Healthcare in Motion

[construction icon]


Due to the aging population there is an increase in chronic diseases which require a more complex intra-, extra- and trans-mural organization of healthcare. The challenges for coordinating care and cure are increasing, but healthcare is ill-equipped to cope with complex multidisciplinary processes (lack of horizontal and vertical integration of processes and information). Another issue is the need for clinical quality improvement (QI) by means of practical tools for measuring and managing health care delivery processes to deal with the problem of poor quality and/or inefficiency waste in health care (B.C. James, 1989). The W.H.O estimates show that in developed countries as many as 1 in 10 patients is harmed while receiving hospital care (WHO, 2017). Major contributors to these hospital-acquired conditions (HACs) are adverse drug events, catheter associated urinary tract infections, patient falls, pressure ulcers, surgical site infection, central line associated infections, venous thrombo-embolism and ventilator associated pneumonia (VAP). So, let us take a look at some ideas and principles which may improve healthcare and avoid harm to the patient in the 21st century.

The article deals with the evolving dynamics of the healthcare process, meaning the flow of interactions between cure and care providers and patients due to the evolving healthcare landscape. The focus is on how to avoid the loss of efficiency and value due to gaps in the flow of people and information in healthcare. Healthcare is seen as an ecosystem in a landscape in which people and information flow in-between locations of cure and care providing. The healthcare ecosystem is the community of patients, physician and nurses in conjunction with the nonliving actors in the healthcare environment (intelligent and dumb machines, instruments, ... ), interacting as a system. At each point of action, value is created by the cure and care provider for the patient (man and machine). Information in the ecosystem should flow regardless of the physical barriers (distance, walls) at the right time and the right place to those people who need it to do their work (physicians, nurses,..) regardless of their location (intramural, extramural, trans-mural). The information reaches the people and the machines wherever they need it and where they need it (identity and access management for man and machine). It requires process integration and point of care integration in an ecosystem suitable for both man and machine.

Healthcare overall will need to develop more horizontal and vertical integration as it will require a more process driven approach across the boundaries of institutions (transmural) and across disciplines (transdisciplinary) to cope with the increasing complexity of care for our patients, more polypathology, and over longer periods of time (extended scope, extended time). In addition the architecture of the cure and care process will require transformation to align it with the components of the technology supporting it. A process is not independent of the supporting technology, be it supported by paper or more advanced technology. The more dimensions and features of a situation (status) and a process a technology is capable to capture and process, the better becomes the representation of reality, both in space as well as in time. Our new healthcare will need to improve its capacity to maintain more complex relations between patients and cure and care providers and other process participants (relationship management, process management). Besides this improvement of vertical and horizontal (human) relations, we will need to improve our management capacity towards our (internal and external) resources to provide cure and care (resource planning) with the appropriate means at the right time and place. Constructing the medical network of tomorrow will not be possible with the tools of yesterday, a complex (new) process requires innovated processes and instruments to operate (logistics and resources, management and monitoring tools). We should always keep in mind that "it is the process stupid".

A more project driven (iterative, agile) approach to healthcare will also provide more tansparancy and manageability of cure and care to our patients. Modern day healthcare management improvment efforts are already being inspired by enterprise architecture (EA) standards such as TOGAF and project standards such as PMBOK , CMMI and the Deming cycle (PDCA), which have their (medical) equivalent in standards such as JCI (accreditation) and clinical pathways (based on CPM) as part of the growing need for international healthcare accreditation to improve the quality of care. Healthcare providers will be managed based on their preformance for both quantity and quality of care in a transparent way. The so-called Project Management Triangle (called also Triple Constraint or the Iron Triangle) can be translated in the Healthcare Management Triangle (HMT) based on "scope" (features and quality, quality-adjusted life-year (QALY)), "time", and "cost". Healthcare improvement efforts require both process and outcome monitoring and control, which is cumbersome in a paper based process. Digitization requires process redesign, because simply translating a paper based process into digital leads to locking in inefficiencies, which were invisible due to the non-transparent nature of a paper based process. The life of man as such becomes the program encompassing all intra life events and sickness periods (the program of a human life is constituted of several interrelated projects and periods).

The technology supporting a more project driven healthcare should match the way we work with our patients and not the other way around. Technology supports people and processes and should not hinder their activities. Nowadays the interaction between man and ICT is still rather awkward and primitive. A keyboard and a screen are not an example of a rich interface for exchanging complex healthcare related information. The data of life should flow back an forth between the analog and digital world with as little deformation and distortion as possible. Digital systems equipped with rich interfaces back and forth to clinical reality can provide us with assistance beyond the limitations we have to work with nowadays in day to day practice. More processing power should be present within the clinical encounter and trajectory as we have at our disposal for the moment. More transparency is required from the interfaces between the analog (clinical) world and its digital counterpart.

