Data ethics in Healthcare: Challenges and solutions

This report presents the main ethical problems that have been identified in the literature and identifies key solutions to ethical problems concerning health care data.

1. Introduction

From diagnostics to clinical decision making, the ability to process and analyze large amounts of data seems to be unlimited. Big data, machine learning and artificial intelligent (AI) have been having strong impact in healthcare (Kubitschke et al. 2010). For example, AI have been used to check drug allergy and deliver information about drug interactions and corollary disorders to guide the prescribing physician, resulting in a significant decrease in medication error rates. (Rainu et al., 2003). In addition, big data and machine learning enhance nursing care (Simpson et al. 2015, Westra et al. 2015). Patients themselves can self-manage of chronic conditions (Vayena et al. 2018). Clinical diagnostics, cancer detection, and robotic surgery are also exciting application from AI (Liu et al.2020). AI can also help to reduce missed appointments (Anom 2020). These applications depend heavily on health data. These data conventionally include test results from clinical laboratory, diagnostic images, medical records, public health registry data, and data produced in the context of biomedical or clinical research. In other words, all information about the health state of an identifiable individual (at any time) obtained through the analysis of health data. However, human data are sensitive data, not only when they come from humans as direct subjects of research, but also when doing small-area geography (Vayena et al. 2017). The following section approaches a few challenges and corresponding solutions.

2. The not exhaustive present challenges in collecting and using heath care data

The following section will discuss about ethical issues on the data collection and data sharing.

2.1.Data collection

Ethnicity data (the ethnicity of the patient populations) have been used to identify barriers to care based on ethnicity and tailor services to meet the needs of diverse ethnocultural groups (Varcoe et al. 2009). Collection of human data have been increased tremendously due to the growth of research activities. For example, collections of human biological material has been spectacular in the last two decades (Hirtzlin et al. 2003). In some cases, patients with a terminal illness want their information to be used to benefit others. It has been show that not terminally ill patients who are often willing to give their information for use in research and in research that does not benefit them but may help others (Wendler et al. 2008). However, collecting ethnicity data can cause judgements. These judgments arise from assumptions and stereotypes. For example, Aboriginal patients and visible minorities feel sometimes vulnerable to the effects of inequities. It has been showed that a variety of sociologic and demographic characteristics affected the treatment modalities. This bias against certain groups or beliefs about the health behaviors or conditions in these groups appeared to have discriminatory patterns of care and mistrust about the clinician’s advice (Schwartz et al. 1999, Grant et al. 2010, Kosenko et al. 2013)

Sorell et al. reported that passive data can improved patient privacy by allowing patients to receive treatment in their own homes, rather than in the hospital setting (Sorell et al. 2012). However, collecting of passive data by wearable are expected to raise express concerns, such as informational privacy and informed consent. These authors observed the difficulty in obtaining adequate informed consent through smartphones or smart devices. For example, most people do not read or understand consent forms. Because of the inherent nature of passive data, participants may not be aware of the type, amount, or implications of the data that is collected and gathered. In addition, people can not always foresee and predict the knowledge derived from passive data and the associated implications predicted. This leads to the loss of autonomy.

2.2. Secondary used of data- Sharing data

TranSMART is a well known open-source for knowledge management and data analytics platform, which uses open-source software to put the company’s preclinical and clinical trials data into the cloud and integrate those data across therapeutic regimes and stages of drug development (Scheufele et al. 2014). Today, this platform is administered by the tranSMART Foundation and is in use at no charge by 300 to 400 different governments, agencies, and nongovernmental organizations around the world. Despite the usefulness of such these platforms, ethical issues such as privacy and autonomy with the wider public benefits and data sharing have been risen. Although much of clinical data mining to date has used anonymous or de-identified data, there are ongoing active discussions about the application of electronic phenotyping, a process used to identify individual characteristics that might be useful for clinical prediction, management, and decision making (Herland et al. 2014). In other words, using advanced statically based methods to predict characteristics of a patient based on another patient’s de-identified data. Moreover, Klitzman et al. have pointed out, ethical issues arise when electronic health record data such as genetic testing results are available to determine differential access to life insurance (Klitzman et al. 2014). Another issue is related to secondary use of data. For instance, patients can contribute passive data to enhance their postoperative monitoring in the clinical research, whereas passive data used for research or public health purposes might not directly benefit the ones who are contributing. In addition, electronic data are being progressively used to identify individual characteristics, which can be helpful for clinical prediction and management, but were not previously to a health care professional . Recently, Parkerb et al. observed that there is no clear ethical framework has been established that stimulates passive data-driven innovation while protecting patient integrity (Parkerb et al. 2019).

