Projects / Programmes
Methodology for data analysisi in medical sciences
January 1, 2022
- December 31, 2027
| Code |
Science |
Field |
Subfield |
| 3.08.00 |
Medical sciences |
Public health (occupational safety) |
|
| 1.01.00 |
Natural sciences and mathematics |
Mathematics |
|
| Code |
Science |
Field |
| 3.03 |
Medical and Health Sciences |
Health sciences |
| 1.01 |
Natural Sciences |
Mathematics |
Biostatistics, survival analysis, relative survival, explained variation, high-dimensional data, classification, imbalance, boosting, regression models, quality, bibliometry, literature discovery, epidemiological models, initial data analysis
Data for the last 5 years (citations for the last 10 years) on
April 26, 2024;
A3 for period
2018-2022
| Database |
Linked records |
Citations |
Pure citations |
Average pure citations |
| WoS |
556 |
13,488 |
12,676 |
22.8 |
| Scopus |
569 |
16,390 |
15,520 |
27.28 |
Researchers (24)
Organisations (1)
Abstract
We work with methodology related to various areas of data analysis in medicine. A Survival analysis Our group has established itself as one of the world's leading research groups in the field of relative survival through its research and important publications, and we will continue our work in this area intensively. We currently work on the first book on relative survival (Chapman&Hall/CRC). Our future goals are to: (i) shed light on the key assumptions of net survival; (ii) define a new measure and estimator for the number of years lost or gained in comparison to the general population; (iii) incorporate relative survival into multi-state models and library mstate; (iv) use pseudo-observations in regression models for the number of years lost or gained; (v) evaluate the effectiveness of cancer screening programmes according to the number of years gained; (vi) generalise the measure of explained variation of time to event and to eliminate its limitations. B Estimation in regression models In this broad field, our goals are to: (i) eliminate bias in the estimation of the (conditional) odds ratio in case-control studies; (ii) solve the problem of data separation in the analysis of binary outcome for conditional (mixed) and marginal regression models; (iii) develop a new approach to the evaluation of the penalty parameter that eliminates the shortcomings of the existing one; (iv) continue research in the field of fitting regression models, where we have already developed a new method for ordinary linear regression and are now exploring its properties and developing a new approach for linear mixed models, together with a comprehensive software library for fitting regression models. C Initial data analysis We are part of the STRATOS group for Initial Data Analysis (IDA), which takes place between the end of data collection and the beginning of analysis. Our goal is to provide recommendations how to plan and perform IDA for specific types of studies, first for longitudinal studies. D Knowledge technologies We are focused on the development and integration of statistical approaches, machine learning, and complex networks to solve open problems in biomedicine, biology, and drug repositioning. We will continue our research in the fields of: (i) complex networks analysis, literature-based discovery; (ii) heterogeneous data analysis, fusion of massive datasets. E Scientometrics, bibliometrics and informatics In the last funding period, we published four papers in prestigious journals. We will develop a model of representational learning that considers not only network topology, but also attributes on edges. F Statistical support in medical research Our research group will continue to provide statistical support to Slovene medical researchers, which is the main practical purpose of this group. In the period from 2015 to date, we have published 155 scientific papers in collaboration with Slovenian medical researchers in impact factor journals.
Significance for science
The primary goal of our research group is the development of statistical methodology related to data analysis in medicine. Our knowledge and understanding, past experiences and publications, staff and the established close collaboration with foreign scientists enable us to contribute to the very best world development of biostatistical science in our subfields. Moreover, we are the only Slovenian programme group that deals with the methodological development of medical statistics. This means that our work is crucial for the development of this scientific and professional field in Slovenia. As the main practical purpose of the group is statistical support for Slovene medical researchers, our work consequentially contributes to the development of science in many areas of medicine. Without proper planning and analysis, exceptional research in many fields of medicine is impossible. We are part of the STRATOS initiative (STRengthening Analytical Thinking for Observational Studies), a large collaboration of experts in many different areas of biostatistical research with the objective to provide accessible and accurate guidance in the design and analysis of observational studies. It consists of nine groups, and we have members in three of them (invited members only): Survival analysis group, High-dimensional data group, and Initial data analysis group. Through research within these groups, we co-create new statistical guidelines. In addition, we are members of numerous programme committees of international conferences and editorial boards of international journals. Our research interests and commitments are extremely broad given the current size of the team. The choice of research areas depends heavily on the topics that prove important during our work with medical researchers. Below, we present the potential impact of our research by individual fields. For many years, the main research topic of our programme group has been survival analysis. This results also from close cooperation with Slovenian Cancer Registry. Our work in the fields of explained variation and relative survival place our research group among the best in the world. In the last decade, a big step forward has been achieved in the field of relative survival and we can proudly say that our research group has played a key role in its development. In 2012, after several decades of using biased estimators, members of our research team proposed a consistent estimator for net survival, i.e. Pohar Perme estimator. The article was highly acclaimed both among statisticians in the field of relative survival as well as among end-users (epidemiologists), also on the account of the user-friendly library. In the future, we will continue our work in this field, and we will focus on new measures and estimators in the area of relative survival (number of years lost or gained in comparison to the general population). Due to the importance of these questions, the international status of our research team, and our rich experience, we believe that the new methodology will be recognized by the relative survival community and influence further developments in this field. This is further justified by our current work (as editors and co-authors) on a book focusing on relative survival that will be published by Chapman & Hall/CRC. This will be the first book on relative survival and will fill an important gap in the literature. We will use the aforementioned survival analysis experience to evaluate the effectiveness of cancer screening programmes, which is a burning issue, currently without clear answers. The results of this research will certainly be of interest to both the scientific community and general public. On the other hand, the development of topics such as properties of multi-state models, pseudo-observations, etc., is crucial for the theoretical development of the different fields and thus the development of statistical science in general. Nevertheless, we will illustrate the use of all methods using practical examples and make sure that all obtained knowledge will be transferred from theory to practice eventually. Another area where we will be very active is the broad field of regression models, where we want to solve some open problems and extend our previous work. There is currently no unbiased estimator of the conditional odds ratio in case-control studies, which are one of the principal and most commonly used study designs in medicine. We are developing an approach that eliminates bias in collaboration with recognised foreign researchers (Georg Heinze and Ioannis Kosmidis), which ensures the visibility of our work in the international community. Similarly, we will collaborate with researchers from Medical University of Vienna to solve the data separation problem. Finally, the area of fitting regression models has been the focus of our group for many years. We will thoroughly investigate the properties of the methods that we have already developed, and develop new methods for the next group of regression models. Since there is no comprehensive software library in the area of fitting regression models yet, we intend to create one and thus contribute to the broader application of the new and better statistical approaches. With the development of methods in the area of complex network analysis, machine learning, and multivariate analysis while merging heterogeneous data types (e.g., structured, unstructured, relational), we are setting the stage for an integrative paradigm in modern data analysis. In the context of scientometric and bibliometric research, the results of our work will directly benefit researchers from different scientific fields, policy-makers, maintainers of bibliographic databases, science-of-science researchers, and last but not least entrepreneurs looking for new market opportunities in high technology.