This article deals mainly with improving process management and data flow and less with artificial intelligence (AI). Although in recent years artificial intelligent (AI) systems based on deep learning have made great progress, there is still a lot of work to do before they become anything more than "idiot savants". Winning a game of Go or chess is still far away from the complexity of real world problems. AI systems have a lot of specialized knowledge, but lack common sense. The basic principles of AI go back to the work of people such as Geoffrey Hinton and the principle of backpropagation, but will need a paradigm shift in order to succeed in broad real world situations. Backpropagation enables computers to learn by iteratively adjusting the weights of a neural network in order to minimize the error between the model's prediction and a ground truth comparison. Simplifying reality to fit AI by throwing away complexity is the wrong way around to solve real world problems. AI still breaks when the requirements of the problem it's trying to solve are slightly changed (which is why they work well in Go and chess). Modern systems may beat man on well defined problems, but they are still far away from general applicability in health care. AI will need to move beyond backpropagation if we want to teach computers to achieve unsupervised self-learning like that of human infants, according to Geoffrey Hinton (Artificial Intelligence Pioneer Says We Need to Start Over). Humans can learn concepts quickly in unstructured, unsupervised learning environments. Although AI systems are already capably to support medicine, they do not replace a specialist or a general practitioner who can deal with multimodal and unstructured information all at once. People have a (philosophical/intuitive) model in their mind about how the whole world works, which is (still) lacking for AI. Conceptual and symbolic thinking is a human capacity which AI systems (still) lack. Humans see both the trees and the forest with their capacity for multiscale and multimodal thinking. Well designed systems however could take some of the burden from our physicians and allow them to focus more on what they can do better than machines. Let us not wait for AI to catch up and let's build better data and processes already, in order to deal with our present day challenges and to prepare ourselves for the future. Better data and better processes will benefit both man and machine, but first of all our patients. We should try to avoid a "shazai kaiken" for not improving our healthcare data and processes to meet the challenges and demands of modern day healthcare.

Go with the flow

Healthcare embedded in a matrix of interacting systems
Figure 1: Healthcare embedded in a matrix of interacting systems.
Patient and care provider become embedded in a supporting process filled with communicating systems.
Increasing participation of interconnecting systems
Figure 2: Increasing participation of interconnecting systems.
The capabilities of the interposed systems to create true added value increases with their
capabiliteis to process the content of the transfer.

Essentialy the care-relation is a request and answer process between an individual asking for the solution of a problem related to its health. In primitive societies this one-on-one relation remains more or less intact. In modern society however, due to the increasing cost of healthcare, this relation becomes embedded into large networks of interrelated stakeholders. Only low-tech care (family) stays largely out of this increasingly complex web of interacting individuals and organizations.

Due to the level of complexity this process has reached nowadays, the patient itself becomes hidden under layers of organizational complexity which also become increasingly unmanageable to sustain. The overhead of the system and the resources required to keep the system affordable are reaching their limits in developed countries while in the developing world access to healthcare suboptimal. As there is no way back to the primitive situation, the only way out is to improve the efficiency of the entire process surrounding the patient care-provider relation (e.g. physician, nurse, ...).

Healthcare, like any human activity, is a process which is defined by a set of transformations of input elements into output elements with specific properties, with the transformations, embedded in a web of interactions, characterized by parameters and constraints. In order to be able to manage the complexity of modern healthcare, we need a process ontology as a model of the structure of the healthcare universe. This healthcare ontology is itself part of a larger set of ontologies, which provide us with a structural design pattern, out of which empirical phenomena can be explained and put together consistently in order to manage the healthcare process itself (like a set of matryoshka dolls). The (healthcare) process ontologies have to provide us with a description of the components and their relationships that make up a healthcare process, e.g.a Process Specification Language (PSL). Modern healthcare is not a linear unidirectional process, but a web of interactions between analog an digital actors, so the process ontology should be capable to deal with an hybrid, web-like process.
(see also Ontologies in Medicine, Domenico M. Pisanelli, IOS Press, 2004 and Medical Ontology: Approaches to the Metaphysics of Medicine, Jeremy Rosenbaum Simon, New York University, Graduate School of Arts and Science, 2006).

A healthcare ontology will have to define the capabilities of each healthcare actor at different levels of an organizational hierarchy (hospital, physician, ....). As such it will create a topology of the healthcare system which can be fed into an AI system capable to support the process flow of the system. One of the components of such a system is an intertwined set of NoSQL databases, each semantically linked to the analog-digital healthcare web.

The ExR universe

Managing digital relations
Figure 3: Managing digital relations.
Questions and answers in the digital world
Figure 4: Questions and answers in the digital world.

Part of the solution of the healthcare quagmire is to reinvent the entire process flow, with the goal to make it both perform more efficient (reduce cost) and more effective (reaching the right goal). This will require us to re-engineer the process and make it into a sustainable, lean, efficient and flexible system which is capable to deliver performance at the right place, time and person. Anything including the relation between care taker and provider should be optimized dramatically to reduce the overhead of the entire process and to refocus on the essential relation between patient and care provider (physician, nurse, ..). But simply converting the paper-based or digitized but siloed process to a networked workflow creates more even overhead and inconsistencies.

It may be better to think about the new process as a digital universe which needs to be shaped according to its own laws and inhabitants. An E(P)HR (electronic (personal) health record) is such an inhabitant. A personal health record (PHR), is a health record where health data and information related to the care of a patient is owned and maintained by the patient. PHRs are not the same as electronic health records (EHRs), which are software systems designed for use by health care providers. An EHR can be part of an PHR. Electronic medical records (EMRs) are a digital version of the paper charts in the clinician's office. An EHR focuses on the total health of the patient-going beyond standard clinical data collected in the provider's office and inclusive of a broader view on a patient's care. EHRs are designed to reach out beyond the health organization that originally collects and compiles the information. The information in the EHR also goes with the patient. The personal health record will over time evolve into a more complete representation of a person and become an ExR or ELR (Electronic Life Record) in itself capable of interacting with other systems or call it inhabitants of the digital universe. A universe of bots connected with the internet of things mirroring the physical universe of people and resources. Creating the digital process then becomes equivalent to creating an interrelated ExR world. The semantic web is a step in this direction.

