1 Introduction
As editors of the Clinical Information Systems (CIS) section, we annually apply a
systematic approach to gather articles for the International Medical Informatics Association
(IMIA) Yearbook of Medical Informatics. Over the past eight years, we have consistently
employed a specific query to identify relevant publications in the CIS field, resulting
in over 2,400 papers each year. These publications undergo a rigorous selection process
to identify the best CIS papers.
During this period, we have observed a notable shift in CIS from primarily focusing
on clinical documentation to a greater emphasis on generating patient-focused knowledge
and facilitating informed decision-making. CIS have evolved beyond being mere tools
or infrastructure for healthcare professionals and hospitals, becoming the foundation
for a complex trans-institutional information logistics process. The patient has taken
center stage, and patient data is utilized to create value for their benefit. Consequently,
research in the CIS domain has increasingly focused on trans-institutional information
exchange, data aggregation, and analysis [[1]
[2]
[3]
[4]
[5]].
The COVID-19 pandemic has significantly influenced scientific research in the CIS
field, as evident from its extensive impact [[6]]. However, in the past year, we also observed no notable breakthroughs or innovative
changes regarding methodologies, algorithms, tools, or applications for diagnostic
or therapeutic purposes. This prompted us to question whether our long-standing query
had become outdated [[7]].
With the change in our editorial team and the new section editor BP, we took the opportunity
to update our query. Recognizing the significance of trans-institutional data exchange
and patient-centeredness, we noticed that “telemedicine” was not explicitly included
or adequately addressed in our existing systematic query. To rectify this, we conducted
a thorough investigation by searching PubMed using the MeSH Major Topic “telemedicine,”
resulting in nearly 30,000 publications. These publications were subjected to bibliometric
analysis [[8]]. Based on our findings, we decided to include terms related to telemedicine, such
as “Telemedicine [MeSH Major Topic]”, “Telemedicine majr/OT”, “Telehealth [OT]” and
“Telemonitoring [OT]” in our PubMed query. Conversely, we removed the term complexes
surrounding “geographic information systems”, as they were increasingly nonspecific
in relation to CIS.
By incorporating these new search terms, we anticipate a significant increase in the
number of identified papers, subsequently increasing our workload. To manage this
challenge, we have appointed a third section editor (SBN) to ensure efficient handling
of the expanded number of papers during the systematic selection process for the best
CIS papers.
Each year, the IMIA Yearbook editorial board defines a special topic to highlight
current aspects relevant to medical informatics. Each section focuses on these aspects
when reviewing the past year's literature. For the 2023 edition, the special topic
is “Informatics for One Health”. Consequently, we were eager to assess whether this
topic is reflected in the papers found during our selection process.
About the Paper Selection
The process of searching for relevant publications and selecting the best papers in
the CIS section follows a well-defined systematic approach. After using the same query
for eight years, we decided this year, following an extensive analysis [[8]], to incorporate the topic of telemedicine in our queries and remove search terms
related to geographic information systems. In mid-January 2023, we conducted the queries.
We retrieved 5,206 unique papers, with 5,020 from PubMed (1,543 from the legacy query
and 3,477 from the telemedicine-related search terms) and an additional 186 papers
from Web of Science® (WoS). These articles were published in 1,500 journals, and [Table 1] presents the Top-15-ranked journals with the highest number of resulting articles.
Table 1 Number of retrieved articles for Top-15 ranked journals.
Despite changing the query, we noticed that the relative frequencies of publications
from different countries remained similar to previous years. Most of those papers
whose publication records included location information came again from the United
States (45%, n=2,327). England was again second (26%, n=1,351), followed by Switzerland
(6%, n=329), Canada (4%, n=222), Germany (3%, n=179), the Netherlands (3%, n=176),
Ireland (2%, n=111) and Australia (2%, n=96).
Following our annual practice, we proceeded to categorize all the papers we found
systematically through multiple rounds, resulting in a shortlist of up to 15 potential
contributions. External experts and yearbook editors then reviewed these selected
papers. Subsequently, the IMIA Yearbook Editorial Board conducted a selection meeting
to choose a maximum of four best papers for each section. To comprehensively understand
the articles' content in the CIS section, we employed text mining techniques and term
combination mapping.
We used RAYYAN[1], an online systematic review tool for the multi-stage selection process of the best
papers. The legacy query results from PubMed and WoS (n=1,729) were reviewed independently
by two section editors (WHO and BP), and the telemedicine-related search results (n=3,477)
were reviewed separately by all three section editors (WHO, BP, SBN).
