Big data and health

Abstract

The digital health revolution is here. Innovations include not only the collection and analysis of electronic health records and personal genomes, but also diverse physiological and molecular measurements in individuals at a level that has not previously been possible. Our recent studies,1Zhou WY Sailani MR Contrepois K et al.Longitudinal multi-omics of host–microbe dynamics in prediabetes.Nature. 2019; 569: 663-671Crossref PubMed Scopus (237) Google Scholar, 2Schüssler-Fiorenza Rose SM Contrepois K Moneghetti KJ et al.A longitudinal big data approach for precision health.Nat Med. 2019; 25: 792-804Crossref PubMed Scopus (198) Google Scholar in which we deep-profiled 109 people for a median of nearly 3 years and made 49 major health discoveries (67 if hypertension is included) affecting 53 people, shows the value of big data and active monitoring. Many health discoveries involved disease risk prediction from genome sequencing, but most involved early detection of disease before symptom onset. Many of these findings were highly impactful, such as early detection of cardiomyopathy, lymphoma, and two precancerous conditions (monoclonal gammopathy of undetermined significance and smoldering myeloma).2Schüssler-Fiorenza Rose SM Contrepois K Moneghetti KJ et al.A longitudinal big data approach for precision health.Nat Med. 2019; 25: 792-804Crossref PubMed Scopus (198) Google Scholar However, the health data revolution is just beginning as more direct-to-consumer devices that measure health information become available (eg, electrocardiogram monitors for smart watches and blood pressure cuffs),3Steinhubl SR Muse ED Topol EJ The emerging field of mobile health.Sci Transl Med. 2015; 7283rv3Crossref PubMed Scopus (409) Google Scholar, 4Li X Dunn J Salins D et al.Digital health: tracking physiomes and activity using wearable biosensors reveals useful health-related information.PLoS Biol. 2017; 15e2001402Crossref PubMed Scopus (220) Google Scholar, 5Dunn J Runge R Snyder M Wearables and the medical revolution.Per Med. 2018; 15: 429-448Crossref PubMed Scopus (218) Google Scholar omics measurements become more sensitive and facile, and new imaging technologies emerge. How effectively society as a whole is able to capitalise on big data in health care will depend, in part, on how well these data are integrated and communicated to clinicians and the public. It will be important to have digital health information collected in a format that clinicians and consumers can easily interpret and query. Much has been written about the shortcomings of electronic health record systems.6Zulman DM Shah NH Verghese A Evolutionary pressures on the electronic health record caring for complexity.JAMA. 2016; 316: 923-924Crossref PubMed Scopus (48) Google Scholar Improvements in such systems for clinicians' purposes and a new commitment to sharing personal digital health information with consumers are needed. Currently, much of health information is organised around systems—eg, cardiovascular, gastrointestinal, and neurological. However, we expect that health information can also be organised according to underlying human molecular biology, and this might lead to novel health management approaches. For example, much as molecular information is grouped into networks based on interaction (physical or other) and co-expression data, digital health data might be grouped into entirely new interactive entities that might enable more efficient diagnoses and help to optimise disease management strategies (figure). The organisation of health information into interactive clusters and other novel methods for stratifying health data will complement existing approaches and potentially lead to improvements in health care. Tumour agnostic therapies in oncology represent early evidence of how the health-care paradigm might shift as health data management matures. We expect that there will be an important role for artificial intelligence in health data organisation and interpretation, with the caveat that clinical applications will need to be subject to rigorous validation procedures.7Beam AL Kohane IS Translating artificial intelligence into clinical care.JAMA. 2016; 316: 2368-2369Crossref PubMed Scopus (112) Google Scholar In addition, with the emergence of more longitudinal digital health information, it will be increasingly feasible to track nuanced changes in health conditions. New visualisation strategies for following health over time and disease progression will be developed. Health and disease monitoring will be increasingly personalised based on an individual's risk factors, personal molecular profile (eg, omics), imaging, and other health data.8Swan M Health 2050: the realization of personalized medicine through crowdsourcing, the quantified self, and the participatory biocitizen.J Pers Med. 2012; 2: 93-118Crossref PubMed Scopus (268) Google Scholar, 9Chen R Mias GI Li-Pook-Than J et al.Personal omics profiling reveals dynamic molecular and medical phenotypes.Cell. 2012; 148: 1293-1307Summary Full Text Full Text PDF PubMed Scopus (907) Google Scholar The construction of personalised health models along with personal health dashboards will be crucial for individualised health monitoring and management (figure). Although many have embraced the so-called data gold rush in biomedicine, there are others who have raised concerns. Some critics argue that there is too much data and that most of it is irrelevant. Moreover, there is concern that health information will overload the medical system—eg, more information will lead to more unnecessary follow-up tests. These are important ongoing concerns; however, we believe they should not serve as barriers to pursuing the potential benefits of big data in health. To address the challenges of the data deluge, our deep-profiling studies,1Zhou WY Sailani MR Contrepois K et al.Longitudinal multi-omics of host–microbe dynamics in prediabetes.Nature. 