Applying Soft System Methodology for a clearer understanding of the future Intensive Care Units

Intensive Care Units (ICUs) generate a large amount of valuable data related to the health status of patients. In addition, ICUs can leverage other sources of big data, such as structured data, text, video, and images. In this work, a framework for the future ICU system is proposed, which is based on the Soft System methodology (SSM) and the use of big data technology. The framework and the related activity models ensure that the ICU can have its particularities and specialties, as well as its core services and functions. The application of the framework also implies that ICU can provide ongoing expertise and training to upgrade its staff, can improve interoperability with the National Health System, and ICU staff intercommunication and remote services.


INTRODUCTION
The experience of recent years with the pandemic brought to the surface the complex problem of an insufficient number of ICUs.It is common knowledge that we face an ever-increasing need to rapidly increase the number of units (meaning wards, beds, equipment, staff).The need to increase the staff working in ICUs and their training and expertise is also important and urgent [17].The problem is exacerbated because the application and utilization of new technology in the ICU, such as big data, data analysis, and IoT, while providing great potential for quality and cost-effective operation is not at the desired level today.Furthermore, according to the literature, there is a lag in the development of a methodology for the application of new technology in solving the problem of assisting the operation of the ICU [1,2,8].
In this paper, we propose the use of Soft System Methodology (SSM) for a clearer understanding of the future ICU which should be based on Big Data and Big Data Analytics (BDA).SSM is also helpful for the creation of a framework for the operation of a modern ICU.ICUs have their particularities such as specialization in specific needs of patients including heart attacks and respiratory problems.An ICU framework would allow the continuous support of specialization and training to upgrade and increase the number of specialized people (doctors, nursing staff), interoperability between ICUs and National Health System, improvement of ICU staff intercommunication, and remote services and in patients transported by ambulances.
Section 2 describes the new possibilities of the multi-disciplinary team running the ICU which are offered by the state-of-the-art technology, e.g., communication possibilities and use of big data.Section 3 presents our research method and explores the relationship between big data and the future ICU.Section 4 explores how the SSM is applied to the Health Care sector.Section 5 discusses why SSM is a methodology suitable for understanding/designing the future ICU.Section 6 proposes a rich picture and activity models of the ICU operation.Eventually, section 7 concludes the research and outlines future activities.

NEW POSSIBILITIES FOR THE MULTI-DISCIPLINARY TEAM RUNNING THE ICU
The operation of the ICU is the work of a multi-disciplinary team that includes nursing staff, specialized medical staff and doctors of other specialties, technical staff, pharmacists, data analysts, etc.The technological framework of operation of the smart ICU intersects with the context of e-government, smart city, and smart health.It is based on cutting-edge and emerging technologies, e.g., interoperability, cloud computing, and big data.These new technologies improve and support the participation of personnel in the work of the ICU in the following directions: 1) Remote communication with the ICU, e.g., specialized staff who are not present on the unit during a patient crisis can provide services remotely, communication of ICU physicians with a patient transported by ambulance for emergency readmission, etc.
2) Optimal in-hospital communication of the ward, nursing staff, and specialist doctors based on state-of-the-art technologies.
3) Communication of ICUs in a wider coordination context and efficient functioning of Communities of Practice at national and international levels for continuous training, access to valuable structured data (e.g., MIMIC III/MIMIC IV database), utilization of the literature, good practices, and case-based reasoning.
4) Big data management using AI/ML techniques that provide many new possibilities for providing intensive care services.
5) Engagement of Visitors to the ICU, while in the past they were more or less excluded from their people, they are now called upon to play, after training, some role in boosting the morale of the patients.However, this issue remains controversial to this day.2017) state that the healthcare sector is one of the promising areas for the application of big data solutions.However, they point out that there is a lack of specific methodology for the implementation of Big Data architectures and projects.

