Data

This monocentric, observational study on strategic planning was conducted in the rheumatology department of the University Hospital in Lausanne (CHUV), Switzerland from 2017–2023. It included restrictions and subsequent new digital developments during the SARS-Cov2 pandemic. Data were used from a general retreat, working and focus groups as well as semi-structured interviews with different stakeholders and a literature and social media review. The methodological process is illustrated in Fig 1.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Methological view on the implementation of the Balanced Scorecard.

https://doi.org/10.1371/journal.pdig.0000385.g001

Retreat and working groups

An initital retreat with all institutional stakeholders on a cadre level (doctors, nurses, adminsitrative) was performed 2017 to define visions, missions and clinical priorities. A SOAR analysis (Strengths, Opportunities, Aspirations, and Results) has been performed [7]. Clinical needs and potential digital solutions were further elaborated in monthly working groups over six months. Subsequently, existing key processes were analyzed in four working groups (patient care, organisation, research and education).

Semi-structured interviews, focus groups and conferences on an institutional level

We performed interviews with all stakeholders, including the informatic department, administration, clinicians, nurses and researchers. The status quo of implemented digital solutions and ongoing projects in terms of digital transformation were investigated. Data access and interfaces on an institutional level were discussed. From a clinical side, we organised a yearly interdisciplinary conference between rheumatology, gastro-enterology and dermatology on immune-mediated diseases to understand the clinical needs and developments of each speciality (www.common-ground-meeting.org).

Technological review

Information on existing digital solutions for health care insititutions was obtained by literature review on Pubmed, Google, social media (mainly LinkedIn) and podcasts (e.g. Medical AI Podcast, DTx Podcast, Faces of Digital Health, Deep Minds). We created the « Digital Rheumatology Network » (www.digitalrheumatology.org) as international platform to educate digitalization in rheumatology along with a conference series called Digital Rheumatology Day was started 2019. This platform served inform and connect clinicians and researcher with digital companies and start-ups in the field.

Adapted Balanced Scorecard and Strategy Mapping

The Balanced Scorecard was adapted to measure or project performances of digital solutions according four perspectives: financial, customer, internal processes and learning and growth. Vision, mission and have been elaborated in the working groups. Internal and external customers have been defined as patients, clinicians, health care professionals and administration. Financial performance has been shifted to « values », which were defined as: patient empowerment, clinical-decision support, time-saving, cost-effectiveness and quality.

Variables and KPI

KPI have been identified and assessed for measurability and feasibility in the working groups based on a clinical, internal, learning, quality of care or innovation perspective. According the clinical focus of this work KPIs shown in Fig 1 concern efficiency and effectiveness, qualiy of care, time management, scientific development of health care professionals, patient-centeredness, technology and information systems and interdisciplinary communication.

Strategic mapping

The primary result of this work is a strategic map based on an adapted Balanced Scorecard according key procedures for digitalization (Fig 2). Previously, missions, visions and values facing new medical trends have been elaborated in a retreate and subsequent working groups. Internal processes and knowledge in terms of digitalisation have been worked out in semi-structured interviews, literature and social media search. Each chosen subset of the BSC was aligned with current developments in healthcare and digital technology.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. A strategic map based on the Balanced Scorecard for the implementation of digital health in a rheumatology department in an academic center.

From bottom-up the figure shows people & organisation, internal processes, customers, values and missions. Values have been adapted to current healthcare trends. Exemplary indicators are listed on the right. Artificial intelligence based models are indicated in orange. DTX: Digital therapeutics. PRO: Patient reported outcomes. EMR: Electronic medical record.

https://doi.org/10.1371/journal.pdig.0000385.g002

Mission and vision

We defined the mission of our academic department as cost-effective and state-of-the-art patient care, combined with a high level education and research. The vision was to overcome the challenges that come along with demographic changes of our society and staff shortage by becoming a smart hospital. We envision a countinously fluctuating healthcare system, e.g. due to the occurrence of infection waves, accidents or natural phenomena (heat or cold waves etc.). We aim to foster digital transformation but also to celebrate human interaction and contact with patients that are in control of their decisions and data.