Each participant of the process, represented by its Electronic x Record or virtual persona (bot) becomes connected to the other stakeholders as in the physical world. The way they interact of course will be different and the interaction process differs (no paper shuffling equivalent). Some of the stakeholders in this flow may have no real world equivalent, but only act as virtual intermediates. Due to the internet, they can interact with the speed of light, so distances are no longer an issue. There should be a dynamic balance between the physical and virtual world and a topological map between both representations of the healthcare continuum. The inhabitants of this digital world, will be linked to real-world counterparts in the real world. The interfaces between this digital world become windows through which we interact to create a mesh of flows which run both in our physical world and in the digital world (a PC with a keyboard is such a window, albeit primitive). The internet of things is one such step towards this digital parallel world which becomes an internet of relations. The boundaries between the digital and physical world will become transparent and process flow will run back and forth through the physical analog and the digital world (e.g. 3D printing and robot surgery). The way we interface with the digital world needs a big improvement in both usability and capacity as today we are still not capable to interact in a way natural enough to allow for the required ease of use.

An ExR memory and digital equivalent creator
Figure 5: An ExR memory and digital equivalent creator.
An ExR becomes a supporting memory for us and from which a digital dataset can be reconstructed
representing our health status and history.

A patient his or her ExR will be part of the "database record" representing complex and multiple aspects of a person in the digital environment, with healthcare-related information (CT, NMR, lab, ...) part of this digital persona (an advanced database system with sockets linking it to its surroundings).

Each ExR acts like an entity which itself belongs to multiple types (situational types) and its relationships present themselves as roles. Ideally, the digital representation should accompany each individual throughout its life and provide a "memory" for all healthcare related events (live events would be a more general description). The internet becomes connected with the human net and physical and digital reality become intertwined through a rich back and forth transformation of analog and digital information through interfaces capable to transform complex information, not just multimedia as we have today buth true polymedia (more than text, sound and images). So we provide our digital mirror image with a rich representation to work with.

As a personal assistant the ExR also acts like the thread of Ariadne within the labyrint of healthcare. It can signal the need for medical care either on-demand or triggered by smart devices (e.g. telemonitoring) and guide the patient and the care provider(s) towards each other when necessary. The care provider with the right skill profile will be contacted and guided towards the patient in case of an emergency. Wherever we go and wherever we are, nobody has to worry about finding appropriate care, the ExR will take care of that within the boundaries the patient allows it.

During a medical contact or intervention a physician or (his) assistant digital care-system could "ask" complex questions to the digital representation, e.g. in case of an emergency. The willingness of the ExR to share information will depend on the trust it has to its digital counterpart. Our digital ID becomes the key to our digital companion and the secure link through which interfaces can connect. A digital ID may in the end resemble a sort of xPod which we carry around and plug into the digital web when we want to identify ourselves and interact digitally. At that moment our digital persona is activated and ready to go (as a genie coming out of a lamp). A patient going through a CT-scan gets the result attached to its digital representation, although the data can stay at a server in the hospital or any secure vault. Whenever a physician needs the data, the virtual companion remembers the location and acts as an intermediary to ask for the data and provides them to the physician through its digital companion. The patient-centered healthcare system becomes a virtual reality. The data are the equivalent systems representing ourselves in-silico and the infrastructure nourishes and connects the data.

Unified Process Management

Healthcare process managment
Figure 5: Healthcare process mangagement system at work.

Electronic health records should not just replicate dumb paper charts and paper-based processes, otherwise they will stay as dumb as paper. An EHR-based system should also be patient centered as the data and the process has to follow the patient through the entire process and across the boundaries of health care organizations. The paper-based healthcare ecosystem lacks data analysis and management capabilities and stumbles around blindly, wasting resources and harming patients. In order to make the transition from a passive paper based process to an active digitized process, one should first and foremost analyze and model the workflow according to best practices, because automating a broken paper based process will ony get you an automated broken process in which workarounds become much harder. A common mistake happens when organizations believe they are unique and should do things in a special way and waste time and resources to create a Byzantine complexity instead of implementing best practices across industries or applying Occam's razor when redesigning processes.

An enterprise architecture (EA) approach could bring all components of the organization in line and guide them through the business, information, process, and technology changes necessary to execute the transition from paper based (incl. dumb EHR) to an integrated enterprise information management (EIM), healthcare orientend business process management (BPM), workflow management, business intelligence (BI) and decision support system. Enterprise information management would structure all data-systems and analytics systems to drive better healthcare outcomes. An active digitized process will enable process mining (Automated Business Process Discovery), which allows for developing and translating insights into process optimization and actions. This will improve health care outcomes for patients and reduce risks.