Sharing data between medical doctors and other healthcare providers, such as nurses and laboratory technicians as they examine patients, can lead to reduce duplicate tests and yield efficiencies in patient care (Anom et al. 2020). Recently, data could be traded and marketed. When health care delivery systems become more global, protecting data becomes more difficult, complicated, and expensive. Security breaches tend to have more serious impacts. Data repositories represent only one mode of data sharing. Data posted in academic or researchers’ websites can provide data to requestors personally, making data accessible through publications (e.g. through supplementary files). Sharing data through well-curated online data repositories presents opportunities as well as challenges. For example, a distinct advantage is that online data repositories create a central “pool” of data and make the data easily discoverable for bona fide researchers worldwide to access and re-use. A corresponding challenge concerns questions about the appropriate governance mechanisms for data repositories, including questions about who will be able to access the data and what (if any) levels of restriction should be applied. Although there is a general ambition in the scientific community to strive for a model of Open Data sharing, ethical considerations sometimes call for access restrictions where human subject data is concerned, especially in the health and biomedical sciences (Merrett et al. 2018, Boulton et al. 2012.). A key consideration here is whether the data that is to be shared consists of aggregate research data or of individual participant data (IPD). The sharing of IPD, even if de-identified, may give rise to re-identification concerns in the context of big data. In contrast, the sharing of aggregate data would generally not disclose information about individuals and, hence, would be safer to share openly. However, aggregate research data does not always allow for full reproducibility of results and is less beneficial for future research use.

In healthcare research, data subjects (patients, volunteers) are either healthy or diseased volunteers who participated in clinical studies. Paticiapants can also donate samples and health information irrespective of their health status. However, the way in which these data subjects can exert control over their data, however, is mainly indirect. For example, data subjects should know how their data and information will be used, stored, distributed, and protected from unauthorized access through informed consent, which explain what the conditions of exposure for health data and personal health information in a are given study. Therefore, participants will be able to decide whether those conditions correspond to their expectations and best interests. However, there are different levels of consent forms. For example, participants who volunteer to provide data cannot know in advance who will have access to their data in the future and for which scientific purposes. In this context, volunteers generally sign broad consent forms, that do not specify the conditions of data exposure in any detail. In such a scenario, participants may experience a substantial lack of control over the flow of their data (Vayena et al. 2013).

3. Solutions

The people and organizations involved is responsible for the collection and the analysis of patient data. Many factors such as data ownership, privacy, security, and administrative issues appear to affect the collection process of patient data . Agencies and government should consider concerns about privacy. Regarding collecting of ethnicity data, the conflation of race and ethnicity must be raised for people involved in this process. The aim for collecting ethnicity data should be clearly articulated by healthcare organizations. Purpose must be communicated to both care providers and patients. Medical doctors are responsible to ensure that they understand how AI systems will be implemented in patient care. and they must obtain informed consent from their patients (Varcoe et al. 2009). Some researchers suggested that the data collected must be used solely for the purpose it was originally intended for. Others suggest that if the data is to be used for a secondary purpose informed, consent must be received again.

Patients must have the right to decide Who will own their data, Where that data must be stored, and What their data must be used for. Privacy concerns access to, and protection of patients’ information while autonomy has to do with patients’ control of their own data and treatment.

Data governance policies must be established to protect patients’ data (Anom et al. 2020). Policies must be sufficiently established to ensure that all patients have total control over their own data and trust that the data will not be used to harm or discriminate against them. Government and health care stakeholders must examine solutions that ensure security safeguards to protect patient data, patient privacy, and health information.

Supporting Information

Mittelstadt proposed nine ethical principles for the design of H-IoT devices and data protocols (Mittelstadt et al. 2017) :

1. Facilitate public health actions and user engagement with research via the H-IoT;

2. Non-maleficence and beneficence;

3. Respect autonomy and avoid subtle nudging of user behaviour;

4. Respect individual privacy;

5. Respect group privacy;

6. Embed inclusiveness and diversity in design;

7. Collect the minimal data required;

8. Establish and maintain trust and confidentiality between H-IoT users and providers;

9. Ensure data processing protocols are transparent and accountable.

References

• B.Y. Anom, Ethics of Big Data and artificial intelligence in medicine, Ethics, Medicine and Public Health 15 (2020) 100568.

