that would potentially enhance student learning outcomes. This article concludes by discussing the implications of these findings, identifying challenges and strategies for developing effective collaborative AR-enabled learning content.

Asunto(s)

RESUMEN

To develop independent healthcare professionals able to collaborate in interprofessional teams, health professions education aims to support students in transitioning from an individual perspective to interprofessional collaboration. The five elements that yield the conditions for effective interprofessional collaboration are: (1) positive interdependence, (2) individual accountability, (3) promotive interaction, (4) interpersonal skills, and (5) reflection on team processes. The aim of the current study is to gain insights into how to design tasks to assess a student team as a whole on their interprofessional collaboration. This was a pilot study using a qualitative design to evaluate an interprofessional assessment task. Four interprofessional student teams, comprising physiotherapy, occupational therapy, arts therapy and nursing students (N = 13), completed this task and five assessors used a rubric to assess video recordings of the teams' task completion, and then participated in a group interview. The completed rubrics and the interview transcript were analyzed using content analysis. Findings showed that the combination of individual preparation, an interprofessional team meeting resulting in care agreements and team reflection was a strength of the assessment task, enabling the task to elicit sufficient promotive interaction between students. Areas for improvement of the assessment task were however, due to a lack of interdependence, the care agreements which now proved to be the sum of students' intraprofessional ideas rather than an interprofessional integration of agreements. Additionally, assessors suggested that a series of varying assessment tasks is required to draw conclusions about students' interprofessional competence.

Asunto(s)

RESUMEN

BACKGROUND: Interprofessional education (IPE) is essential to foster collaboration among healthcare professionals for holistic patient care. However, Malaysian dental education remains discipline-centric, hindering multidisciplinary learning approaches. Hence, this study aimed to explore Malaysian undergraduate dental students' perceptions of IPE. METHODS: The present cross-sectional study employed convenience sampling to survey undergraduate dental students from four Malaysian institutions using a modified questionnaire with 20 close-ended and 2 open-ended questions. The questionnaire covered three domains (effectiveness, preference, importance) to assess students' perceptions using a five-point Likert scale. Psychometric validation was performed to assure validity and reliability of the modified questionnaire. Quantitative analysis (descriptive and inferential statistics), and qualitative analysis (content analysis) were subsequently performed. RESULTS: 397 students responded, and positive perceptions were generally noted with mean scores ranging from 4.13 to 4.35 across all domains. Questions 2 and 3, assessing the improvement in understanding the roles and responsibilities, and communication among healthcare professionals, received the highest mean scores. Meanwhile, Question 15 concerning the incorporation of IPE into educational goals received the lowest mean score. Regression analysis identified gender and clinical phase as significant factors, with females and preclinical students exhibiting more favourable perceptions. Motivators for IPE included a keen interest in diverse perspectives and recognising the importance of teamwork, while barriers encompassed tightly packed schedules, lack of understanding about IPE, misconceptions regarding dental education, and students' nervousness and fear of participation. CONCLUSION: This study produced a valid and reliable instrument to measure undergraduate dental students' perceptions towards IPE. Strategic planning, such as overcoming logistical challenges, improving awareness, and creating a supportive learning environment are crucial for successful IPE integration into existing curricula, especially in resource-constrained developing countries like Malaysia.

Asunto(s)

RESUMEN

Introduction: The rapid development of artificial intelligence (AI) in healthcare has exposed the unmet need for growing a multidisciplinary workforce that can collaborate effectively in the learning health systems. Maximizing the synergy among multiple teams is critical for Collaborative AI in Healthcare. Methods: We have developed a series of data, tools, and educational resources for cultivating the next generation of multidisciplinary workforce for Collaborative AI in Healthcare. We built bulk-natural language processing pipelines to extract structured information from clinical notes and stored them in common data models. We developed multimodal AI/machine learning (ML) tools and tutorials to enrich the toolbox of the multidisciplinary workforce to analyze multimodal healthcare data. We have created a fertile ground to cross-pollinate clinicians and AI scientists and train the next generation of AI health workforce to collaborate effectively. Results: Our work has democratized access to unstructured health information, AI/ML tools and resources for healthcare, and collaborative education resources. From 2017 to 2022, this has enabled studies in multiple clinical specialties resulting in 68 peer-reviewed publications. In 2022, our cross-discipline efforts converged and institutionalized into the Center for Collaborative AI in Healthcare. Conclusions: Our Collaborative AI in Healthcare initiatives has created valuable educational and practical resources. They have enabled more clinicians, scientists, and hospital administrators to successfully apply AI methods in their daily research and practice, develop closer collaborations, and advanced the institution-level learning health system.

