Knowledge development,technology and questions of nursing ethics

Introduction

We are now in what many are calling the Fourth Industrial Revolution, with its melding of the physical, digital, and biological worlds. ¹ Knowledge generated through machine learning is being used to make decisions about our everyday lives, including healthcare. This knowledge is treated as truth. Yet how and why this knowledge is developed is not always known nor critically analyzed. There is also a growing recognition that living systems are not fully controllable, nor replicable because of their dynamic nature. Instead of absolute, never wavering results, in complex biological models there are patterns, where what was known as error may instead represent normal variation and therefore truth. In these instances, our fundamental ways of developing knowledge and knowing truth are under question, and with these questions come ethical concerns.

Melding the physical, digital, and biological approaches to knowledge development is possible only because of the advent of new computer driven technologies. The application of these developments comes at a time when fundamental changes are needed in the organization and delivery of healthcare. ² The promise to radically and positively transform healthcare through technology may seem compelling especially when it is couched in terms of safety, efficiency, and effectiveness but its role in the provision of ethical nursing care remains uncertain.

Nursing has the ethical obligation to develop, synchronize and exchange significant clinical and technical knowledge with the goal of supporting and coordinating safe, effective patient care. Understanding how the progression of technology fits into the development of nursing knowledge while staying true to the goals of nursing science is crucial. Given the profound moral and clinical implications of how today’s knowledge is developed and utilized, it is time to reconsider the relationship between ethics and knowledge development in this new uncharted era.

Knowledge: past and present

Our epistemology has been strongly influenced by empiricism, Newtonian mechanics, and linearity. Within the post-Cartesian worldview the “best” scientific knowledge was derived from objective, controlled experimentation. The linear statistics that forms the backbone of nursing science is predicated on the numerical average, the concept of statistical error, and the ideal of complete predictability and control. To nursing’s credit, both qualitative and quantitative methodologies have been thoughtfully used to explore our domain of knowledge. The qualitative approaches have produced a richness of understanding through the words and artifacts used to know our world. In both approaches, nursing has come to rely upon standard methods to determine the reliability and validity or reliable truth of the findings.

The scientific method is based on thoughtful hypotheses, design, intervention, and analysis. When well done, it represents high individual achievement in the quest for knowledge. On the other hand, the new technologies, augmented by supercomputing are driving knowledge development that has little human input into the data collected, the questions asked, and analysis undertaken. Because of these differences in knowledge development new ways of ascertaining truth need to be explored in both machine learning and biological modeling.

Nursing cannot utilize knowledge from unknown sources without critical analysis as it has the potential to jeopardize nursing theory-guided practice. ³ Up to this point, we have looked for the traditional markers of ascertaining truth including clinical trials, experimentation, peer review, reliability, validity, error rate, sampling, and so forth. It has been assumed that given the correct information and enough of it, nurse scientists should be able to discern the truth.

Knowledge is both individual and shared; a continuum from which we operate in our personal and professional lives. As the philosopher Phillip Kitchener says, knowledge represents a cognitive division of labor. ⁴ All fields of inquiry possess both a unique body of knowledge as well as a shared understanding of the world from other perspectives. In nursing, we rely on the knowledge, obtained from alternate disciplinary sources, such as psychology, pharmacology, physiology, and so forth, to be reliable and truthful. However, knowledge from other disciplines may lack transparency, control, and accountability in its design, and reporting.

Zanotti and Chiffi ⁵ note that epistemic values with their focus on truth and objectivity can be at odds with nonepistemic values, where nonscientific influences come into play. They state that nursing values should be inherent in our knowledge development. ⁵ We argue that technologically assisted ways of knowing should not be value free, but rather invested in ethics; reflective of nursing’s social, political, and cultural values. In the following sections, we discuss the importance of each of these new ways of knowing to nursing, along with our concerns.