• Geoffrey Boulton , Philip Campbell, Brian Collins, Peter Elias, Wendy Hall, Graeme T. Laurie, Onora O’Neill, et al. 2012. Science as an open enterprise. The Royal Society Science Policy Centre report 02/12. The Royal Society

• J. M. Grant, L. A. Mottet, , J. Tanis, J. Herman, J. Harrison, M. Keisling National transgender discrimination survey report on health and health care (2010) Washington, DC: National Center for Transgender Equality and National Gay and Lesbian Task Force.

• Matthew Herland, Taghi M Khoshgoftaar, and & Randall Wald, A review of data mining using big data in health informatics, Journal of Big Data 1 (2014).

• Isabelle Hirtzlin, Christine Dubreuil, Nathalie Préaubert, Jenny Duchier, Brigitte Jansen, Jürgen Simon, Paula Lobato de Faria, Anna Perez-Lezaun, Bert Visser, Garrath D Williams & Anne Cambon-Thomsen, . An empirical survey on biobanking of human genetic material and data in six EU countries, European Journal of Human Genetics 11 (2003) 475.

• Rainu Kaushal , Kaveh G Shojania, David W Bates, Effects of computerized physician order entry and clinical decision support systems on medication safety: a systematic review, Arch. Intern. Med. 163 (2003) 1409.

• Robert Klitzman, Paul S. Appelbaum, and Wendy Chung, Should life insurers have access to genetic test results? The Journal of the American Medical Association, 312 (2014) 1855.

• Kami Kosenko, Lance Rintamaki, Stephanie Raney, Kathleen Maness, Transgender patient perceptions of stigma in health care contexts, Medical Care 51 (2013) 819.

• Lutz Kubitschke, Sonja Müller, Karsten Gareis, Ursula Frenzel-Erkert, Felicitas Lull,

• ICT & Ageing: European Study on Users, Markets and Technologies–Final Report, European Commission: Brussels, Belgium, 2010.

• Rong Liu, YanRong, Zhehao Peng, A review of medical artificial intelligence, Global Health Journal 4 (2020) 42.

• Kirsty Merrett, Zosia Beckles, Stephen Gray, Debra Hiom, Kellie Snow, and Damian Steer. 2018. The administrative load of sharing sensitive data—challenges and solutions? Paper presented at the International Digital Curation Conference (IDCC), 19–22 February 2018, Barcelona, Spain.

• Brent Mittelstadt, Designing the Health-Related Internet of Things: Ethical Principles and Guidelines, Information 8 (2017) 77

• Michael Parkerb, Jukka-Pekka Onnelac, Annelien L. Bredenoordd, Timothy R. Smitha,, Marike L.D. Broekmana, , Passive data collection and use in healthcare: A systematic review of ethical Issues, International Journal of Medical Informatics 129 (2019) 242.

• Elisabeth Scheufele, Dina Aronzon, Robert Coopersmith, Michael T. McDuffie, Manish Kapoor, MS, Christopher A. Uhrich, Jean E. Avitabile, Jinlei Liu, MS, Dan Housman, and Matvey B. Palchuk, tranSMART: An Open Source Knowledge Management and High Content Data Analytics Platform, AMIA Jt Summits Transl Sci Proc. (2014) 96.

• L. M. Schwartz, S.Woloshin, H. G. Welch, Misunderstandings about the effects of race and sex on physicians’ referrals for cardiac catheterization. New England Journal Medicine, 341 (1999) 279.

• Roy L. Simpson, Big data and nursing knowledge, Nursing Administration Quarterly 39 (2015) 87.

• T. Sorell, H. Draper, Telecare, surveillance, and the welfare state, Am. J. Bioeth. AJOB., 12 (2012) 36.

• Colleen Varcoe, Annette J. Browne, Sabrina Wong, Victoria L. Smye, Harms and benefits: Collecting ethnicity data in a clinical context, Social Science & Medicine 68 (2009) 1659.

• Effy Vayena, Alessandro Blasimme, I. Glenn Cohen, Machine learning in medicine: Addressing ethical challenges, PLoS Med 15 (2018) e1002689.

• Effy Vayena , Alessandro Blasimme, Biomedical Big Data: New Models of Control Over Access, Use and Governance, Bioethical Inquiry 14 (2017) 501.

• David Wendler and Tammara Jenkins, Children’s and Their Parents’ Views on Facing Research Risks for the Benefit of Others, Arch Pediatr Adolesc Med 162 (2008) 9.

• B.L. Westra, L. Pruinelli, C.W. Delaney, Nursing knowledge:2015 big data science, Computers, Informatics, Nursing:CIN 33 (2015) 427.