It is unlikely to be news to anyone in the United States that we are living through a data revolution. As more and more of our routine tasks and activities move online or are undertaken via our phones, most people understand that their numerous small, daily actions, interactions, proximity and location all contribute data to systems that have the capacity to retain, combine, analyze and learn from them as never before. Such massive datasets provide the fuel that powers modern informatics: the increasingly sophisticated use of computers to understand, personalize, organize and predict human behavior and its results.
Compared to other industries, healthcare has been relatively slow to make use of the vast potential of informatics, but that gap is now closing rapidly. At universities across the US, health informatics is one of the fastest-growing disciplines. Tom Andriola is Vice Chancellor, Information, Technology and Data, and Chief Digital Officer, at the University of California, Irvine, and identifying how the institution can be strategic about data is an issue that concerns him daily.
“At UCI, we’ve taken a unique approach to how we think about data,” he explains. “It is more than an asset; it’s a way of thinking about how we can expand our mission for research, education, and patient care. How can we use data to facilitate the right people coming together for better patient insights? How do we enable nurses, clinicians, researchers, and data scientists to leverage the power of data and create an environment that allows transdisciplinary experts to work together for new scientific discovery, and increased quality of care?”
Quality of care was very much the motivation behind nurse researcher Jung In Park’s initial engagement with nursing informatics. Now an assistant professor at the Sue & Bill Gross School of Nursing, Park was originally an operating room nurse in a transplant unit in Seoul, South Korea. She gained her PhD in nursing informatics at the University of Minnesota, where her work concerned care factors affecting incidence of hospital-acquired Catheter Associated Urinary Tract Infection (CAUTI) – an extremely common problem among intensive care unit patients that can delay their discharge from hospital or cause their readmission soon after their return home.
Using data on nurse staffing – such as the percentage of registered nurses with specialty certifications, their education level, the skill mix, and total nursing hours – Park was able to identify that higher levels of specialty nursing certification among nursing patient-day, were associated with lower incidence of CAUTI. In addition to previously known factors, the implied knowledge of the better qualified nurses shone through the data.
Park developed a highly accurate predictive model that could identify a patient at risk of CAUTI, enabling hospitals to provide the necessary nursing interventions in advance. She explains that “CAUTI is a useful indicator of patient experience but it is just one measure. Similar studies could be used to identify factors affecting many different nursingsensitive outcomes.”
“I’m a nurse,” says Park, now a decade into her career as a nursing informatician, “and I have a nurse’s point of view. CAUTI is one of the concerns that nurses have, because it is associated with nursing care.”
This very first example of Park’s work demonstrates a key feature of nursing informatics: its potential to unpack the mystery of care. People who are, and who feel, cared for have – statistically speaking – better health outcomes and, ultimately, longer lives than those who do not. This phenomenon is widely acknowledged yet, despite extensive literature on the observable results, its causes and components have historically been seen as opaque. The intangible but appreciable benefits of “good care” remain hard to discuss empirically.
Randomized controlled trials, the gold standard of proof in healthcare, are rarely ethical or practical when it comes to testing what matters in terms of nursing care. Nurse researchers, whether arguing for change or seeking funding, often have to base their case on qualitative research. As Park’s work exemplifies, however, those opaque aspects of the care relationship are revealed in the vast datasets made available through modern electronic record keeping.
Park’s post-doctoral research at Stanford University involved developing a predictive model for the health outcomes of prostate cancer patients – specifically, to support patients and clinicians in making informed, individualized decisions as to whether to opt for treatment or active surveillance in the management of their cancer. Researchers such as Park test their initial analyses by creating new predictive models that are themselves measured against existing models and incoming data. Machine learning uses such comparisons and further incoming information to continue to refine the accuracy of the model. Because nurse informaticians are still rare, much of this work – if it is done at all – is currently designed by computer scientists. That’s why Park insists, “It’s time for nurses to jump in. I want to see nurses leading these efforts. Clinical problems don’t need binary answers! Nurses will be the ones using this technology in their jobs, and nurses see what really matters to people. That is absolutely missing in the development of AI unless nurses are involved.”
Introducing nurses to informatics comes with its own challenges, however, particularly in the early stages of training. As Andriola puts it, “people didn’t become a nurse to do a lot of advanced stuff with data. The challenge is how do you give them enough data literacy to help them to deliver the best care to the patient and practice at the top of their license.”
It’s an issue that UCI is tackling proactively and with energy. The close interdisciplinary tradition of the campus has translated naturally into opportunities to pair experts from health sciences and computer science at every level, from graduate researcher to professor. Park herself continues to work on predictive models for cancer, using data from the National Cancer Database, but is simultaneously involved in multiple collaborations with nursing colleagues, across disciplines and between organizations.
A key collaborator on campus is Amir Rahmani, a joint associate professor at the Sue & Bill Gross School of Nursing and the UCI Donald Bren School of Information and Computer Science. Rahmani, a computer engineering graduate from the University of Tehran in Iran, explains that he was “always interested in applying what I do to real world problems.” At UCI, he is a founding member of the Institute of Future Health, which opened in 2021. The IFH operates independently of any other school on campus, and exists specifically to promote interdisciplinary collaboration and learning between the schools of health sciences and of computer science and engineering. Faculty members of all these schools are members of the IFH, too.