Values

Apart from a financial aspect, we defined four health care developments as the key values in our department:

1. 1. Patient empowerment and convenience

Healthcare is developing towards a consumer-driven service. Patients want to actively chose their site of patient care and doctors. Less wating times, accessiblity via email ect. appointments online ect. This also applies at least for a portion of the patients that prefer a more active role in managing their chronic disease by monitoring their symptoms, providing patient reported outcomes (PROs) or even taking over clinical decisions after appropriate training [8]. Patient convenience from the medical and digital side (not in terms of hospitality) has so far remained largely unexamined in hospitals, although it has a significant impact on the quality of life as a measurable indicator. Of note, convencience can be more important to patients than quality of care [9].

1. 2. Personalized medicine

Personalized medicine approaches aim to streamline individualized diagnostic and therapeutic procedures for high efficacy and safety of care. This means a higher diagnostic effort (e.g. by -omics) and the integration of digital biomarkers and real world data in our workflow [10]. The exploitation of biomedical and clinical data by machine learning algorithms allows the prediction of individual disease courses, treatment response or phenotyping (clustering) as clinical decision support systems [11]. Thus, higher costs in diagnostics and data analysis might be compensated by better efficiency, lower complications and secondary healthcare costs.

1. 3. Interdisciplinary care

In many chronic diseases, notably immune-mediated diseases or cancer, symptoms of different body systems often are compulsorily connected e.g. skin and joint inflammation in psoriasis arthritis, immune dysregulation due to immunotherapy in cancer or chronic pain and depression [12]. Quality of care depends on an optimal exchange of data and clinical decision between medical and paramedical information which bears a substantial number of barriers in terms of data interoperability. The individual ’point of care’ which is not always clear to the patient (private practice, hospital, pharmacy, online), needs to be redefined. Digital platforms or EMR with interdisciplinary dashboards may support an efficient and time saving exchange [13].

1. 4. Sustainability and automatisation

The shortage of healthcare professionals, notably nurses, doctors or administrative staff is an emerging problem which leads to employee dissatisfaction and sick leave. Automatisation of simple procedures such as voice, image or text recognition has largely been implemented e.g. for medical documentation. There are several FDA-approved algorithms for automated evaluation of radiographs e.g. for fracture detection on the market. So far, those solutions are not an active clinical decision support but they likely increase quality as ‘double check’ or if no radiologist is available. Finally, advanced technology in chatbots such as ChatGPT from OpenAI could potentially support administrative tasks such as writing and correcting medical reports. Care robots are slowly touching ground in hospitals and care homes in order for transport but also for vital signs and PRO assessment [14].

Customers

Customers of our structure are patients (whether face-to-face or remote), clinicians, healthcare professionals, administrative staff and scientists. Accordingly, the needs of those internal and external customers on the BSC are anchored as clinical decision support, patient support, administrative support and research infrastructure. New types of customers are entering healthcare systems such as remote healthcare professionals (e.g. telenurse or online coaches), DIGA (Digital health application) providers ect. [15].

Internal processes (Digital Care Pathways)

Internal processes in hospitals are responsible to maintain and improve quality, efficiency and safety. Fig 2 (left) shows various forms of digital data collection (telemonitoring, digital biomarkers, biosensors ect.) as the basis for a personalized medicine. AI-models likely integrated in the EMR permit to define phenotypes, disease predictions and to perform transfer learning from concomitant (autoimmune) diseases. This supports the creation of digital care pathways that organise individual patient monitoring and treatment and orchestrate ressouces optimally. Data accessibility and connectibility is a main process for horizontal digitalization, both in terms of providing user interfaces, data privacy and legal certainty. Usability and intepretability of data is considered as a key process to avoid data and knowledge silos. This concerns structured data at the moment, but can be enlarged unstructured data e.g. by large language models. Clearly, external registries or other data sources should be interoperable with hospital systems, especially the EMR.