Feeding Subjective, Objective, Assessment, and Plan (SOAP) information into an intelligent system instead of putting them to sleep on paper or a dumb EHR, would benefit both the clinician, the nurse and the patient. A Clinical Decision Support System (CDSS) containing a knowledgebase, an inference engine and a mechanism to communicate alerts or clinical knowledge, should support the clinicians and nurses in their work. The CDSS should check conformance with clinical guidelines and protocols, protect patients against errors in prescribing, highlight critical laboratory results and display relevant clinical knowledge resources. By embedding the healthcare process into an intelligent system which plans, manages and controls resource usage and deployment, healthcare could become an outcome oriented analytic process. In most EHR Systems, managing the workflow is the problem which they cannot deal with. Dumb data capture and sharing is not enough, it is the process optimization which will lead to improved outcomes (e.g. Meaningful Use criteria). Each request for healthcare would launch a system which manages the entire process. Artificial Intelligence (AI) based systems would perform the support process based on structured input from care providers and devices. The AI system would organize patient routes and treatment plans, and also provide physicians with the information and analyses they would need to support their decisions and actions. A hospital would become an organization running multiple cure and care projects in parallel managed by an intelligent program managment system. Appendicitis projects run alongside pneumonia and cholecystis projects, etc.. All projects are closely monitored and analysed for quality and performance.

Unified Process Improvement Management - PDCA

Multiscale PDCA cycles
Figure 6: Multiscale PDCA cycles iteratively improving a multiscale process.

The system should be capable of self-management, self-control, self-improvement and (iterative) process optimization by means of machine learning. There are many types of machine learning techniques available, such as supervised, unsupervised, semi supervised, reinforcement and evolutionary learning (see also computational learning theory). Compared to machine learning methods, statistical models are less successful to hold categorical data, deal with missing values and large data points. A combination of machine learning and statistics could be used, such as statistical learning.

Self-management, self-control could be built upon the principles of cybernetics. Cybernetics, a term coined by Norbert Wiener (1894-1964 CE), refers to organizations or systems which depend on each other to function, and whose interdependence requires flexibility of response. A patient embedded in an health care system becomes part of such an organization and system. The health care system monitors the desired outcome of the patient's health or at least the (operational) stabilization of his situation (e.g. ICU). The entire system incorporates a closed signaling loop, where a change in the patient health status triggers a system change and an appropriate response by the system (physician, nurse, device) (feedback loop). These control loops either run entirely in digital-mechanical systems, or apply a human-machine hybrid process. The system can operate at the level of a Point of Care (PoC), organization, region or national (population, disasters, terrorism, epidemics, etc...). When you take a look at accreditation of health care providers, such as the Joint Commission International (JCI), it resembles cybernetics in the ways it intends to improve monitoring and control of the healthcare process. It prepares them for data driven health care, process monitoring and control (PMC).

Self-improvement and (iterative) process optimization should be built into the system. The system is embedded in an ever improving iterative Deming Cycle operating on multiple organizational levels, based on machine-learning and other artificial intelligence techniques (figure 6). The Deming Cycle operates at several levels of process integration and balances optimizations at (interconnected) multiscale levels: Point of Care (PoC), hospital, local, regional, national, etc. ... . Healthcare processes could be designed by means of Activity Diagrams (UML), etc., embedded into the Healthcare Management System (national, regional), Hospital Management System, and PoC Management System. Data aggregation and disaggregation meet the management needs at different levels. The PoC itself could be intramural and extramural (primary care) of course. The process diagram should be both man and machine readable. Process management and control should be generated from the diagram itself. Optimizations should avoid point-optimization (fragmented point solutions) resulting in global process deterioration as is often the case in self-interest driven optimizations (see also Eight Levels of the Analytics Adoption Model).

Unified Project Management

Unified Process Management
Figure 7a: Unified Project Management.
Each care episode becomes a small project.
Project progress
Figure 7b: Project progress
Health status evolution with and without complication.

Each patient becomes part of a project to be dealt with in an efficient and effective way (scope or health outcome, budget, time and quality) (figure 7a). The project has a scope (target), which is to restore the health status of the patient in the example from 40% to 100% (simplified) (figure 7b). Certain forces will improve the health status of the patient (green arrow pointing upward), while other forces will decrease the health status (red arrow pointing downward). At t=9 an avoidable error happens, which causes the health status of the patient to deteriorate (figure 7b). Although the problem is solved (t=11), the health status of the patient never reaches the intended scope (red line) compared to the original plan (black line). In project management terminology the event (complication) is an issue, which is something negative to the project that is happening. Before this, it was a risk, which is something that might happen. With an issue happening, the team has to figure out how to resolve it for the project to move forward. With an identifiable risk, a healthcare provider (physician, nurse, hospital, ...) should establish mitigation plans that will (hopefully) eliminate the possibility of the risk occurring or reduce the impact if it does occur (risk mitigation). Examples of healthcare associated risks are so-called hospital-acquired conditions (HACs) such as adverse drug events, catheter associated urinary tract infections, patient falls, pressure ulcers, surgical site infection, central line associated infections, venous thrombo-embolism and ventilator associated pneumonia (VAP). Potentially Preventable Complications (PPCs) are harmful events or issues (e.g. accidental laceration during a procedure, improper administration of medication) or negative outcomes (e.g., hospital-acquired pneumonia, C. difficile colitis) that develop after hospital admission and may result from processes of care and treatment. Risk mitigation (risk reduction) is a systematic reduction in the extent of exposure to a risk and/or the likelihood of its occurrence. Once a risk occurs, it becomes an issue, which is what project management and professionals try to avoid. As an example accreditation is based upon risk mitigation (risk reduction) techniques in order to avoid harm to the patient. What would professional project management bring us, inspired by accreditation and supported by intelligent systems? Some thoughts on using project management principles and intelligent systems to support healthcare workers to achieve their goal and to reduce risk exposure for our patients.