RESEARCH METHOD FOR APPLYING BIG DATA IN ICU
Reiz, de la Hoz and Garcia, [1] review and evaluate different research techniques of big data analysis and machine learning and explore how these techniques could be used in the ICU field.They also suggest that clinicians work with a "data scientist" specialized in health data analysis to help develop big data analysis and machine learning techniques.
Markopoulos et al. (2020) describe a Conceptual Framework and propose its use in designing architectures and big data applications in ICUs.The Conceptual Framework is based on Big Data Analytics (BDA), Machine Learning (ML), Natural Language Processing (NLP) and consists of the following subsystems: The "Big Data Integration and ICUs" module, the "ICUs and critical care services" module, the "Use of standards and ICUs" module, the "Machine Learning and ICUs" module, and the "NLP and ICUs" module.
The application of the seven (7) stages of the Soft System methodology (SSM) is proposed in order to define conceptual frameworks (models) for complex information systems [5].
Based on the SSM, a conceptual framework for the operation and management of ICUs could be presented that aims to meet the needs but also to reduce the difficulties faced by medical and nursing staff and patients.Emphasis could be placed on exploiting new technology in the use of such systems, improving data integration, information retrieval methods, enriching original data sources and reducing the workload of doctors and nursing staff.The methodologies described in [1][2][3][4]6] focus on applying new technology in ICU, e.g., big data analytics, and on covering the needs for remote communication, and in-hospital communication.They, also, offer a preliminary description of data flow in ICU.Our method, which is based on SSM [5], covers the same topics and also presents the stakeholders' view of the situation, describes in more depth data flow in ICU, examines co-operation with other ICUs, and offers detailed activity models for covering needs in information and training.Therefore, a deeper and clearer understanding of the future ICU is ensured by the proposed method.

SSM AND HEALTH CARE
SSM methodology [5] uses a never ended "learning cycle", which is based on the multiple activity models that correspond to the stakeholders' perspectives during implementation.A discussion/debate is organized "which will surface multiple worldviews and generate ideas for change and improvement" of the problematic situation.This process seeks "common ground" for different perceptions and finds, implements and evaluates "versions of the to-be-changed situation".To create any specific purposeful activity model root definitions, CATWOE techniques, etc. could be used.Hanafizadeh and Mehrabioun (2018) believe that the iterative process of SSM, which includes observation and rich diagram creation of the problem situation and of ideal models, can provide an ideal vehicle for redescribing an issue, and "at arriving at the questions to be asked".But it is also necessary to focus the discussion and incorporate results of the literature from the field of performance measurement and management [11,15].
The use and contribution of SSM to health care was studied by Augustsson, Churruca, & Braithwaite.(2020).According to their review of important case studies the SSM was used in a variety of problem situations and was found to be of great help in understanding them and in addition was used to suggest possible improvements.To a lesser extent the methodology contributed to the implementation and evaluation of these improvements.Kotiadis (2006) presented a simulation study of a complex integrated health care system for older people in which SSM is used: to understand the problem and determine the conceptual model, to explore the Decision Making process involved in IC services, to explore the services utilization, and also explores if there are service gaps.SSM is useful in the conceptual modelling phases, particularly in determining the overall objectives, and there are benefits to applying SSM to determine the simulation study objectives [13,14].
The application of SSM in new fields, such as sustainable development, knowledge management and project management, is also described in the literature.In addition, it is proposed to apply mainly the processes of inquiry used by the methodology or a hybrid use of SSM that includes the process of inquiry and elements of the methodology that are action-oriented [11].
Armstorng (2019) presents a "critical realist approach to SSM" which incorporates existing knowledge into the treatment of problematic situations.
Sharma et al. (2020) after finding that SSM has been criticized for its lack of "rigour" believe that this is a fallacy due to a lack of understanding of its purpose.The purpose of SSM is exploration that will yield a clearer understanding of complex, socio-technical systems.In contrast, the purpose of quantitative modeling is confirmation, not understanding.The purpose of their paper is to describe the application of SSM to address the problematic situation of low participation rates in the promising practice of medicine known as Precision Health-Care (PHC).Their empirical study proves that applying the methodology ensures that participants, such as patients and their doctors, are more willing to share their medical data and diagnosis for the purpose of precision health modeling.