Big Data

The use of Big Data, coupled with increasingly powerful computing ability, is fundamentally altering healthcare. The term Big Data is vaguely defined; in some cases, it represents large purposeful data collection such as census data, but increasingly it has come to mean “found” data. Found data is also called “digital exhaust”—the data trails produced by all online digital activities, related or not. ⁶

The Big Data proponents cite several important benefits of large data sets and machine generated knowledge. ⁷ First, analysis of large data sets often produces accurate and sometimes unexpected results. It is not unusual for massive amounts of data to provide insights that were previously unrealized. Second, because the data sets are so large, sampling protocols are considered unnecessary. The idea that “N = All” is compelling and is assumed to eliminate the biases inherent in samples. Third, simpler correlational analysis is proposed as sufficient to generate answers and fourth, the scientific method is thought to be obsolete as “the numbers speak for themselves.” ⁷

The benefits of Big Data are critical to the future of healthcare. Large data sets make it possible to uncover previously unknown patterns revealing new avenues for research as well as clinical care. However, the assumptions about Big Data can and should be challenged. When research is designed a priori, the data to be collected is thoughtfully considered prior to the study. Traditional research methodologies have a frame of causality; as either the real or ideal end game. With large data sets, all information is considered equally significant and not until the correlations are completed does a picture of relationships emerge. And while causality is implied in results from large data sets, and the results are treated as causal, they are in fact not causal but rather correlational. ⁷

We argue that sampling bias can exist even in Big Data sets; N does not really equal “All,” just the “All” that was captured. Spurious patterns exist and correlational false positives are possible even in large data sets. As to the minimization of the scientific method, at least three interconnected questions arise with the knowledge generated by machine. Is the correct information being collected, is it verifiable, and who makes the decision as to what information is collected? ⁸ Experts have warned that there is built-in hidden, and perhaps unrealized sexism, racism, and other discrimination because those building the systems may choose to dismiss or not even see the issues involved. ⁹ It should be remembered that the algorithms used to understand Big Data reflect the values of the developers and sometimes the companies with which they work. It is not value free and must not be treated as such even though it may contain millions of data points; it is still a snapshot of the data someone deemed important to collect in the first place. The machine-generated data may give an answer, but not necessarily an understanding. Dennet refers to this as an inversion of reasoning, where computers are capable of achieving competence without comprehension. ¹⁰

Dennet ¹⁰ has voiced concerns with the increasing transparency of everyday life in which the “membranes protecting our lives and institutions have been permeated and secrets are almost impossible.” The collected information used in Big Data analysis comes from everywhere: grocery shopping, pharmacy purchases, doctor visits, cars, online shopping, and even the robots used to clean our floors. Even more disturbing is that most of this information is collected without any permission.

In some ways, the knowledge generated by Big Data resembles Plato’s Allegory of the Cave where what was viewed appeared real, but was a shadow of reality. Big Data provides a worldview that is a reflection of everyone but of no one. Decisions about many aspects of life, from buying habits to technology usage, may soon be predicated on these massive data sets. But we question whether the depth, breadth, and variety of information necessary to make critical decisions are captured by this artificial reality? The revealed truth may not be the whole truth but rather a shadow of truth.

The influence of Big Data on nursing seems evident. Digital exhaust is used to make potentially life-altering decisions about individual healthcare, as well as providing guidance to healthcare businesses and providers. We do not deny that there are many benefits from Big Data including the uncovering of unknown correlations and hidden patterns. Unfortunately, there are few who have voiced concerns about the downside of Big Data, including the ethical implications related to privacy, harm, autonomy, and justice. Massive data collections are not morally neutral for either patient or nurse. Some basic questions that need to be asked are as follows: who owns an individual’s data, is it right for companies to profit from data collected without the individual’s approval for its further use, are proprietary algorithms ethical, can the information be traced back to the individual, how can privacy be protected, and so forth.

Nurses should be aware of the benefits of Big Data but also concerned about how it is collected, by whom, and for what reason. The truth of Big Data and its use in healthcare can and should be discussed. To be unaware of the utilization of Big Data in the provision of care is not ethical, nor is it safe. Nurses are positioned at the axis point between the patient and the healthcare system and as such must take a primary role in interpreting and influencing the relationships among technology, healthcare praxis, and the human experience. ¹¹ Although computer generated knowledge is rampant, its progress is not inevitable. As Dennet noted computers do not have agency, they can be unplugged. ¹⁰

Artificial intelligence

Artificial intelligence (AI) or as some prefer to call it, machine learning, is learned problem solving by an artificial entity; generally assumed to be a computer or machine. ¹¹ Most recent advances in AI have come from tremendous computer power combined with very large data sets (Big Data).