An intelligent system which never sleeps and doesn't get tired becomes the project manager based upon adaptable diagnostic and therapeutic trajectories and in close operation with its human partners monitors and controls the project phases. Each care episode becomes a (small) project, easily traceable because of its associated data following the patient. Transfers of a patient between healthcare providers or in between departments of hospitals make no difference, as the data are attached to and owned by the patient. The patient and his data are inseparable. Rich data on project status flow continuously in an out the management system. Outcome measures are to be used for defining quality and cost targets (mortality, complications, readmission, patient experience, etc.). Project planning, monitoring and control and lessons learned analysis, have to reduce unacceptable healthcare errors, accidents, and injuries and as a result also reduce the per capita cost of healthcare.

Each interaction between healthcare professionals and with patients are regarded as transactions of which the efficiency and effectiveness needs to be guaranteed and therefore monitored (surgery, medication, ...). The health condition of the patient should improve graciously throughout the entire process. The system and its professionals involved in patient care should keep in mind the principle "Primum non nocere" or non-maleficence. The project should evolve towards a positive evolution of the health status of the patient without harm caused by process failure. Healthcare providers should to be in control of the project aimed at improving the health status of their patients and not let the project control them. Decent execution of basic processes such as surgical and other invasive procedures, medication management, etc. should be self evident and not add problems to the patient in need of professional help. Several aspects of a project contribute to failure of success of disease treatment: the level of accuracy of execution (professionalism), efficient and effective organizational procedures, control thresholds (variance thresholds, alerts), units of measure which make sense (speed of change and risk based, ER, OR, ICU), effective rules of performance measurement (earned health value rules), reporting formats which the prime actors understand and acknowledge (multilevel measurement and feedback loops), and finally practical process descriptions which are aligned to reality at the work floor.

Earned Health Management (EHM) becomes possible, because of the traceability of patient data and health status. Dealing with traceable patient data makes possible to evaluate the performance of healthcare with regard to its efficiency and effectiveness. For instance a patient presenting with symptoms such as a dull pain near the navel or the upper abdomen that becomes sharp as it moves to the lower right abdomen (usually the first sign). In addition loss of appetite, nausea and/or vomiting soon after abdominal pain begins, abdominal swelling, fever and the inability to pass gas. The working hypothesis in this case is appendicitis, which sets in motion a diagnostic and therapeutic project. For such a project we should take into account life-cycle costing, where you look at the "total benefit and cost" when a treatment succeeds or fails for the patient, his family, the healthcare provider and for society. In this case for instance the overall cost versus benefit of a laparoscopic versus a traditional appendectomy. Take into account direct costs and indirect costs or the patient, his family the healthcare provider and for society. Cost estimates are a continuing prediction, based on information known at a given point in time and attached to the patient data (scope, time, budget, quality). This becomes more important in complex diagnostic and therapeutic trajectories. The cost estimates are refined during the course of the treatment as additional detail becomes available and should be continued even after the patient has left the care setting or finished treatment (continuous when dealing with a chronic disease). Health value engineering should deal with getting more out of the treatment for the patient and society in every possible way. A health status baseline is made is made at each transition in the process (should include assumption about direct and indirect costs) and the Diagnostic and Treatment Breakdown Structure (DTBS) is put in place for the next phase in order to achieve the health scope baseline (a DTBS is a Work Breakdown Structure or WBS). Execution of the scheduled DTBS according to plan and appropriate change management along the way gets the patient to the achievable health scope baseline, from "as is" to "to be". The health scope baseline lists the limitations for the period under consideration for expenditure of funds, funding constraints, and diagnostic and therapeutic assumptions (requires qualified professionals and supporting systems). The working hypothesis (differential diagnosis) at the start of the trajectory provides the initial guidance for diagnosis and treatment, e.g. appendicitis based on symptoms and signs. A project schedule, in this case for appendicitis, contains all the (initial) activities, types of labor (surgery, ...) and resources (personnel, lab, OR, ...) which the (intelligent) healthcare management system monitors and controls. The healthcare management system guides the patient through the entire project using RFID and/barcodes to track and trace the patient through his diagnostic and therapeutic schedule supported by a Gantt chart or a PERT diagram. The healthcare management system continuously monitors the consumption of all resources compared to the work required to be accomplished for the diagnostic and therapeutic project schedule. The system continuously monitors and controls the cost variance where positive variance is good, and negative cost variance is bad. Ideally the actual costs match what was budgeted and the cost variance is zero in an ideal situation.

A dynamic change management system keeps track of the diagnostic and therapeutic project along the way and continuously monitors and controls risks and issues which would require an adaptation of the diagnostic and treatment plan. Both man and machine work together to keep the trajectory of the patient on track. The monitoring and control system performs failure mode and effects analysis (FMEA) on the diagnostic and therapeutic activities. The system should continuously identify and prevent diagnostic and therapeutic problems before they occur, reducing the risk of sentinel events and system error-related occurrences (delay in treatment, negligence, wrong-patient, wrong-site, wrong-procedure, medication errors, op/post-op complication, unintended retention of a foreign body, fall, nosocomial infections, ... ). Rate-based and sentinel indicators allow for monitoring the entire project. They can be generic or disease-specific, and related to either structure, process, or outcome (e.g. rate of contaminated wound infection, hospital-acquired infection, ...). Sentinel events indicate poor performance by man and/or machine and they are used for risk management and process improvement (e.g., patient misidentification, wrong surgery performed, patients who die during surgery, patients who die during the perinatal period, ...). The system monitors events and triggers further analysis and investigation (e.g. based on the Hospital Consumer Assessment of Healthcare Providers and Systems (HCAHPS) outcome measures grouped into seven categories: Mortality, Safety of care, Readmissions, Patient experience, Effectiveness of care, Timeliness of care, Efficient use of medical imaging).