Generally, the EMR can be considered as the main tool to integrate new digital solutions horizontally and vertically (Fig 3). For example. EMR can be extended vertically by integrating PROs, Patient reported experiences (PREs), digital biomarker or apps [16]. According to the ‘Internet of Things’ (IoT), the concept ‘Internet of Medical Devices’ (IoMD) should be promoted. Horizontally, dashboards increase usability of EMR between specialities and AI-generated predictions (including transfer learning algorithms) can be included in those interdisciplinary dashboards. New generation clinical dashboards also should show a holistic patient journey including predictions and outlier analysis etc. In the sense of horizontal integration along the value chain, the prediction should be connected to data on the availability of follow-up care facilities. The first step of AI-guided clinical pathway / decision support will likely be auto-care loops in stable disease courses. A maximal amount of collected data will be analyzed by clustering and prediction models, a digital treatment plan will be established and automatically monitored and reported to health care professionals including remote nurses. Smart agenda planning allows to identify and planify patients with a high flare risk for a conventional visit.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Horizontal and vertical extension of Electronical Medical Records (EMR).

Elements around the EMR have been taken from the Balanced Scorecard (internal procedure perspective) and regrouped for a better demonstration.

https://doi.org/10.1371/journal.pdig.0000385.g003

Finally, automatisation is a sine qua non to encounter the shortage of rheumatologists and health care professionals. Large language models (LLM) such as GPT4 can be used to screen EMR and hospital databases for data or for the generation of discharge letters [17]. Voice recognition systems have widely been introduced in the clinic. New generation chatbots are able to extract the necessary information to create discharge reports and thus save time for the clinician to spend more time with patients or in interdisciplinary discussion. It seems primordial that all staff are equipped with smartphones and apps for those dashboards. Communication between colleagues in- and outside the hospital is via Email, which seems not adequate for several reasons (high number of unreplied messages, spam, security ect). It is obvious to connect case-based communication to dashboards in order to save time and to get relevant Information in a few clicks. All those digital processes together with their KPIs illustrate digital care pathways.

People and organisation (learning organisation)

In order to implement (digital or other) innovations, the human factor remains indispensable. This is especially true in medicine, where human interaction and ethical concerns are or paramount importance. Therefore, change management is the basis for creating readiness and acceptance for new technologies. In chronic diseases, a holistic view on patient journeys (or digital clinical pathways) with patient-centered endpoints and indicators are of increasing importance. Both patients and healthcare professionals need to understand how to access and inteprete this data and how to guarantee data privacy. To this end, knowledge about the inhouse databases (data warehouse etc.) and external data sources (registries or apps) as well as basic knowledge on data science is required and should be educated as a first step. The national or federal strategy for education of digital health is currently being actively discussed, but is only interoperable to a limited extent due to the different health systems [18]. Thus, a regular exchange between institutional, pan-departmental data scientists and collaborators within clinical departments ‘closer to clinical action’, shoud be established. Clinical-informaticians within medical departments are key elements for horizontal digitalization. For exemple they can educate clinicians use EMR more efficiently, including processes such as billing or implementing new processes such as prescribing DIGAs (Digitale Gesundheitsanwendung, digital health application) [15].

Human clinical know-how is also required to build and to control machine learning algorithms. As for input variables for algorithms (e.g. for clinical predictions), it is obviously not just about the availability but also the quality of the clinical data (e.g. what, when and how was measured). Clinicians need to specify the relevance of the data. As an exemple in image recognition, clinically defined pre-processing increases the prediction performance. As an example focusing on the hip shape to predict radiographic progression of hip ostearthritis [19].

No-coding platforms for image recognition or natural language processing can support clinicians in creating on predictions based on the insitutional data. For natural language processing, chatbots such as the openAI chatbot can leverage the work of doctors and scientists by support of writing articles or create automated reports and save precious time. Those tools are typically explored in defined research projects or audits before integrating them in the clinical workflow. Clearly, data safety issues have to be addressed rigourously and with new technological tools, its strengths and weaknesses are yet to be fully understood–and there appears early indication that such chatbots can be prone to confabulating information. An example of a medical report generated by the chatGPT3 chatbot can be found in S1 Table.

Patient education in the use of wearables, apps, digital therapeutics or biomarkers is pivotal. However low adherence to digital tools such as DIGAs is a widely undadressed problem [20]. Beside gamification, avatars or better human companion (e.g. health coaches) are necesary to assure adherence and quality of care. For mental health care, which also affects a substantial part of patients with immune-mediated diseases in form of secondary fibromyalgia, the human companionship seems even more important. Therefore, the need for trained remote health care professionals, e.g. as health coaches, will increase strongly in the future. A new profession of online nurses is also emerging among nurses, who work for hospitals or insurance companies and are connected to patients via a wide range of digital tools [21].