AI was created with the sole aim of mimicking or outperforming humans. It creates knowledge processing systems with knowledge representation, acquisition, inference, search, and control. ¹² AI has two approaches to learning: bottom up and top down. Bottom up, which is also called deep learning, occurs when raw data are used without supervised direction about how the machine is to learn. The top down uses a Bayesian approach, using the known to guess the unknown. These are not new approaches to understanding the world; Plato, Descartes, and the other rationalists could be considered “top down” philosophers through their use of reasoned deduction. Bacon, Hume, and other empiricists could be classified as “bottom up” philosophers who advocated the use of raw data to build theory. ¹³

With respect to AI, healthcare has been quick to adopt its positive potential. The ever-increasing complexity and demands of healthcare including countless and ever-changing treatment options has led to AI serving as a framework to help clinicians address problems in what appears to be a timely, effective, and efficient manner. In some cases, AI is created to think as a clinician, such as IBM’s Watson, in others, it serves as a sorting aid. The principle behind most of AI is simply that machines can be made to act as if they are intelligent; but are they? It has been suggested that machines have the potential to develop an alternate form of intelligence that could pose a threat to humans. While AI now involves human interface, a future without human exchange is possible.

Dennet ¹⁰ says we should not put our whole trust in machine-generated knowledge if we don’t understand how it came to its decision because “If it can’t do better than us at explaining what it’s doing, then don’t trust it.” In AI, it is often unclear how decisions are made. Without knowing how decisions are made it is impossible to know what and how information is being used in the decision process. In both top down and bottom up learning, decisions reflect a multilayered neural network where it is hard to know where the important decision points are found. New efforts are underway to build AI programs that elucidate their reasoning. Researchers exploring explainable AI, or what is called X.A.I., hope to reconcile the transparency issue and allow humans to make informed decisions about the AI directives. ¹⁴ As Barzilay notes for healthcare to be truly effective “…you need a loop where the machine and the human collaborate.” ¹⁵ Without this key understanding, there are diagnostic as well as ethical issues. It is not difficult to imagine that AI decisions designed around efficiency could create very different decisions than those based on more patient-centered values. Unless the nurse is aware of how the computer arrived at the decision, there is no way to know if it is an appropriate patient-centered conclusion.

Nurses should remember that the artificial nature of this intelligence not only has significant potential benefit but also carries with it the risk of serious harm. As nurses interface with AI they should endeavor to understand the genesis of the knowledge: who created the initial algorithms, what sources of information fed into the system, what was the initial goal, what controls are in place, does it reflect top down or bottom up learning and what mediation is available if a dispute occurs. AI generated knowledge should never be accepted as absolute truth, any more than we accept research generated truth as absolute. As the future of AI unfolds nursing involvement in its development and implementation is paramount to prevent the advent of unwanted, machine driven events unmediated by ethical and humane oversight. ¹⁶

Algorithms

Algorithms are a procedure or formula for solving a problem based on a sequence of specified actions or calculations that translate input into output. ¹⁷ Its origins go back to the Persian mathematician, Mohammed ibn-Musa al-Khwarizmi (AD 780–850).

Algorithms shape knowledge because there is a fixed start; initial codes determine what will serve as the input. Algorithmic thinking is essential to comprehending how and why information technology systems work as they do. ¹⁸ Big Data and AI both depend on algorithmic direction. One common benefit of data-driven algorithms is the limitation of subjective inclinations by the end-user. Such limits are expected to reduce error and bias. But as previously noted, algorithms are developed by humans and as such carry with them biases built in at inception.