During the first pass of screening, articles were excluded based on their titles and
abstracts. The agreement rate for “exclude” decisions between WHO and BP was 95.5
percent, resulting in 4,970 articles being excluded out of the total 5,206 articles
in consideration.
For the telemedicine-related search results, there was a 95.6 percent agreement rate
(n=3,324) between BP and SBN regarding article exclusions out of the total of 3,477
articles. Between WHO and SBN, the agreement rate for “exclude” decisions was 96.8
percent, leading to the exclusion of 3,368 articles out of the total of 3,477 telemedicine-related
articles.
These agreements reflect the level of consensus among the section editors regarding
the exclusion of articles based on title and abstract review. From this first selection
pass, 350 papers remained for the next screening rounds, in which consensus was jointly
reached to narrow down the selection further. The second selection pass yielded 136
possible candidates which were reduced to 52 in a third and 25 in a fourth pass. For
these potential candidates, the full texts were obtained and reviewed.
Finally, we selected 15 candidate papers for the CIS section on mutual consent. Six
candidate papers stemmed from the legacy query and nine were among the telemedicine-related
articles. Each candidate paper underwent a rigorous review process, with at least
seven independent reviews collected for each paper.
The selection meeting took place on May 5, 2023, in Bordeaux, France, and was conducted
in a hybrid format, allowing both online participation and in-person attendance. The
meeting involved the IMIA Yearbook editorial board, which decided on the final selection
of papers. After careful deliberation and discussion during the meeting, four papers
were ultimately chosen as the best papers for the CIS section. Content summaries of
these four best CIS papers can be found in the appendix of this synopsis.
2 Findings and Trends: Clinical Information Systems Research 2022
During the selection process, the increase in the number of publications in the CIS
field, particularly with the inclusion of telemedicine, posed a challenge in keeping
track of the content of all the relevant literature. To cope with this abundance of
publications, we employed additional methods such as text mining and bibliometric
network visualization for several years in our section [[9], [10]]. These techniques allow for the efficient extraction of relevant information from
a large corpus of articles and help us obtain a quick understanding of the content
of the articles, enable meaningful comparisons between them, and provide us with a
visual representation of the relationships between publications and their content.
The visualizations also assist in identifying clusters of terms and emerging trends
within the field.
2.1 Overview of the Content of All Found CIS Papers
First, we extracted the authors' keywords (n=38,184) from all articles and presented
their frequency in a tag cloud (cf. [Figure 1]). We found 5,393 different keywords, of which 3,200 were only used once and 728
were used twice. As in the previous years, the most frequent keyword was “Humans”
(n=3,914). This year “Telemedicine” was the second most frequent keyword (n=2,209),
followed by “Pandemics” (n=1,030), COVID-19 (n=705), “Female” (n=683), “Child” (n=571)
and COVID-19/epidemiology (n=546). “Electronic Health Records” is ranked 8th (n=546) versus 2nd last year.
Fig. 1 Tag cloud illustrating the frequency authors' keywords (top 250 keywords out of n=38,184
are shown) within the 5,206 papers from the CIS query result set. Font size corresponds
to frequency (the most frequent keyword was “humans” n=3,914).
In contrast to the keyword tag cloud, a bibliometric network can reveal more details
on more items and interrelationships between the publications. We used VOSviewer [[9]] to create a clustered co-occurrence map of the keywords depicted in [Figure 2].
Fig. 2 Clustered co-occurrence map of the Top-690 keywords (keywords with the greatest total
link strength n=690 of 5,393 distinct keywords) from the 5,206 papers in the 2023
CIS query result set. Only keywords that we found in at least six different papers
were included in the analysis. Node size corresponds to the frequency of the keywords
(“humans” n=3,914). Edges indicate co-occurrence (only the top 1,000 of 22,262 edges
are shown). The distance of nodes corresponds to the link strength between the keywords.
Colors represent the six different clusters. The network was created with VOSviewer
[[9]].
Cluster 1 (in red) indicates the clinical perspective. Both in application and in
research. The most common keywords in this cluster include “electronic health records”,
“technology”, “hospitals”, “(health) communication”, “medical records”, or “information
systems”. Terms such as “machine learning” or “artificial intelligence” suggest current
research activities in this area. Cluster 2 (in green) contains as most frequent keywords
terms like “humans”, “telemedicine”, “pandemics”, “covid-19”, “covid-19/epidemiology”,
“telemedicine/methods”, “sars-cov-2”, “cross-sectional studies”, “delivery of health
care”, “referral and consultation”, “patient satisfaction”, or “primary health care”.