Earned Health Management (EHM) as a tool allows for measuring the actual performance of the diagnostic and therapeutic activities against the original plan for this patient. The healthcare management system continuously identifies areas where the diagnostic and therapeutic project is different from the original plan, and identifies variances and trends that influence the diagnostic and therapeutic costs. Planned Health (PH) can be compared to Earned Health (EH), etc. . All diagnostic ans therapeutic projects and activities are being planned, monitored and controlled in a program management system which is capable of managing resources (man and machine).

For those who are familiar with professional project management it should be clear that the principles of project management can be applied to healthcare taking into account the sometimes dynamic nature of a diagnostic and therapeutic plan. The main issue is that healthcare workers are not accustomed to the principles of project management and its language. The tools which support projects in other industries need to be tuned to healthcare projects and its users. Clinical pathways and accreditation are examples of an attempt to bring project management to healthcare. However, the productivity gains which can be achieved in healthcare are limited because of the Baumol effect. It is mainly the cost of poor quality (COPQ) or cost of diagnostic and therapeutic failure, which can be reduced by process improvement at least for the near future.

Unified N-dimensional Analysis

Treatment vector
Figure 8: Health status change vector (scope, z-axis), as a function of time (x-axis) and cost (y-axis) of treatment.
Origin (0,0,0) is the "as is" situation at intake (baseline), while the vector points at the desired,
and in this case, achieved "to be" situation (time, cost, health status).

The analysis of health care data (health care analytics) is required at different levels and situations. First of all we have Point of Care (PoC) analytics, which support identitovigilance (patient identification tracking), patient history and current status, clinical decision support and reporting. Hospital and population based analytics deal with trend analysis, pharmacovigilance and clinical audits. Analytics for clinical research deal with the identification of clinical trail candidates and predictive medicine. Point of Care (PoC) analytics should assist physicians and nurses in remembering preventative measures, identify patients with potential care gaps and risk factors, monitor patient compliance with prescribed treatments and finally report (anonymous) clinical data to disease, treatment and outcome registries. A Point of Care (PoC) process analysis system documents the "as is" state at patient intake (baseline) compared to the "to be" or future state in relation to the real gain in health status at the end of the process (fit gap analysis, figure 8). It evaluates if the health care process was capable to achieve the goal of improving a patient's health status (effectiveness and efficiency). Keep in mind the Triple Constraint of project management when planning and performing a project (diagnostic and therapeutic) to treat a patient (multi-factorial optimization of treatment, revenue cycle management (RCM), earned value method (EVM)). Trend analysis should assist long term population health monitoring, prediction of demand, and effective resource allocation. Trend analysis should also assist short term resource planning and allocation, such as in case of epidemics, natural disasters or terrorism (ER, ICU and OR resources) on a local, regional, national and international scale. Pharmacovigilance should support collection, detection, assessment, monitoring and prevention of adverse effects with pharmaceutical products. Clinical audits support improving patient care and outcomes through systematic review of care against defined standards and the implementation of change (Plan-do-check-act (PDCA)).

Ontology-Oriented System (OOS)
Figure 9: Ontological web - Ontology-Oriented System (OOS).
Red dots (new data), blue dots (old, pre-existing data), white dots (nodes in the ontological web).

Symptoms, signs, etc. become embedded into an Ontology-Oriented System (OOS), which becomes the basis of a knowledge-based diagnosis built upon on a well-defined ontology. The ontology transforms the relationships among symptoms, signs, personal history, etc., into a machine-interpretable model (figure 9). The analytical system has to take into account the diagnostic criteria, on which specific patient conditions may be classified under a specific disease, into its inference engine (association of signs and symptoms=diagnostic criteria). An inference engine could apply genetic reasoning, meaning crossmatching both regressive analytic and forward synthetic reasoning in order to analyze the supplied information. The OOS includes a specification of conceptualizations that constitutes evidence-based clinical practice guidelines. It represents the elements of an evidence-based guideline by specifying its attributes and defining the relationships that hold among them.

N-dimensional space
Figure 10a: N-dimensional space
N-dimensional space
Figure 10b: A.I. Experiment example, as part of TensorFlow (animation).

The data attached to the patient create an N-dimensional space or landscape which can be used by A.I. systems for visualization, analysis, diagnosis, treatment support and process management (figure 10a). Feeding the system with clinical data such as SNOMED CT based data (325,000+ concepts), laboratory data such as LOINC, etc., would create a vast data universe upon which intelligent systems could be built. As such the system not only supports individual diagnosis and treatment, but also population studies and epidemiology on a regional, national and international scale (figure 10b). It will be very important to avoid vendor lock-in due to proprietary or regional standards, because this will create unacceptable risks for our patients. For instance, infectious diseases act on a global scale, just as criminal organizations or global-scale cybercrime, but we still deal with medical data in a very primitive way. Analyzing and fighting diseases in a certain way resembles analyzing and fighting criminal organizations.