2019; 569: 663-671Crossref PubMed Scopus (237) Google Scholar, 2Schüssler-Fiorenza Rose SM Contrepois K Moneghetti KJ et al.A longitudinal big data approach for precision health.Nat Med. 2019; 25: 792-804Crossref PubMed Scopus (198) Google Scholar as well as a similar study by Price and colleagues,10Price ND Magis AT Earls JC et al.A wellness study of 108 individuals using personal, dense, dynamic data clouds.Nat Biotechnol. 2017; 35: 747-756Crossref PubMed Scopus (238) Google Scholar are intended to test feasibility of information collection and determine what types of information might be most useful. These research studies might also be useful for discovering new potential biomarkers, new molecular pathways that co-associate, and previously undiscovered changes that occur upon disease onset. Indeed, we found a candidate marker, the cytokine MIG1, which tracked with lymphoma onset and disappeared after treatment. Only some of the information collected in these studies is relevant to health monitoring, particularly in the study by Price and colleagues. However, the studies provided valuable insights into what types of information are expected to be most useful for risk prediction and health monitoring, from a combined technical, economic, and scientific perspective. These insights will be valuable as these technologies move into mainstream health care and the big data approach is commercialised. Smart and relevant data, not just data for data's sake, will be crucial for health-care adoption. Another major concern related to the advent of big data in health care is data privacy. Many people worry that these data will be acquired by organisations that will use the data to discriminate against them. Currently, in the USA, it illegal to discriminate in employment and health insurance on the basis of genetic information, but such information can be considered in long-term disability and life insurance contracts. Moreover, there is currently debate about protecting the ability of people with pre-existing health conditions to obtain affordable health insurance coverage. Because one of the goals of precision health is to identify risk early to better manage health, ensuring affordable health care for people with pre-existing conditions is crucial. An important aspect of these precision health technologies, especially wearables and genome sequencing, is that they can provide valuable information about individuals when they are healthy. They can help to define health for an individual and provide actionable information related to reducing an individual's personal disease risks.9Chen R Mias GI Li-Pook-Than J et al.Personal omics profiling reveals dynamic molecular and medical phenotypes.Cell. 2012; 148: 1293-1307Summary Full Text Full Text PDF PubMed Scopus (907) Google Scholar These technologies permit a greater focus on health preservation, rather than the current focus on disease diagnosis, monitoring, and treatment. Furthermore, the highly personalised health information provided can motivate individuals to take the steps needed to help preserve their own health, as demonstrated in our study.1Zhou WY Sailani MR Contrepois K et al.Longitudinal multi-omics of host–microbe dynamics in prediabetes.Nature. 2019; 569: 663-671Crossref PubMed Scopus (237) Google Scholar, 2Schüssler-Fiorenza Rose SM Contrepois K Moneghetti KJ et al.A longitudinal big data approach for precision health.Nat Med. 2019; 25: 792-804Crossref PubMed Scopus (198) Google Scholar, 4Li X Dunn J Salins D et al.Digital health: tracking physiomes and activity using wearable biosensors reveals useful health-related information.PLoS Biol. 2017; 15e2001402Crossref PubMed Scopus (220) Google Scholar, 9Chen R Mias GI Li-Pook-Than J et al.Personal omics profiling reveals dynamic molecular and medical phenotypes.Cell. 2012; 148: 1293-1307Summary Full Text Full Text PDF PubMed Scopus (907) Google Scholar When a health-care system is set up primarily to finance providers of downstream services such as disease or injury diagnosis and treatment, rather than on health preservation, the full societal benefit of precision health technologies might not be realised. We need to find creative ways to incentivise broad adoption of precision health, which will require participation from insurers, health-care providers, employers, and consumers. Transformational change also requires evidence, which is rapidly accumulating. Robust evidence of the health benefits of big data together with financial incentives will make precision health a reality. MS reports grants from the US National Institutes of Health. MS is a cofounder and scientific adviser for Personalis, SensOmics, Qbio, January, Mirvie, Filtricine and Protos. MS is an adviser for Jupiter and Genapsys.