The most common keywords in cluster 3 (in blue) are “mobile applications”, “quality
of life”, “pilot projects”, “feasibility studies”, “randomized controlled trials as
topic”, “telerehabilitation”, “chronic disease”, “exercise”, and “treatment outcome”.
Cluster 4 (in yellow) contains general contextual factors of the studies with keywords
such as “female”, “adult”, “male”, “aged”, “retrospective studies”, “middle aged”,
or “adolescent”. More specific context factors of the studies can be found in cluster
5 (in pink) with keywords such as “point-of-care systems”, “prospective studies”,
“reproducibility of results”, “emergency service”, “hospital”, “only child”, “emergencies”,
“infant”, “infant”, “newborn”, “child”, “preschool”, and “point-of-care testing”.
Finally, the keywords in cluster 6 (in turquoise) reflect specific application perspectives
in the country with the most frequent publications, the USA. The most frequent terms
in this cluster are: “united states”, “medicare”, “prescriptions”, “pharmacists”,
“analgesics”, “opioid/therapeutic use”, “drug prescriptions”, “practice patterns”,
“physicians'”, “opioid”, “prescription drug monitoring programs”, “electronic prescribing”,
“opioid-related disorders/drug therapy”, “medicaid”, “buprenorphine/therapeutic use”,
and “nonprescription drugs”.
To ensure consistency with previous years' investigations, a parallel analysis was
conducted using the terms extracted from all publications' titles and abstracts. The
resulting co-occurrence map of the top-690 terms is visualized in [Figure 3].
Fig. 3 Clustered co-occurrence map of the Top-690 terms (top 60 percent of the most relevant
terms: n=690 of 95,937) from the titles and abstracts of the 5,206 papers in the 2023
CIS query result set. Only terms that we found in at least 25 different papers were
included in the analysis. Node size corresponds to the frequency of the terms (binary
count, once per paper, “covid”: n=1,627). Edges indicate co-occurrence (only the top
1,000 of 131,853 edges are shown). The distance of nodes corresponds to the association
strength of the terms within the texts. Colors represent the four different clusters.
The network was created with VOSviewer [[9]].
Compared to the analysis of the keywords, a different picture emerges here. The cluster
analysis of titles and summaries resulted in four distinct, similarly large clusters.
Cluster 1 (in red) is heavily populated with terms related to the COVID pandemic.
But also, terms from the field of telemedicine, its application, and research are
strongly represented. For example: “video”, “appointment”, “encounter”, “patient satisfaction”,
“telehealth service”, “telephone”, “telemedicine visit”, “telehealth visit”, “telemedicine
use”, “teleconsultation”, “care delivery”, “phone”, “telemedicine service”, “telehealth
use”, “healthcare delivery”, “video visit”, “virtual visit”, “virtual care”, “video
consultation”, or “patient portal”. Cluster 2 (in green) and cluster 3 (in blue) again
contain various contextual factors, objectives, and methodological aspects of the
studies. The remaining cluster 4 (in yellow) also includes such terms. Here, however,
we also find numerous terms that can be attributed to the outcomes of the studies.
The impact of the COVID-19 pandemic continues to be profoundly felt in the field.
Modifying the query with an increased emphasis on telemedicine has notably influenced
the results. Nevertheless, the trends and research focus observed in recent years
are still evident in the generated maps. The selection process for the best papers
has yielded a substantial number of captivating and high-quality contributions. In
the following sections, we will briefly introduce the candidates and highlight the
best CIS papers chosen for recognition.
2.2 Insights into the Candidate Papers and Best Papers
Secondary use of clinical data is still an essential and prominent research field
within CIS. Although the term “data science” is not yet often explicitly mentioned
in the keywords or the titles or abstracts of the found CIS publications, we found
many examples and inspiring, promising data science applications. One of them is one
of the best papers, a contribution by Guardiolle et al. [[11]], which is about linking biomedical data warehouse records with the national mortality
database in France using a large-scale matching algorithm. The algorithm was found
to be reliable for both sensitivity and specificity evaluation, and it was able to
link a large number of records with a high degree of accuracy. The paper also discusses
the potential benefits of linking these two data sources for medical research and
public health decision-making.
Another contribution from the field of data science, specifically about using machine
learning and natural language processing techniques to analyze electronic health record
(EHR) data is the best paper by Zou et al. [[12]] who use an end-to-end knowledge-graph-informed topic model. They discuss the challenges
of extracting clinical knowledge from EHR data and propose a new method called the
Graph ATtention-Embedded Topic Model (GAT-ETM) to address these challenges. The paper
also compares GAT-ETM to other methods in terms of topic quality, drug imputation,
and disease diagnosis prediction, and explores how GAT-ETM can be used to discover
interpretable and accurate patient representations for patient stratification and
drug recommendations.