Unified Information Flow Management

Unified Information Flow Management
Figure 11: Unified Information Flow Management.
Information follows the patient independent of an extramural or intramural setting.

Every patient and every healthcare provider is in a supply and demand combination requiring resource management and capability management. Current healthcare system suffer from vendor lock-in as they are application centered and not process and data centered also due to a lack of decent standards. Information about care demands and supply capabilities flows though the system attached to the patient and provided to the appropriete healthcare provider and institution (figure 11). One could imagine an encrypted blockchain of medical data (Summarized Electronic Health Record) stored in a distributed database or cloud which maintains a patient's continuously-growing list of health data records. The lingua franca of the system could be based on for instance SNOMED-CT, LOINC, DICOM and HL7. The clinical data handover at each transition of healthcare station (intra- and extramural) takes care of the ISBAR (Identify, Situation, Background, Assessment and Recommendation) process. This way the walls between intramural (departmental) or extramural healthcare disappear as it is no longer the physical organization of healthcare which determines the flow of data, but the demand (based on the position within the healthcare chain or web) of the patient for the appropriate type of healthcare from the cloud of healthcare service providers. Online reputation management will have to be part of this system. A decent encryption standard such as Rijndael (AES) is also crucial for the system to be safe. Blockchain technology is also capable to transform the healthcare system into a patient centered system. Blockchain technology could provide a new model for health information exchange (HIE) by making electronic medical records more efficient, disintermediated, and secure.

Unified Resource Management

Healthcare resource demand and supply management web
Figure 12: Healthcare supply and demand management web for a patient centric healthcare.
Patient and care provider resources become embedded in web containing connected healthcare units (society, hospitals, primary care, ... ).

Management of healthcare resources should become more dynamic and responsive. A unified system should be capable to respond in a dynamic way when a patient demands a healthcare service. The information flow should be independent of either an intramural or extramural environment. The medical demand management system manages the push and pull of intramural and extramural medical resources (dynamic healthcare resource monitoring and management) (figure 12). The resources present and available for providing care, should be summoned for service in a dynamic way by a system which is capable to manage the human and non-human resource pool in a healtcare providing entity. A Unified Resource Management (URM) system keeps track of each type of resource and its characteristics and in this way is capable to optimize the response to patients entering the entity even in an unplanned way (about 30 percent of cases on average). The activity itself should generate its own data, so no overhead is being created to translate activity into data. This should minimize the slow response rates due to the complexity of coordination in for instance emergency rooms. Everything and everybody is considered as a resource with its own characterisitcs and is actively assisted in its work planning and execution. The process of information gathering and treatment providing develops into a hybrid system consisting of an interaction between physical and digital data extraction and treatment generation. Both clinical decision support systems (CDSS) as clinical treatment support systems (CTSS) such as robots and bots interact with patients and (human) healthcare providers, building at transparent and unified system. The system will have to be a learning system with the capacity for improving its performance. This way the unified healthcare organism intertwining human and non-human resources in a dynamic way could become a reality.

The way forward

How do you eat the elephant which is healthcare? One bite at a time. When you need to do something that is difficult, do it slowly and be careful. What and how this digital presentation and Unified Resource Management (URM) operates is part of ongoing development, from an P(E)HR to a complete virtual companion or ExR evolving in a Unified Resource Managed environment. The flexibility with which the digital components are capable to interact with its (medical) environment will to a large extent determine the feasibility of such a system. In a patient-physician encounter for instance, the patient explains his symptoms which are digitized into his digital companion to be remembered as part of its ExR, and who transmits them also to the digital companion of the physician which tells its real world counterpart what the anamnesis could suggest. The physician performs its clinical and technical exams and its digital counterpart acts as its agent in the digital world to manage the interactions with the radiologist and the lab. A stethoscope or any other device not only transmits the information to a physician, but also to his digital companion for assistance - the Web of Things (WoT) and the Web of Ontologies interconnect to the web of persons (healthcare professionals). The more sophisticated the digital companions become the more they are capable to create added value for the overall system by moving away from the mostly passive digital systems of today towards true physical-digital partnerships. The WWW evolves from a web of devices (things) into a hybrid and intertwined part of everyday healthcare reality.

In order to construct a new health care system, it is more important to understand the true needs of our patients and health care workers, than to introduce a multitude of new technologies, tools and gadgets. Understanding, analyzing and designing a truly integrated system which unites man, machine and technology from a process oriented (efficient) and result driven (effective) perspective is the alpha and omega of success. As Henri Poincaré once wrote in La science et l'hypothèse (1902): "Le savant doit ordonner; on fait la science avec des faits comme une maison avec des pierres; mais une accumulation de faits n'est pas plus une science qu'un tas de pierres n'est une maison."

Closing remark. The new healthcare model will only succeed when the younger generation of digital natives comes of age to be the new healthcare consumers (patients), (policy) providers, managers and politicians. As Max Planck once said "Eine neue wissenschaftliche Wahrheit pflegt sich nicht in der Weise durchzusetzen, daß ihre Gegner überzeugt werden und sich als belehrt erklären, sondern vielmehr dadurch, daß ihre Gegner allmählich aussterben und daß die heranwachsende Generation von vornherein mit der Wahrheit vertraut gemacht ist." (Wissenschaftliche Selbstbiographie. Mit einem Bildnis und der von Max von Laue gehaltenen Traueransprache, Max Planck, Johann Ambrosius Barth Verlag, Leipzig 1948, p. 22).