Algorithms are designed to recognize patterns, but not necessarily all of the subtle data points in individual encounters with the subject matter; to distinguish gross patterns but not meaning. ¹⁹ According to Danaher et al., ¹⁷ algorithmic translation includes two distinct questions; first, how to convert a task or problem into defined steps and second, how to convert the steps into computer code.It is within translation that additional issues arise as it involves human judgments and discernments.

As healthcare systems adopt more and more algorithmic decision support programs, knowledge generation runs the risk of intellectual reductionism with a drive to the mean, rather than expansion. If care is predicated on the average, what happens to the other, the different, the unusual, and the individualization of care? Will all patient care be viewed through the lens of the average or is there room for nurses to care for the uniqueness of the person? Nurses from the bedside to those in hospital administration need to be cognizant of these algorithmic issues otherwise we fall subject to Algocracy or rule by algorithms. ²⁰

Schuppli ²⁰ has argued that we also need to consider the legal ramifications of the actions that result from algorithmic code.Legal regulations have been developed by some countries to protect the right to privacy in the age of algorithms and Big Data; at its root is the protection of human dignity. ²⁰ While legal issues may be a point of concern, the impact of algorithms which have the potential to limit nursing actions and cause the loss of human dignity should be at the forefront. ²¹

Biology and biological modeling’s influence on knowledge development

The scientific process and its derivative, the nursing process, reflect classical mechanics. The appeal of a mechanistic world, of Newton’s clock, and Laplacian determinism, has held strong appeal: one of absolute predictability along with reductionism, determinism, and objective knowledge. Starting in the 1970s, a new view has slowly began to emerge as more and more scientists from disparate disciplines began to recognize that living systems, even plants, are fundamentally different from nonliving systems such as machines. New approaches have emerged to explain these differences. Today complexity theory, intuitive knowing, genomics, and neuroethics provide nursing and healthcare with both benefits as well as ethical challenges.

Complexity theory

Complexity theory has gained traction as an important approach to the study of dynamic adaptive systems such as humans. Complexity theory is the study of how complex systems, which are far from equilibrium, contain order, patterns, and structure yet appear chaotic. ²² The philosophical change from a mechanistic to a dynamic adaptive worldview changes the approach to scientific research from the questions posed to the analysis used.

All living systems are complex and as such are sensitive to initial conditions, unpredictable, dynamic, self-organizing, and scalable. ²³ Until recently, science attempted to predict behavior based upon individual components, missing at times the dynamic nature of interactions between the many components. While traditional science is still critical for questions of a linear nature, it is proving limiting for dynamic systems. With an understanding of complex dynamic systems behavioral predictability becomes moderated; not an absolute. That which was previously considered error thus can become normal pattern variation.

As Maziak ²⁴ writes without new approaches, such as complexity theory, epidemiology may have reached the boundary of understanding and achievement. He notes that the simple problems in epidemiology have been answered, and we are now left with the complex. ²⁴ If we do not address complex questions using appropriate methodology then the answers are suspect and could ultimately negatively affect patient care.

Thompson ²⁵ observed, “Complexity theory offers a perspective to studying complex systems in a manner that does not reduce the system to individual components” (p. 2), but rather considers how interactions between components create specific and unique outcomes. In healthcare, the individual components of disease are often studied separately but considered predictive of the trajectory of illness and treatment for all patients. However, not all patients respond to treatment as predicted.

Supercomputing, algorithms, nonlinear statistics, and AI facilitate the understanding of the complexity of human disease. For example, Multiple Organ Dysfunction Syndrome (MODS) represents a condition with features consistent with complexity theory: sensitivity to initial conditions, interdependent elements, and disproportionate response. ²⁶ Traditional ways of knowing might dismiss the difference in patient response to MODS protocols as error; complexity theory tells us that the so-called error could represent fundamental differences in understanding the syndrome’s trajectory because it is a complex response and not a linear one. Complexity theory also elucidates why seemingly small events in MODS and other diseases can result in catastrophic results in patients. ²⁶

Although subtle, there are ethical ramifications in the choices of research approaches. First of all, complexity may compel us to revisit the meaning of error. If error is not really error but system variation how should it be handled? If the question is one that includes dynamic movement and pattern then nonlinear statistics and design will provide the best answer. If instead linear statistics and traditional methodologies are used then a suboptimum answer may be obtained. If this information is then used to influence patient care there is a potential of an ethical problem. We propose that it is essential for nurse scientists to be cognizant of the nonlinear approaches to complex questions. More controversial, it may also be time to revisit the nursing process as it represents a limited, if not fading mechanistic worldview.

