Dr. Cristina Richie, Lecturer in Ethics of Technology, Department of Philosophy, Edinburgh University Future’s Institute, looks at the ethical considerations raised by the environmental impact of “digital twins”.
Technology has revolutionised health care. As medicine and science intersect, more health care adopts and uses technological innovations from other scientific disciplines. From gene-editing to robotic surgery, health care in most of the industrialised world is highly technological. One of the newer advances in health care is the use of “digital twins” (DT).
Essentially, DT use computer modelling to “provide digital representation of the equipment that can mimic properties and behaviours of a physical device.” In health care, the physical device is the person. Like a computer-generated model of a building that an architect may use to record and test design, DT capture health care information about a person. Koen Bruynseels, Filippo Santoni de Sio, and Jeroen van den Hovendefine medical digital twins as “an emerging technology that builds on in silico representations of an individual that dynamically reflect molecular status, physiological status and life style over time.” DT allow the users to run simulations on how the product/ person will perform; DT store and maintain data on the equipment and tests, and DT offer preliminary results on how the model/ prescription plan/ biodata will perform based on algorithms. When DT are translated to health care, the doctor becomes the “human user” and the patient the “model.” DT rely on medical tracking through the use of medical devices that transmit biodata to computers. Visual images of DT in health care can be found in a number of journals.
While DT may be used over the life course of an individual, they may also be a time-targeted tool. In the latter, DT would capture relevant health care information about individuals—for instance, glucose levels for diabetics, and can record real-time bioinformation during a period of medical risk. DT can assist predictive oncology and can track “the whole human body, one body system or body function (e.g., digestive system), one body organ (e.g., stomach or liver), one cell of a given type, or even simply some specific subcellular (organelle/sub-organelle) or molecular level of interest within a cell.” DT offer the potential for better precision and personalised health care, but also “precision public health” through aggregated data on patient populations. If, for instance, Hispanic males age 30-60 in Arizona with high cholesterol agree to digital tracking and monitoring, then, perhaps other groups of Hispanic males can benefit from the bioinformation, if it is shared and accessible to clinical researchers. In this way, the more people who use these technologies, the better it is for those who must wait for access to better health care and roll-out periods of DT services.
Digital twins are exciting to many in health care who see the use of technology as beneficial in servicing patients, facilitating precision medicine, and maintaining personal health. However, they also raise a number of ethical concerns about accessibility, cost, privacy, and benefit to user. DT are a sophisticated technology that is not globally accessible and deepens growing divides in medical access. The technologies which interface with DT—like personal wearable sensors and software on phones—may be cost prohibitive if they must be purchased by the individual (e.g., Apple Watch) instead of given as part of subsidised health care (e.g., at home cardiac monitors).
Privacy is not only compromised, but totally lost, when biodata is used for aggregate population predictions, even when it is voluntarily shared. Biodata—like all data—runs the risk of being datamined by hackers and may include sensitive personal information like geographical movement and financial transactions. Moreover, DT may not benefit the individual using them, unless they are tied to a clinical concern; people enjoy health technologies like Fit Bits simply out of curiosity. Even when DT are employed to assist patient treatment and prevention plans, they are not totally accurate and can lead to misdiagnosis. As these ethical concerns have been somewhat addressed in literature,attention needs to be paid to the ecological implications of use and dissemination of DT.
Personalised health monitoring, maintenance of electronic records, and stored patient data banks all have a carbon footprint. Although lifecycle assessments (LCA) of individual medical procedures—whereby the total carbon footprint of an item is calculated—are becoming more common, most health care services have not been a calculated. Therefore, best estimations between DT infrastructure and parallel services that have a LCA attached to them—or life cycle thinking—is one approach to understanding the possible environmental footprint of DT. For instance, one outpatient appointment emits 50 kg of CO2 equivalent; DT would add to these carbon emissions when they are used in both outpatient and inpatient services.
Moreover, DT which relies on artificial intelligence—for triage algorithmsor predictive artificial intelligence (AI) for health care disease—must go through programming, running, and training. By way of estimation, 40 days of training Google’s AlphaGo Zero game had the carbon impact of 1,000 hours of air travel. While DT AI infrastructure has not been calculated or predicted, simply the knowledge that AI is carbon intensive is an environmental consideration, in addition to other well-known environmental aspects of AI and technological use, like the carbon associated with extraction of minerals, metals, and plastics necessary for AI capable hardware.
Additionally, many of the necessary minerals are mined in conflict areas and mining often takes place under poor labour conditions. In “Anatomy of an AI System,” Kate Crawford and Vladan Joler tracked the environmental and labor resources required to develop, produce, maintain, and dispose of an Amazon Echo, illuminating the far reaching impact of technology. Fossil fuel use, mineral mining for chips, exploitative human labor, and the significant waste produced by consumer gadgets designed for planned obsolescence are often at work in medical technologies, undermining the ethics of their development and production even before they are in use.
Beyond the carbon from in/outpatient use and the carbon from AI in DT, one might also look at parallels with telemedicine, which DT uses when communicating with personal monitoring devices, operationalising patient consultations, allocating patient flow to practitioners, and securing patient data. In 2012, PLoS One recorded that the carbon cost of 238 telemedicine appointments in Sweden was 602 kg CO2 with a range of 1.86–8.43 kg CO2 per 1‐hr telemedicine appointment.
As digital twins evolve, as well as an awareness of carbon emissions and calculation of carbon emissions of health care, more data will be available with which to determine just how carbon intensive DTs are. Even so, the metric of carbon emissions is ethically and scientifically problematic and using CO2 as a proxy for environmental sustainability is insufficient, as safe amounts of carbon has been exceeded.
Moreover, a carbon number assigned to a procedure does not account for other ethical aspects of health care, carbon emissions, or health care delivery like distributive justice and is therefore morally reductionistic. Hence, principles for sustainability, ecological wisdom (reduce; reuse; recycle), and individual action are likely to be needed in making DT more sustainable as society awaits more carbon data.
No comments:
Post a Comment