See also


  • Quality Management for Health Care Delivery, B.C. James, Hospital Research and Educational Trust [American Hospital Association], Chicago, IL, 1989
  • To Err Is Human: Building a Safer Health System, Institute of Medicine (author), National Academies Press, 1 apr. 2000
  • Cost of Poor Quality or Waste in Integrated Delivery System Settings, Agency for Healthcare Research and Quality, September 2006
  • Patient Safety: An Engineering Approach, B.S. Dhillon, CRC Press, 8 nov. 2011
  • Patient Identification Errors, ECRI Institute, June 2016
  • Clinical Decision Support Systems: Theory and Practice, Eta S. Berner, Springer Science & Business Media, 3 apr. 2007
  • The Healthcare Imperative: Lowering Costs and Improving Outcomes: Workshop Series Summary, Institute of Medicine, Roundtable on Evidence-Based Medicine, National Academies Press, 17 dec. 2010
  • Philosophy of Artificial Intelligence: A Critique of the Mechanistic Theory of Mind, Rajakishore Nath, Universal-Publishers, 2009
  • Medical Applications of Artificial Intelligence, Arvin Agah, CRC Press, 6 nov. 2013
  • Artificial Intelligence in Medicine: Expert Systems, M. Fieschi, Springer, 11 nov. 2013
  • An Introduction to Statistical Learning, G. James, D. Witten, T. Hastie, and R. Tibshirani, Springer, 2013
  • Rumelhart, D., Hinton G., Williams, R., Learning representations by back-propagating errors, Nature, 1986, Oct;323, 533-536
  • Taylor MJ, McNicholas C, Nicolay C, Darzi A, Bell D, Reed JE., Systematic review of the application of the plan-do-study-act method to improve quality in healthcare, BMJ Quality & Safety, 2014, 23(4):290-298
  • Hederman L, Smutek D, Wade V, Knape T., Representing clinical guidelines in UML: a comparative study, Stud Health Technol Inform., 2002, 90:471-7
  • Isern D, Moreno A., Computer-based execution of clinical guidelines: a review, Int J Med Inform., 2008, Dec;77(12):787-808
  • Vinterbo S, Ohno-Machado L., A genetic algorithm approach to multi-disorder diagnosis, Artif Intell Med., 2000 Feb, 18(2):117-32
  • Celi, L. A., Marshall, J. D., Lai, Y., & Stone, D. J., Disrupting Electronic Health Records Systems: The Next Generation, JMIR Medical Informatics, 2015, 3(4), e34.
  • Kononenko, I., Machine Learning for Medical Diagnosis: History, State of the Art and Perspective, Journal of Artificial Intelligence in Medicine, 2001, 1, 89-109
  • Singh, Y., Bhatia, P.K., and Sangwan, O.,A Review of Studies on Machine Learning Techniques, International Journal of Computer Science and Security, 2007, 1, 70-84
  • Archana, S. and Elangovan, K., (2014) Survey of Classification Techniques in Data Mining, International Journal of Computer Science and Mobile Applications, 2014, 2, 65-71
  • Fatima, M. and Pasha, M., Survey of Machine Learning Algorithms for Disease Diagnostic, Journal of Intelligent Learning Systems and Applications, 2015, 9, 1-16
  • Rosenbloom ST, Miller RA, Johnson KB, Elkin PL, Brown SH, Interface Terminologies: Facilitating Direct Entry of Clinical Data into Electronic Health Record Systems, Journal of the American Medical Informatics Association: JAMIA, 2006, 13(3):277-288
  • Chute CG, Clinical Classification and Terminology: Some History and Current Observations, Journal of the American Medical Informatics Association: JAMIA., 2000, 7(3):298-303
  • Cimino JJ, Desiderata for controlled medical vocabularies in the twenty-first century, Methods Inf Med., 1998 Nov, 37(4-5):394-403
  • Cimino JJ, In defense of the Desiderata, J Biomed Inform., 2006 Jun, 39(3):299-306
  • Copyright notice and disclaimer

    These webpages of course represent only personal interests, opinions and ideas and were created without a commercial goal. You may download, display, print and copy, any material at this website, in unaltered form only, for your personal use or for non-commercial use within your organization.
    Should these webpages or portions of these webpages be used on any Internet or World Wide Web page or informational presentation, that a link back to this website (and where appropriate back to the source document) be established. Send a short notice by email when you copy these webpages, or part of it for your own use.
    Any information here is provided in good faith but no warranty can be made for its accuracy. As this is a work in progress, it is still incomplete and even inaccurate. Although care has been taken in preparing the information contained in my webpages, I do not and cannot guarantee the accuracy thereof. Anyone using the information does so at their own risk and shall be deemed to indemnify me from any and all injury or damage arising from such use.
    To the best of my knowledge, all graphics, text and other presentations not created by me on my webpages are in the public domain and freely available from various sources on the Internet or elsewhere and/or kindly provided by the owner.
    If you notice something incorrect or have any questions, feel free to send me an email.

    The author of this Webpage is Peter Van Osta
    Email: pvosta at gmail dot com

    Latest update: 22 January 2018.