Rahmani is involved in a dizzying number of healthcare and research projects in the US, but nonetheless continues to work on a long-term public health scheme run from the University of Turku in Finland, where he first became involved in nursing informatics. This project, founded in 2016, uses mobile technology to monitor the physiology and wellbeing of expectant mothers and provide appropriate individual support via public health nurses and community workers. With the Finnish project still underway, Rahmani came to UCI in 2018 and approached the school of nursing to find a collaborator, having secured National Science Foundation backing for a similar scheme for underserved pregnant mothers in California.
Rahmani is animated by the potential of informatics to spread the benefits of tailored, evidence-based healthcare to all people. He sees some of the current challenges to do with storing and using personal data melting away with each new generation of both patients and nurses. Although the anonymized data that researchers work with today is an unprecedented resource, he points out, there are many purposes for which ditching anonymity will make it more useful still, enabling researchers to create increasingly nuanced and personalized models.
“There are very stringent permissions in place for access to and use of personal data,” Rahmani explains, “but when people see the value, they give permission. And younger people aren’t bothered. For them, the information is already there, collated by their Apple watch.”
“Keeping people healthy is a business opportunity.”
Amir Rahmani
It’s clearly true that over the last decade, the practice of accessing personal health data has become a routine part of many people’s lives. Few of us who use a “wearable” – such as a Fitbit or Apple watch – throughout the day to help increase our step count, monitor our heart rate or record our sleep patterns would be concerned about sharing this information with a clinician to benefit our own health. After all, a desire to stay healthy is usually why we have a wearable in the first place.
The move to relinquish the adamantine requirement for privacy is thus likely to be driven by the benefit to the individual of access to completely personalized healthcare. In Andriola’s words, the future of health informatics is in supporting “whole person health” that is precisely tailored to the individual, from “deep, acute care, to struggles with chronic disease, to the healthy person who wants to stay that way.” Indeed, the rise of so-called “precision” healthcare has already begun.
A likely evolution, Rahmani suggests, is that “health records will be decentralized, and ownership will move to the user. You will own your own data, and healthcare providers will scan it to gain a full understanding of the person, even between visits.”
This vision of a longer, healthier life, based on disorder prevention and early intervention, has an almost utopian appeal. But what of the danger that the increasing sophistication of precision healthcare for the individual will also increase health inequality at the societal level? If the majority, with the right technology and access to healthcare services, are to be both owners of and participants in the system, will the lives and deaths of the disadvantaged minority slip increasingly from view? How will the data enable researchers to identify the needs of people who aren’t included in the first place?
In this, nurses have a critically important role to play. Health equity, the health effects of racism, environmental degradation and the experiences of underserved communities are central themes of nursing research, particularly at the Sue & Bill Gross School of Nursing. Again, it is nurses who are motivated to ensure that the right data is recorded, the right questions asked.
“I am concerned about representation bias,” says Park, whose most recent work has been to create a predictive model of cancer survival rates among Black and Hispanic patients, based on national data. A general model already existed, based on the same data, but when Park tested her model against it for Black and Hispanic patients specifically, “it outperformed that model.”
“There’s an expression in machine learning,” she says. “It’s, ‘Garbage in, garbage out.’ The accuracy of the data is very, very important. What has been included in the initial survey of participants? We know that income affects health outcomes; acculturative status, both real and perceived, does, too. And there are many other features of disadvantage that need to be considered and represented. There have to be nurses checking that the research has been properly constructed, looking into the data using a nurse’s lens.”
“How do we blend the worlds of technology and data science with the
Tom Andriola
complexities of healthcare?”
Andriola agrees. He sits on the board of a health technology company dedicated to health equity and underserved populations. “Health equity cannot be an afterthought. Nationally, it’s a top of the list issue,” he says. “With data, we can look at the differences across our patient population, not just in terms of their health but also in how easy it is for them to access the technology. Things we take for granted can’t be taken for granted when caring for underserved communities.”
As Rahmani points out, the potential of informatics opens up areas for commercial investment that are already proving beneficial in the field. “Keeping people healthy is a business opportunity,” he declares. “The pregnant mothers in our study are from underserved communities. And for example, applications related to mental health can now be prescribed – telehealth, CALM, PTSD counseling, stress management. It’s happening.” New sources of funding are always good news, particularly in a field in which, as Park puts it, “The data exists – I’m working at a national level already. There’s no limit except money. Eventually, I think informatics will be included in all nursing research that uses data. Nursing undergraduates are interested in new technology – they’re interested in the future.”
Andriola’s optimism is equally robust. “The question for someone in my job is how do we blend the two worlds – the worlds of technology and data science with the complexities of healthcare. We’ve seen a lot of effort happening with medicine. But now we are finally seeing the same energy to work with nurses and other health professionals. We’re figuring out that multidisciplinary approaches are possible, and we’ll get there.”