Information exchange between different stakeholders or institutions has become a core
CIS topic in recent years. In the past years, we always had papers from the FHIR (Fast
Healthcare Interoperability Resources) context in our selection. This year, a contribution
by Bialke et al. [[13]] on the topic of standardized exchange of informed consent in an extensive network
of university medicine made it into the best paper selection. The authors discuss
the use of FHIR in facilitating the exchange of informed consent and provide insights
into the potential implications of using FHIR for informed consent in the field of
university medicine. Another interesting contribution to the FHIR context comes from
Ayan Chatterjee et al. [[14]]. The paper is about achieving semantic and structural interoperability in personal
health data through the use of HL7 FHIR with SNOMED-CT. It presents a proof-of-concept
study that explores innovative solutions to the problem of heterogeneity in digital
health information systems. The study focuses on designing and implementing a structurally
and logically compatible tethered personal health record (PHR) that allows bidirectional
communication with an electronic health record (EHR).
For a successful, seamless exchange of information, not only technical aspects are
relevant. The publication by Pylypchuk et al. [[15]] provides valuable insights into the significance of EHR developers in facilitating
hospital patient sharing, which involves the seamless transfer of patients between
different healthcare facilities.
In another candidate paper worth reading, Kryszyn et al. [[16]] explore the question of how the performance of an openEHR-based hospital information
system can be compared with a proprietary system. They evaluate the benefits and drawbacks
of using the openEHR standard and suggest that the benefits of using openEHR may outweigh
the found performance issues, especially for more complex hospital information systems.
Based on our observations, there is a growing convergence between clinical information
systems and telemedicine. This overlap is leading to not only the exchange of data
between different healthcare facilities but also the development of increasingly beneficial
applications that leverage this data for the well-being of patients. Particularly,
the term “mHealth,” which refers to mobile health technologies, is gaining prominence
within this context. As a subset of eHealth, mHealth focuses on utilizing mobile technologies
to improve healthcare delivery and patient outcomes. Integrating mHealth solutions
with CIS and telemedicine is driving innovations and advancements in patient-centric
healthcare. In their candidate paper, Alenoghena et al. [[17]] provide a comprehensive review of trends and advancements in three aspects of an
eHealth system and service delivery: contemporary architectures for eHealth designs,
mHealth technologies, and security concerns. It is recommended to all who want to
refresh their knowledge in these areas since reading this article.
Also an interesting read is the candidate paper by Li and You [[18]] who propose an intelligent mobile health monitoring system and establish a corresponding
health network to track and process patients' physical activity and other health-related
factors in real-time. The authors state that this system can help patients monitor
their personal health in real time and can help healthcare providers identify potential
health issues early on and intervene before they become more serious. This can lead
to improved patient outcomes and better overall health management.
The focus on patient-centric outcomes is crucial in driving meaningful advancements
in healthcare informatics. To emphasize the practical impact of research efforts,
the selection process included concrete examples of applications that directly benefit
patients. A truly impressive example of this is the fourth paper in the best papers
roundup. Poelzl et al. [[19]] demonstrate the feasibility and effectiveness of a multidimensional post-discharge
disease management program for heart failure patients in clinical practice. The study
evaluated the benefits of a telemedical monitoring system incorporated into a comprehensive
network of heart failure nurses, resident physicians, and referral centers. The study
found that the multidimensional post-discharge disease management program was feasible
and effective in clinical practice, resulting in a significant reduction in the primary
endpoint of death from any cause and readmission for acute heart failure at six months,
as well as improvements in patient empowerment.
Graetz et al. [[20]] could demonstrate that video telehealth improves access to healthcare for people
with diabetes by offering them a new, convenient way to access healthcare without
arranging transportation or taking time off work. Video visit access was associated
with a statistically significant reduction in HbA1c levels among people with diabetes.
This is particularly important for people with chronic conditions, who require ongoing
monitoring and adjustment by patients and their clinicians. Video telehealth gives
people real-time access to clinicians, which can help them manage their condition
more effectively. Moreover, this approach contributes to reducing a patient's environmental
impact by reducing the need for travel. Consequently, this article is relevant to
the Special Topic “One Health” as it highlights the potential benefits of telehealth
in improving patient outcomes and reducing environmental impacts, as shown in [[21]]. In addition to reducing the time and travel burden associated with participation,
remote technology, and decentralization tools can also help increase patient enrollment
in cancer clinical trials as suggested by the findings of Adams et al. [[22]].