WHY SSM IS A METHODOLOGY SUITABLE FOR DESIGNING THE FUTURE ICU
The problem of designing the future ICU is a typical problem of designing and implementing a new system from those solved with SSM.ICUs encapsulate complex terminology and it combines heterogeneous worldviews, depending on the actors, customers, owners of the system.There are different "customers" (e.g., patients, hospitals, National Health System, manufacturers of medical devices and software, other ICUs, collaborating doctors of various specialties), different "actors" (doctors, nursing staff, technologists, information professionals, data analysts, public health policy makers), and there is need for "service transformation" to take advantage of new technologies.
Here are topics related to "Worldviews" of the various actors, and the customers, e.g., patients, hospitals, other ICUs, National Health system (see also the rich picture in figure 1): 1. Provision of core functions, e.g., traditional intensive care services, and clinical decision support 2. Criteria for clinical support, e.g., mortality prediction, probabilities/possibilities of re-admission 3. Need for a dramatic increase in the number of units but also in the number of qualified doctors, nurses 4. Need for continuous training and seminars 5. Expansion of data available for use and exploitation.We can use structured data, free text, discharge letters, doctors' notescomments, bedside monitoring data, wearables, financial data, etc.The MIMIC III database is an excellent example of data collection and is used in a wide range of scientific research (Markopoulos, Tsolakidis, & Skourlas, 2021) as shown in figures 1, 2 and 3 6.The future ICU is connected to big data and new data flow and processing.
7. Provision of new functions, e.g., Remote access, new possibilities for inter-communication, communication with ambulance for intensive care of transported patients, participation in cooperative schemes, communication with the National Health system, 8. Strategic decision support.Criterion of cost benefit.9. Data analysis and ML, e.g., use of Neural Networks, Hadoop [4] The application of SSM is directly linked to rich pictures of the problematic situation.The advantage of this tool is that it simultaneously depicts many-all worldviews and facilitates the exchange of views.
It should be noted that doubt is expressed in the literature as to whether physicians would want to participate in meetings that use rich pictures.Despite the relative skepticism, rich pictures are a popular tool and lead after discussions to root definitions that make sense for actors (stakeholders).The root definitions are a starting point for a briefly defined objective and activity model and finally in a conceptual model that includes the result of the process and covers a series of activity models.
The completion of the application of the methodology could be finalized with the following stages: 1) Comparison of models with "reality", 2) Defining desired changes.Examination of feasibility, 3) Actions.

RICH PICTURES AND ACTIVITY MODELS OF ICU
In figure 1 a Rich picture of the operation of the new smart ICU is given which illustrates the Worldviews of various actors (stakeholders).This picture should be used for the discussion of ICUs operation and the various Worldviews.Figure 1 also forms a basis for a conceptual map (framework) which can be used for the software applications which will support the core services and the secondary services offered by the ICU.Specialized doctors ("specialists") are mainly interested for clinical support systems, and the possibility of remote patient management.The doctors who specialize ("trainees") are interested for continuous assistance in their daily work (in order to gain experience) and the quality of their training.Nursing staff are mainly interested for the better organization of intercommunication in the ICU and for advanced patient monitoring systems.Doctors of other specialties are interested in communication with doctors and nursing staff of the ICU from a distance and/or a special meeting place in the ICU.The experts in new technology (IT professionals) are interested in upgrading the services provided using new technology (e.g., big data, big data analytics) and they also focus on specialized issues of security, personal data, overall quality of the system, etc.At another level, hospital management, heads of other ICUs and National Health System executives have their own interests and perceptions of the 'reality' of ICU operation.The main problem they face is how to increase the number of equipped ICU beds and wards and how to reduce the ratio between patients and qualified doctors and nursing staff.As a major implication, a Decision Support System is proposed that uses big data from different sources both inside and outside a specific ICU.Access to Electronic Health Record from the primary health and prescription records, laboratory test results, treatment details and demographic data is required during patient admission in ICU.In addition, data from wearables and data from sensors and other devices must be imported.Reference data is also entered for the inpatient setting, i.e., the other clinics or the emergency.These include information about whether the patient has a complete health record, whether it is a readmission and other critical information.In figure 2 we present an activity model in which the data flow, the clinical decision support in the ICU and the strategic decision support at the ICU/hospital level/national health system are illustrated.We see integrated (big) data in two directions: • live data from monitors next to the patient (bedside monitoring), diagnostic data using ICD9/ICD10 coding, data from operations, data on the scale of severity of the incident, complications, infections, etc. • Demographics, ICU admissions data, referral data (attending physician, clinic, etc.), final outcomes.
In figure 3 a detailed activity model for the online processing of the various types of data which are useful for the operation of the new ICU is presented.Emphasis is given to the clinical decision support in the ICU.The ICU system can also use mathematical models, prognostic models and pharmacological models and can use data from external databases, e.g.MIMIC III or MIMIC IV (Medical Information Mart for Intensive Care), which include data curated by experts.Figure 3 also describes the personalized immediate prognosis of the patient's condition, immediate prognosis of treatment response, use of pharmacological models and treatment regimens, comparison of actual treatment response vs predicted treatment response, continuous monitoring using video, image, text, etc.The processing of large-scale data should lead to the calculation of readmission possibilities in the ICU and to mortality prediction and to decision support which feeds back the personalized prognosis of the patient's condition.In figure 4 we present a detailed activity model for upgrading the skills of various ICU stakeholders/actors also covering their information needs (see also [5], pp 72-73).This model illustrates the necessary access control component of the ICU information system that should be ensuring that stakeholdersactors get access to the information and training they really need.Monitoring and the feedback loop is depicted.The never-ending learning cycle of SSM, which is the main philosophy of SSM, is also illustrated.The related root definition follows.Root definition: An ICU-owned and staffed system which develops and trains ICU personnel providing information, skills, and experience in a cost-effective manner.The CATWOE approach could be used to enrich the root definition: C -ICU personnel A -ICU personnel T -Identification of users' role and users' needs W -The mutual working can be an effective and efficient basis for ICU processes (see [5], p. 93-95) O -ICU E -Various conditions: timely, comprehensive, etc.