Rational and intuitive thinking

To process information in complex environments, it is necessary to “think about thinking” and how decisions are made based on self-reflection (meta-cognition). We have begun to understand that there is dual processing of problems, involving both intuitive and rational thinking. Intuitive or fast thinking represents the first impression of the problem while rational or slow thinking is the more considered analytical response. Fast thinking can provide the best or nearly best answer when time is of the essence. Rational or slow thinking can overcome errors in judgment from fast thinking, as it is logical and analytical. ²⁷ However, slow thinking takes more effort, and it is often difficult to discern when to use it with major or critical decisions. ²⁸

While logical, analytical thinking is well accepted in the scientific process and clinical decision-making, intuitive thinking is less well supported. Kahneman ²⁸ has lent considerable support to the value of intuition. He characterizes intuition as fast thinking that is automatic, often unconscious, and yet rich with experiential references that identify patterns in new experiences. These patterns are recognized quickly and decisions are made based on factors that include education, culture, cognitive ability, social learning, and previous exposure to similar situations.

Nursing has long recognized the presence of intuition ^{29 –31} but perhaps not its value. In the healthcare, setting slow thinking and fast thinking are both valuable, depending on the situation at hand and the experience of the practitioner. We realize that deliberative analysis is often perceived as the best way to problem solve, but we do not accept that only slow thinking is acceptable. We argue that it is critical to acknowledge both fast and slow thinking as important in clinical decision-making and not consider one more ethical than the other. Nursing research is needed to explore how and why dual processing thinking influences the ethical decisions of patient care.

Neuroethics

New developments in neuroethics provide another way of examining ethical decisions making. Neuroethics refers to the ethical interface of technology and neuroscience ³² inclusive of moral and ethical cognition and behavior, and its use and meanings in the clinical and social spheres. ^33,34 It is an

interdisciplinary research area that focuses on ethical issues raised by our increased and constantly improving understanding of the brain and our ability to monitor and influence it, as well as on ethical issues that emerge from our concomitant deepening understanding of the biological bases of agency and ethical decision-making. ³⁵

The ultimate goal of neuroscience is to gain a full understanding of the structure of the nervous system and the brain. ³⁷ Recent advances in neuroscience are giving us “unprecedented ways to understand the human brain and to predict, influence and even control it.” ³⁵ The merging of nanotechnology, biotechnology, information technology, and cognitive sciences (NBIC) involves augmentation of neuro-cognition and lends itself to the perception of enhanced human performance. ^38,39 In nursing, we need to be concerned that over reliance on neurotechnologic science may lead to self-deception and the override of free will. If no approaches are adopted to ensure that there is appropriate oversight of these technologies the threat to humanity is imminent. Nursing knowledge derived from these new approaches to neuroscience must be balanced against the real ethical issues involved.

Neuroethics can prove essential in dissecting the impending ethical implications of neuroscience as well as ethics itself. The neuroscience of ethical decision-making has been tied to dual processing thinking; functional magnetic resonance imagings (fMRIs) have shown different parts of the brain are responsible for quick responses versus the more deliberative. Work by Greene ^40,41 using the “Trolley Problem,” ^42,43 have shown that there is an instinctive response by most when faced with the decision to save one person or sacrifice one to save five. When the scenario is altered different responses result. When the problem is deliberated (slow thinking) the answers are generally different than the first (fast thinking) response.

The new neurosciences are in the future of healthcare. The ability to “improve” thinking, change personality, and to remove memories are on the horizon. With these enhancements comes the potential of profound ethical issues, including loss of autonomy, privacy, and questions of social justice. ³⁵ An understanding of neuroethics will give nurses the vocabulary to discuss these new developments. The voice of nurses in defense of patients must be heard; otherwise we are nothing more than cogs in the wheel of enterprise.