RESUMEN

OBJECTIVES: To evaluate the learning gain and students' perceptions towards Jigsaw collaborative learning in comparison with lectures in learning about pharmacokinetic changes in special populations. METHODS: Undergraduates learn about A-D-M-E of specific populations via Jigsaw collaborative learning and didactic lectures. Pre- and post-lesson quizzes were conducted to evaluate the effectiveness of the teaching method in terms of knowledge gain. Surveys comprising Likert scale statements and open-ended questions were conducted to elucidate students' perception towards the teaching methods. RESULTS: From a class of 192 students, 118 (62%) and 110 (57%) students completed the pre- and post-lecture quizzes, respectively, while 176 (92%) and 168 (88%) students completed the pre- and post-Home Group discussion of Jigsaw quizzes, respectively. There was an improvement of 22.2% and 14.3% in median percentage quiz scores for the lecture and Jigsaw method respectively. Most students agreed that they have learned (54-60%) and collaborated (78-89%) through the Jigsaw method and rated Jigsaw as useful for their learning (54%). Open-ended survey responses offered a mixed conclusion where the didactic lecture was perceived to be as, or more effective than the Jigsaw method. CONCLUSION: Learning gains were observed through the Jigsaw collaborative learning method which relied solely on peer-teaching, despite students perceiving it to be not as effective as lecture. The method provided opportunities for active and peer-learning. Further studies are needed to evaluate the long-term effects of this teaching method.

RESUMEN

Deep learning from collaboration occurs if the learner enacts interactive activities in the sense of leveraging the knowledge externalized by co-learners as resource for own inferencing processes and if these interactive activities in turn promote the learner's deep comprehension outcomes. This experimental study investigates whether inducing dyad members to enact constructive preparation activities can promote deep learning from subsequent collaboration while examining prior knowledge as moderator. In a digital collaborative learning environment, 122 non-expert university students assigned to 61 dyads studied a text about the human circulatory system and then prepared individually for collaboration according to their experimental conditions: the preparation tasks varied across dyads with respect to their generativity, that is, the degree to which they required the learners to enact constructive activities (note-taking, compare-contrast, or explanation). After externalizing their answer to the task, learners in all conditions inspected their partner's externalization and then jointly discussed their text understanding via chat. Results showed that more rather than less generative tasks fostered constructive preparation but not interactive collaboration activities or deep comprehension outcomes. Moderated mediation analyses considering actor and partner effects indicated the indirect effects of constructive preparation activities on deep comprehension outcomes via interactive activities to depend on prior knowledge: when own prior knowledge was relatively low, self-performed but not partner-performed constructive preparation activities were beneficial. When own prior knowledge was relatively high, partner-performed constructive preparation activities were conducive while one's own were ineffective or even detrimental. Given these differential effects, suggestions are made for optimizing the instructional design around generative preparation tasks to streamline the effectiveness of constructive preparation activities for deep learning from digital collaboration.

RESUMEN

This article is based on pharmacy students using technology in collaborative learning and offers insights for tutors conducting clinical case study discussions. Reflection revealed that collaborative learning, fostering active interaction and asymmetrical roles, positively impacted student engagement. It proved beneficial when the goal was to enhance learners' engagement in a shared task with mutual accountability. Tutors are also encouraged to view teaching as a means of developing students' learning abilities rather than information transmission. This article emphasised the significance of technology, particularly Padlet, in this process, identified by students as an enabler of learning. Suggestions for future development include exploring variables affecting collaborative learning and understanding technology's role in supporting this method. Overall, this reflective work provided valuable guidance for tutors seeking to optimise collaborative learning experiences in case study discussions.

RESUMEN

In this study, we assessed a highly structured, yearlong, case-based course designed for undergraduate pre-health students. We incorporated both content learning assessments and developed a novel method called Multiple Mini Exams for assessing course impact on the development of skills that professional schools often seek in pre-health students, focusing on students' abilities to collaborate with others, display bedside manners, synthesize patient case details, appropriately use scientific and medical language, and effectively attain patients' medical histories. This novel method utilized a rubric based on desired medical student skills to score videotaped behaviors and interactions of students role playing as doctors in a hypothetical patient case study scenario. Overall, our findings demonstrate that a highly structured course, incorporating weekly student performance and presentation of patient cases encompassing history taking, diagnosis, and treatment, can result in content learning, as well as improve desired skills specific for success in medical fields.