On a national and internationa level, both societies such as EULAR and patient organisations foster education and exchange of digital solutions [22]. To support this endavour, we created the digital rheumatology network (www.digitalrheumatology.org) organising yearly conferences (Digital Rheumatology Days) and regular pod-and webcasts.

Indicators

The BSC has been developed around key indicators so they can and should be obtained at all levels. In other words, what is measurable within a digital care pathway should be measured. In our opinion, extra hours, sick leave and employer satisfaction are among the most urgent KPI, given the shortness of health care professionals and the wish for part-time positions. Patient empowerment and convenience measured by patient satifaction is also notably important to create and maintain trust as the basis of successful care.

For skills, education levels both of patients and clinicians can be measured by the number of attended courses, podcasts ect. Prescribed DIGAs and telemonitoring reports and patient adherence to those can be measured and discussed during face-to-face consultatons. PREs can be used to reflect the mix of human and digital therapeutic services offered by the department. Internal processes, including automatisation procedures, can be monitored by reduced extra-hours. Measuring ’connected care’ is more subtle but could be measurable by reduced length of hospital stay or readmission rates and patient satisfaction. Of note, many aspects of patient care such as empathy, time for listening can not be measured.

Implementation in our department and first results

In our and many other hospitals, the key aspects of digital transformation are initiated top-down from an institutional level. For example, the online assignment of patients for consultations or hospitalisation or telemonitoring via an app for post-operative patients that is supervised by a central nurse team. The access to the data warehouse for scientific purposes has been facilitated and a machine learning team for clinical predictions or biomedical (big) data analysis collaborates closely with medical departments on a project level. To leverage these and other opportunities, we initiated a bottom-up strategy according to the above mentioned factors. On a departimental level, we have introduced regularly scheduled meetings between clinicians and institutional IT specialists. A clinician-informatician consultant provides regular training in basic machine learning coding and institutional data access to our clinical and research team. Clinical data science training using data from registries and our datawarehouse has been added to the curriculum for new assistant doctors. A specialized nurse consultation for rheumatic patients has been implemented to instruct patients in the use of apps with PROs, here within our national registry SCQM (Swiss Cohort for Quality Management), wearable data and the use of digital therapeutics. As our current EMR is lacking clinical scores and indicators such as the DAS28 score, we included the mannequin for tender and swolen joints and were thus able to synchonize EMR with the registry. Patients can send pictures or videos of their hands into our system via apps. We have developed an algorithms that automatically measure the finger folds, joint diameter etc. on hand images to monitor joint swelling [23]. The range of motion can be estimated on self-recorded videos of patients by automatically measuring the angles of the different joints and comparing them with the previous values. In the future, hopefully such disease-specific biomarker will earlier detect of arthritis flares.

A telemonitoring & communication office is currently being set up where medical medical assistants are trained to become ‘clinical workflow & communication manager’ with access to PROs, wearables, photos, videos etc. on specific dashboards and obviously the agenda of our consultation. In parallel to telephone calls and emails from patients they forward information (or not) to specialized nurses or our rheumatologist and to refer patients to the consultation.

Five years ago we started an international conference series called the ’Digital Rheumatology Days’ (www.digitalrheumatology.org) for healthcare professionals to educate them on telemonitoring, digital therapeutics/DIGAs, digital biomarkers, social media usage, machine learning ect. The annual conference was complemented by the Digital Rheumatology Network as an educational platform where information, podcasts, webcasts, etc. are permanently published and distributed via social media. To foster interdisciplinarity as part of the connected care model, the Common Ground Meeting for Immune-mediated Diseases has been created (common-ground-meeting.org), assembling different specialties for disease updates, the use of data (e.g. presentation of digital biomarkers) and case discussions by rheumatologists, dermatologists, gastro-enterologists, immunologists, nephrologists and pneumologists.

So far, the internal processes mentioned above have been partially implemented. Digital tools and data science have mainly been introduced on a project basis, with ongoing studies (S1 Table).