Genetics and genomics

To provide effective care today genomic knowledge is requisite. Genetics is the branch of science, which studies one gene at a time, whereas genomics refers to the study of the entire genome of an organism. ⁴⁴ Genomic developments that are changing the face of healthcare include prenatal and newborn screening, risk identification, screening and diagnosis, disease characterization, individualized therapy, management of symptoms, and end of life DNA retrieval. ⁴⁵ Discoveries such as mapping the human genome and illumination of genomic variation associated with health, disease, and management options are already being translated into nursing practice and no longer always dependent on referral to a genetic specialist. ⁴⁶

As advancements in genomics continue, more questions than answers exist. We are emerging into an era where translation of genomic knowledge to nursing practice is inevitable, and today there are many examples of its use in disease management. The rapid proliferation of knowledge and understanding of genomics makes it clear that understanding heritability and its intersection with the environment has become foundational to nursing science, theory, and practice. ⁴⁶ The current trends in healthcare increasingly demand that registered nurses (RNs) use genetic information and technology to obtain comprehensive family histories of their patients, provide guidance in making informed decisions about genetic testing, and assist at-risk patients to appropriate care and counseling.

While the benefits of genetic/genomic knowledge are known, nurses need to think through how genetic information is used and distributed. There is a real possibility in the future of genetic discrimination including but not limited to loss of insurability and employment for both the patient and family members. ^47,48

There are also profound ethical implications about the use of a current technique called CRISPR which allows DNA editing. ⁴⁹ Scientists are exploring how to edit genes to cure or prevent diseases such as cancer. These efforts appear inherently beneficent. However, the ability to edit genes could alter humanity in ways we do not understand and which may not be good for mankind; with the possible re-emergence of eugenics and the creation of a super class able to afford the cost of genetic alteration. ^47,48 This is not an issue of the future, it is of the present. World scientists have condemned the editing of the human genome, however, it may have already been attempted. ⁵⁰ All genomic research has the potential to threaten, at minimum, autonomy, beneficence, nonmaleficence, and distributive justice.

The distinction between the overlapping concepts of preventing harm and promoting good are sometimes difficult to distinguish in genomics and in fact can be present in the same instance. An understanding of genetics and genomics distinguishes nursing professionals as state of the art clinicians, researchers, and academicians who will insure the best ethical care possible. Genomic literacy has not kept pace with genomic advances, and filling this gap is urgent. We argue that all healthcare professionals must be appropriately prepared to integrate this knowledge into their practice. Perhaps more importantly nurses need to develop genomic knowledge consistent with our disciplinary view. Nurses as advocates should be vigilant; raising concerns about how and why patient DNA is being used. To not have this knowledge may be an ethical breach of our patient compact.

Summary

The positive influences of knowledge developed through large data sets; the quick efficient reduction of data into useable pieces; the possibility of reduction of redundant human tasks by machines, error reduction, pattern recognition, new understanding of decision-making, and so forth are apparent. However, all these new ways of understanding also require skepticism and critique from a profession that purports to care for and protect patients. We must produce knowledge that will guide decision-making and encourage critical thinking; using rather than relying on technological assistance. Wholesale acceptance by nursing of any knowledge without understanding and analysis is irresponsible at the least and ultimately can be unethical. Therefore, we urge the development of ways to critically analyze how, why, and when we use technology and biological models.

Questions about the new ways of knowing, including how we discern truth, remain unanswered and in some cases unasked. Increased reliance on technology may negatively inhibit the human thought process producing what is called “digital amnesia.” ⁵¹ We argue that digital amnesia coupled with noncritical acceptance of technology and biological models is very problematic to ethical care. There is no question that we are in the Fourth Industrial Age. Traditional approaches to education, practice, and research may not provide the knowledge, understanding, adaptability, and innovation needed today. We ask do we have the tools to function in this new age? If we do not then we cannot provide the highest level of ethical care.