RESUMEN

In this manuscript, we develop a multi-party framework tailored for multiple data contributors seeking machine learning insights from combined data sources. Grounded in statistical learning principles, we introduce the Multi-Key Homomorphic Encryption Logistic Regression (MK-HELR) algorithm, designed to execute logistic regression on encrypted multi-party data. Given that models built on aggregated datasets often demonstrate superior generalization capabilities, our approach offers data contributors the collective strength of shared data while ensuring their original data remains private due to encryption. Apart from facilitating logistic regression on combined encrypted data from diverse sources, this algorithm creates a collaborative learning environment with dynamic membership. Notably, it can seamlessly incorporate new participants during the learning process, addressing the key limitation of prior methods that demanded a predetermined number of contributors to be set before the learning process begins. This flexibility is crucial in real-world scenarios, accommodating varying data contribution timelines and unanticipated fluctuations in participant numbers, due to additions and departures. Using the AI4I public predictive maintenance dataset, we demonstrate the MK-HELR algorithm, setting the stage for further research in secure, dynamic, and collaborative multi-party learning scenarios.

RESUMEN

BACKGROUND: Staff shortages limit access to health services. The bidirectional benefits of allied health clinical placements are understood in the domains of student learning, health service delivery, and future workforce development. Still, the benefits to current workforce outcomes remain unknown. This review provides insights into the effects of allied health student placements in acute and primary care settings, particularly on healthcare staff's knowledge and procedural skills. METHODS: This search was based on the integrative review process established by Whittemore and Knafl in 2005. In October 2023, the first author (MH) searched five major electronic databases: Medline-EBSCO, PubMed, CINAHL, Embase, and Scopus. The CLUSTER model was used to track additional references. The first three authors (MH, SM, and SC) were involved in screening, quality appraisal, and synthesis of the studies. Data were thematically synthesised and analysed. RESULTS: MeSH headings and keywords were used in key search areas: health education, health professional training, clinical placements, and allied health professions. The systematic search yielded 12 papers on allied health student placements across various healthcare settings in rural and metropolitan areas, with no high-quality methodologies measuring student placements' impact on staff knowledge and skills. Four main themes were identified from the analysis: meaningful student integration in service delivery, targeted educational support to healthcare staff, development of staff procedural skills and confidence, and the mechanisms of why student placements work in this aspect. CONCLUSIONS: This review suggests that offering allied health student placement could be a promising approach to supporting rural healthcare staff in performing patient assessments and treatments proficiently and collaboratively. However, this requires further investigation to confirm.

Asunto(s)

RESUMEN

BACKGROUND: Consider a setting where multiple parties holding sensitive data aim to collaboratively learn population level statistics, but pooling the sensitive data sets is not possible due to privacy concerns and parties are unable to engage in centrally coordinated joint computation. We study the feasibility of combining privacy preserving synthetic data sets in place of the original data for collaborative learning on real-world health data from the UK Biobank. METHODS: We perform an empirical evaluation based on an existing prospective cohort study from the literature. Multiple parties were simulated by splitting the UK Biobank cohort along assessment centers, for which we generate synthetic data using differentially private generative modelling techniques. We then apply the original study's Poisson regression analysis on the combined synthetic data sets and evaluate the effects of 1) the size of local data set, 2) the number of participating parties, and 3) local shifts in distributions, on the obtained likelihood scores. RESULTS: We discover that parties engaging in the collaborative learning via shared synthetic data obtain more accurate estimates of the regression parameters compared to using only their local data. This finding extends to the difficult case of small heterogeneous data sets. Furthermore, the more parties participate, the larger and more consistent the improvements become up to a certain limit. Finally, we find that data sharing can especially help parties whose data contain underrepresented groups to perform better-adjusted analysis for said groups. CONCLUSIONS: Based on our results we conclude that sharing of synthetic data is a viable method for enabling learning from sensitive data without violating privacy constraints even if individual data sets are small or do not represent the overall population well. Lack of access to distributed sensitive data is often a bottleneck in biomedical research, which our study shows can be alleviated with privacy-preserving collaborative learning methods.

Asunto(s)