Also for hospitals, it may be beneficial to have telehealth services. According to
the candidate paper of Zhao et al. [[23]], hospitals with one or two telehealth services were found to have higher total
performance scores compared to hospitals with no telehealth services. However, the
study did not specify which specific telehealth services were associated with improved
hospital performance.
In any case, the future will bring more mHealth applications, and it will be important
to prepare both patients and hospital operators as well as possible. In their article,
Hamberger et al. [[24]] explore central questions here, like What are some specific challenges that mHealth
apps can help address in the healthcare system? How can patient-centered approaches
be implemented through mHealth apps? What are some potential benefits and drawbacks
of integrating mHealth apps into the healthcare system?
In our opinion, artificial intelligence (AI) can play a significant role in the context
of mHealth apps and the healthcare system. AI can assist in decision-making, data
analysis, and personalized treatment plans. However, the integration of AI into digital
healthcare solutions also poses ethical challenges, such as transparency, accountability,
and bias, such as biased data sources, which are often based primarily on male subjects
and usually do not include minorities, possibly limiting the validity of the data
and derived results. However, before we drift too deeply into such a discussion –
there will be enough to discuss in detail in the coming years – let's take a look
at the last two candidate papers. They exemplify what is already possible with artificial
intelligence now. First, an interesting contribution by Humayun et al. [[25]] about an agent-based medical health monitoring system. Such a system is a group
of intelligent agents that gather patient data, reason together, and propose actions
to patients and medical professionals in a mobile context. The proposed system combines
data mining techniques with a wireless medical sensor module to gather real-time sensory
data from the patient's body and historical data obtained in the past. The system
then categorizes the data into normal and emergency categories and declares an emergency
by comparing the previously described data groups. Of course, there are numerous challenges
here. If you are interested in this, you should read the paper. And last but not least,
we have a paper from Hah and Goldin [[26]] in our selection. It is on AI-assisted decision-making in healthcare. In this article,
the authors explore how clinicians currently use multimedia patient information (MPI)
provided by AI algorithms and identify areas where AI can support clinicians in diagnostic
decision-making.
This year's survey article of the CIS Section fits in perfectly with this. Farah Magrabi,
David Lyell, and Enrico Coiera from the Centre for Health Informatics, Australian
Institute of Health Innovation, Macquarie University, Sydney, present a very interesting
overview of automation in contemporary clinical information systems [[27]]. Their survey explores the use of AI technologies in healthcare settings, including
their clinical application areas, level of system autonomy, and reported effects on
user experience, decision-making, care delivery, and outcomes. We would like to warmly
recommend it to our readers, as we do every year at the end of our review of the results
and trends of the Clinical Information Systems Section.
3 Conclusions and Outlook
The inclusion of telemedicine as a search term in our query turned out to be positive.
Of course, this led to a substantial increase in the number of papers found. But we
could compensate for that by adding a third section editor. On the other hand, we
also found a large number of very substantial papers, which was also reflected in
our selection. Nine of the 15 candidate papers came from this new query part.
The analysis highlights the growing convergence between clinical information systems
and telemedicine, with mHealth technologies gaining prominence as a subset of eHealth.
Data science applications, particularly in the secondary use of clinical data, are
increasingly making a mark in CIS research. Additionally, the integration of artificial
intelligence (AI) in healthcare informatics, mainly through mHealth apps, shows promising
potential for decision-making, data analysis, and personalized treatment plans.
The selected candidate papers underscore the practical impact of research efforts,
focusing on patient-centric outcomes and benefits. They cover a range of topics, from
intelligent mobile health monitoring systems to AI-assisted decision-making in healthcare,
all contributing to the improvement of patient care and outcomes.
As we move forward, it is evident that the field of CIS will continue to evolve, driven
by advances in telemedicine, mHealth technologies, data science applications, and
the integration of AI. This ongoing convergence between various disciplines will pave
the way for transformative innovations in patient-centric healthcare. It will be crucial
to address ethical challenges surrounding AI, ensure transparency, accountability,
and eliminate biases to harness its full potential in improving healthcare delivery.
We think that the ongoing efforts in CIS research will undoubtedly lead to the development
of more efficient, patient-oriented, and intelligent healthcare systems, contributing
to the overall improvement of global healthcare outcomes. The next few years will
show whether we are correct in this assumption.
Table 2 Best paper selection of articles for the IMIA Yearbook of Medical Informatics 2023
in the “Clinical Information Systems” section. The articles are listed in alphabetical
order of the first author's surname.