CONCLUSIONS AND FUTURE ACTIVITIES
In this paper a brief description of applying the SSM for a clearer understanding of the future Intensive Care is given.The never ended "learning cycle" of the Methodology is related to various activity models that correspond to the stakeholders' perspectives.The whole approach includes the organization of a "debate" for understanding the different "worldviews", and seeking the "common ground".It also ensures that the problematic situation is clarified, the stakeholders' needs are specified, and new ideas are generated.A framework of operation of ICU is formed which permits us to define versions of the system which should be evaluated.Therefore, we concluded that we can use steps adapted from the SSM to design From the literature we know that despite the large amount of data that we can exploit, the accuracy of prediction models and clinical and strategic decision support systems in many cases is still not as desired.The main reasons for this are mainly the lack of knowledge about which data and from which sources are useful.It is open to the question of how the human point of view (the point of view of doctors and nursing staff) could be combined with clinical data.The findings of our research also suggest that ICUs should be viewed as part of a broader system of information sharing and data retrieval from both external and internal sources.Furthermore, it is important to point out that in many cases doctors/nursing staff are too busy or even reluctant to implement new systems in their daily routine.Due to its nature, ICU operation requires specialized and patientoriented treatment.In addition, each ICU may operate with different standards and procedures.From this point of view, decision support systems should not only follow certain basic principles but also be easily adaptable to different environments and needs.Therefore, continuous research is needed in order to improve the usability and accuracy of prediction models, and emphasis must be given to the daily changes in the fields of critical care medicine and the evolution of IT.The use of big data technology presents opportunities and challenges in the critical care field not only for its primary use but also for secondary purposes.Recent developments with the covid-19 pandemic require coordination of ICU units with their external environment and dynamic decision support systems with extensive interoperability, friendly user interface.Therefore, there is room for further research in the fields of selecting useful data, building and improving predictive models as well as improving the user interface of such systems so that they are more easily adopted by users such as doctors, nursing staff and patients' relatives.In the future, an extensive continuous inquiry has to be made in related to the topics of types of data and data flow in ICUs, ICD9/ICD10 based classification of patients' data, Electronic Health Records, various prediction models, clinical and strategic decision support, and eventually to the use of big data in ICUs.Specific topics that have to be also inquired include the secondary use of data, the development of specific prediction models e.g., models for patients' mortality, calculation of probability for re-admission, the standardization of data (FHIR and HL7), the ICU severity scales concerning patients' status as well as the new challenges that arise from the use of Big Data and IoT technologies and the new needs related to the pandemics.
Fillion et al. (2015) use steps adapted from the SSM to design an integrated Knowledge Management model that can be applied in a hospital and clinical context.

Figure 1 :
Figure 1: Stakeholders of the ICU system visualized with SSM method

Figure 2 :
Figure 2: Illustration of activity model incorporating data flow, clinical and strategic decision support the ICU/hospital level/national health system

Figure 3 :
Figure 3: Detailed activity model for the online processing of the various data types

Figure 4 :
Figure 4: Illustration of activity model for upgrading the skills of the ICU personnel and covering information needs