Significance for the country
Most of the topics of our scientific research in various areas of methodology for data analysis in medical sciences are derived from practical problems that prove interesting when working with medical researchers. Consequently, our results are directly applicable in biostatistics, but they are usually useful in a variety of research fields where the need for data analysis arises. An example of such a co-dependence is also our main topic of relative survival analysis - close collaboration with Slovenian Cancer Registry and their need for experts in this field have fostered our long-term commitment to this area. The methods that we have already developed in this area (net survival estimator) and the new methods we are currently developing (the number of years lost or gained in comparison the general population) enable the assessment of the cancer burden in Slovenia. Furthermore, we plan to evaluate the effectiveness of cancer screening programmes by the number of years gained, which is a very interesting question, but currently there are no clear answers due to the different types of biases that arise. The findings in this area will be important both for the medical experts and in the long term for the national strategy of screening programmes. The COVID-19 pandemic is another important indicator of the importance of our research. At the start of this epidemic, Slovenia had no experts in the field of epidemiological models. In the light of this need, our research group temporarily devoted all research to the intensive development of expertise in this field. We did not limit ourselves only to the analysis of current data and short-term forecasts, but also prepared two in-depth methodological papers (one already published, the other submitted for publication in an international journal), as only a thorough understanding of the theoretical and practical properties of models allows a high-quality and reliable analysis. The results enable the evaluation of the effectiveness of non-pharmaceutical interventions, giving stakeholders the opportunity to make data-driven decisions. We intend to further develop our expertise in this field at least during this epidemic, possibly even long-term if we will be able to ensure enough resources to maintain so many different research topics at an internationally comparable level. We have cooperated with COVID-19 state decision-makers, both as consultants within COVID-19 Sledilnik (Tracker) and independently (cooperation with COVID-19 expert group, cooperation with the Ministry of Health, participation in a session of the Slovene National Assembly). Our research work has expanded into various areas of data analysis, both in the short and long term, as we are able to quickly adapt to the needs of the field of medicine and others. This is possible on the account of many years of experience and knowledge in the scientific research of statistical methodology. We are the only Slovenian programme group dealing with methodological development of medical statistics, which makes our work crucial for the development of this scientific and professional field in Slovenia. For many years, we have been active in the development of the statistical profession in Slovenia. We are the co-founders of the first and only master's programme in statistics in Slovenia (Applied Statistics at University of Ljubljana). Maja Pohar Perme is the chair of this study programme, many members of our group are lecturers and mentors. We also teach statistical courses on multiple faculties on the University of Ljubljana (Faculty of Medicine, Faculty of Veterinary Medicine, Faculty of Health Sciences) and University of Primorska (Faculty of Mathematics, Natural Sciences and Information Technologies, Faculty of Health Sciences). At the University of Ljubljana, we are involved in two PhD programmes, Statistics and Biomedicine. In the last funding period, we have been mentors to 7 PhDs, 3 of them were a part ARRS Young Researcher Programme. We are members of statistical council and methodological council at Statistical Office of the Republic of Slovenia. Many members actively participate in the Statistical Society of Slovenia. Our members have been on the participation and programme committee of the international conference Applied Statistics. We are co-editors of the journals Advances in Methodology and Statistics (Metodološki zvezki) and Slovenian Medical Journal (Zdravniški vestnik), and Editors-in-Chief of the journal Informatica Medica Slovenica. In 2020, Lara Lusa received the Zois Distinction for scientific achievements in the development of methodology in the field of medical statistics. We are active in the popularisation of statistics and science in general. We are a part of the project Zdrava Glava for popularisation of science, a part of COVID-19 Sledilnik (Tracker), and have been guests in several informative and educational programmes in Slovene media as experts in statistics. We have developed and maintained various data management information systems (secure and reliable multicentre data collection and storage, together with a complex search engine and the possibility to publish statistical analyses) for clinical research, registries and quality management (e.g. Institute of Oncology, pharmaceutical companies, healthcare providers). Information systems are built using our own platform, allowing for a great flexibility in terms of different types of collected data (data models) and simplicity for end users.
