Keywords
Public health - epidemiology informatics - IMIA Yearbook 2022
1 Introduction
There are various types of health informatics, which is the confluence of people, technology, and data to improve the safety and quality of patient care. Patient portals, electronic health records (EHRs), telehealth, healthcare apps, and a range of data reporting tools are all just a few examples. Nowadays, solving pressing public health needs requires access to and cross-referencing of various types of data and the use of approaches and techniques from different fields: epidemiology and population health studies, bio-statistics and machine learning, big data and heterogeneous data integration, etc. In this context, there are currently various initiatives around the world consisting in enabling health related data sharing frameworks. For instance, in Europe, the European Health Data Space (EHDS)[1] is a recent initiative which aims at providing a health-specific data sharing framework that establishes clear rules, common standards and practices, infrastructures, and a governance framework for patients to use electronic health data for research, innovation, policy-making, patient safety, statistics, and regulatory purposes. An important pillar of the EHDS is the secondary use of data for better research and policy-making. Similarly, in the United States (US), the National COVID Cohort Collaborative (N3C) is worth mentioning. N3C maintains one of the largest collections of clinical data related to Covid-19 symptoms and patient outcomes in the US[2] allowing many organisations to give clinical data in near-real time, enabling to learn more about COVID-19 and potential treatment options. Indeed, early in the current COVID-19 pandemic, issues arose that highlighted the need for public health authorities to combine diagnostic, clinical, and demographic data at scale to facilitate real-time monitoring and decision-making [[1]]. Overall, EHRs and databases holding other health-related data (claims, pharmacy) can help support observational research and pragmatic clinical trials, which are both valuable sources of real-world evidence [[2]].
The current Public Health and Epidemiology Informatics (PHEI) synthesis, which follows the last edition of the IMIA Yearbook, reviews the scientific literature published in 2021 in the domain of medical informatics and which addresses the above challenges. Targeted studies focus more specifically on the sub-field of public health and epidemiology. In comparison to past years, it tries to find new subjects and trends within this discipline. It also explains how the best articles published in 2021 were chosen through a selection procedure. To that purpose, the section editors oversaw the peer review of the most representative research papers, which were chosen for their quality and originality.
2 Methods
As with the previous edition of the IMIA Yearbook for the PHEI section [[3]], a comprehensive literature search was performed by the section editors using the PubMed/Medline database from the National Center for Biotechnology Information (NCBI). A large set of MeSH descriptors was used to retrieve relevant studies. The queries targeted public health or epidemiological journal articles over the year 2021 and that included medical informatics topics. Returned references addressing topics of the other sections of the Yearbook, e.g., those related to COVID-19 or sensors, were excluded from our search.
The study was performed early March 2021, and the search returned a total of 782 references, which is lower than for the 2021 Yearbook edition. Articles were reviewed separately by the two section editors, considering the number of citations for each reference in the process. They were first classified into three categories: keep, discard, or pending using the BibReview tool [[4]]. Then, the two lists of references were merged, yielding 88 references that were retained by at least one section editor or classified as “pending” by both of them. The two section editors jointly reviewed these 88 references and selected a consensual list of 11 candidate best papers. These papers are depicted in [Table 1].
Table 1 List of the 11 candidate papers for the PHEI section.
All of these papers were then peer-reviewed by the PHEI section editors and external reviewers. Each paper was reviewed by at least five reviewers to ensure a very informed decision making. Three papers emerged at the end of this process. However, the third paper [[5]], which addresses the detection of medical misinformation, although very relevant to the section, was removed due to the lack of an appropriate discussion section about the ethical consideration for using social media data. The two papers that have been finally selected, respectively [[6]] and [[7]] both rely on Machine Learning (ML) approaches applied to large datasets of patients. A summary of these papers is provided in the appendix.
3 Results
3.1 Overall Observation
Broad trends in public health and epidemiology informatics include an increased interest in surveillance methods based on mobile apps for early detection of disease outbreaks [[8]]. Public health issues related to the opioid drug epidemic [[9]], antimicrobial resistance [[10]], or the impact of environmental pollution [[11]] are more and more analysed on a wide scale using data sources such as social media, internet search data, or Wikitrends and on a variety of modalities including images, text and spatial data. The practice of recruiting participants for large-scale studies through online social media campaigns has seen an increased adoption [[12]]. In terms of informatics approaches, ontologies have matured to the level of worldwide standards for health data exchange [[13]]. The impact of misinformation and rumors during an epidemic outbreak was modeled using social networks [[14]]. In addition, the focus has moved from simply applying and increasing performance of deep learning approaches [[7]] to explainable machine learning and in depth analysis of salient features [[6]].
3.2 Selected Papers
Thus, among the 782 retrieved articles from the PHEI section’s relevant keywords, two best papers have been selected this year ([Table 2]). The first study focuses on explainability in the context of ML [[6]] while the second study is a multi-centric study to identify hepatobiliary diseases from ocular images using deep learning techniques [[7]].
Table 2 Best paper selection of articles for the IMIA Yearbook of Medical Informatics 2022 in the section ‘Public Health and Epidemiology Informatics’. The articles are listed in alphabetical order of the first author’s surname.
4 Conclusion
The 2022 edition of the PHEI section has confirmed the use of ML and deep learning approaches for data analysis to address public health issues.
A concern that has become more apparent in the wake of the recent epidemic and has yet to be addressed by the community is the ability to design and implement individualised public health approaches and tools.
