The year 2019 brought transformative changes in our daily lives thanks to the rapid advancement of big data technology. I think about information pushed to us via social media like Twitter, and products recommended to us by other social media forms based on our user behaviors. Similarly, transformative changes have been happening in the medical and health sciences such as medical imaging, artificial intelligence, auxiliary diagnosis systems, and diagnosis and prognosis prediction based on big data.
Encouraged and inspired by these successful changes, we as clinical health professionals and researchers in the cancer care community are urgently seeking to determine how we could make better use of existing and potential big data resources to explore the secrets behind data that could be the impetus to reshape the model of cancer care. First, however, is to make more explicit that data are the carriers of knowledge. This means that data should be able to pass information to others and generate new knowledge, have such ability is synonymous with our definition of "good data." Data without any information or knowledge is useless and would only bring interpretive burden to both health professionals and patients in oncology. It is easy for us to get lost in the world of data, especially in those "useless data" unless we have a high-level design and clear purpose to guide us in extracting what we really want from data and what data we need.
From data to knowledge, "big data" researchers usually follow the path of data integration, data mining, and data presentation. That is, to reshape the model of cancer care, the first step is about the integration of data. A good cancer data model is characterized by multiple data sources comprised of at least four types of voices including data reported by doctors, nurses, laboratory examinations, and patients themselves. Research has shown that algorithms based on mixed data sources are more accurate than a single database.1 Diverse data sources help to draw a whole picture of a human being reflecting physical, mental, and social health to realize "whole person" care, which is of great significance for the model of cancer care. However, gaps exist as commonly patients' voices are missing from such databases. To find a solution to this gap, our research team is trying to integrate several data sources including clinical medical records, patients' health-related lifestyle data from social media, and patient-reported outcomes data collected using the Patient-Reported Outcomes Measurement Information System (PROMIS)2 to develop a personalized smart recommendation system for symptom management in breast cancer patients.
An updated model of cancer care should be able to reveal the underlying knowledge behind data and achieve in-time personalized clinical decision-making. Improvements in various data mining technologies such as machine learning have made this possible. Similar to a medical imaging artificial intelligence auxiliary diagnosis system, within the smart model of cancer care, recommended nursing prescriptions could be given in real time following input of patients' key characteristics. Another application of data mining is precise classification of patients. For example, the Chinese government has been working on a precise Medicare pricing model with the help of big data-based diagnosis-related groups.3 This model divides patients by diagnosis, treatment, complications, and so on into over 500 000 groups to obtain reasonable and standardized Medicare pricing by backtracking and analyzing thousands of past cases to build a new Medicare model and, thus eventually, to optimize resource allocation.
In terms of data presentation, health professionals and patients expect a more user-friendly product that reduces the learning burden of users and makes the value of data to clinical care easy to maximize. Good solutions to the challenge of data presentation lie in easy to understand data visualizations and good products following person-centered design. We have seen many smartphone applications, wearable devices, even smart nursing robots, which could all be carriers of new knowledge generated from big data.
In summary, to welcome the year 2020 and a more advanced stage of big data era, we should focus on how could clinical practices and studies of cancer care contribute to making big data "good data", and more importantly, how could cancer care further benefit from big data to re-shape the model of cancer care. A good high-level design, a comprehensive integration of diverse data sources, advanced data mining techniques, and adequate forms of data presentation might help us find the secrets behind data. Inevitably, this hunt places higher requirements on nurses as we enrich ourselves with new knowledge and work closely with other disciplines to interpret and use the resulting knowledge. However, what should be kept in mind is the oncology nurse's role as a top designer in health data science from the perspective of nursing professionals wanting big data-based health care models to be well based in clinical problems and patients' needs, thus positioned to ultimately improve the efficiency and quality of cancer care.
Changrong Yuan, PhD, RN, FAAN
Editorial Board Member, Cancer Nursing
School of Nursing, Fudan University, Shanghai, China
References