Health data poverty: an assailable barrier to equitable digital health care

Abstract

Data-driven digital health technologies have the power to transform health care. If these tools could be sustainably delivered at scale, they might have the potential to provide everyone, everywhere, with equitable access to expert-level care, narrowing the global health and wellbeing gap. Conversely, it is highly possible that these transformative technologies could exacerbate existing health-care inequalities instead. In this Viewpoint, we describe the problem of health data poverty: the inability for individuals, groups, or populations to benefit from a discovery or innovation due to a scarcity of data that are adequately representative. We assert that health data poverty is a threat to global health that could prevent the benefits of data-driven digital health technologies from being more widely realised and might even lead to them causing harm. We argue that the time to act is now to avoid creating a digital health divide that exacerbates existing health-care inequalities and to ensure that no one is left behind in the digital era.

Characteristics of publicly available skin cancer image datasets: a systematic review

“…Health data poverty, defined as "the inability for individuals, groups, or populations to benefit from a discovery or innovation due to a scarcity of data that are adequately representative" is an emerging problem in digital health. 14,36 We found unequal geographical distribution of datasets and atlas origin, with 11 (79%) of 14 datasets and 15 (88%) of 17 atlases originating from Europe, Oceania, and North America exclusively. Only one dataset originated from Asia, two from South America, and none from Africa.…”

“…Strategies proposed to achieve this include increasing awareness of health data poverty within the machine learning and digital health com munities, improving dataset transparency, investment into routine prospective collection of digital data from health-care systems, and transparent, effective communication to increase inclusion of all population groups. 36 Our systematic review had several limitations. Although our search strategy incorporated searches from multiple sources, our search was limited to MEDLINE, Google, and Google Dataset Search, and our search terms were in English only.…”

“…75 Geographical and ethnic disparities in digital health data have similarly been identified in other areas of digital health care where certain populations are underrepresented by the amount of digital health data available. 14,36,63 Biased datasets can lead to digital solutions benefitting populations that have advanced data infrastructure that facilitates data collection, but can result in the exclusion or even harm of data-poor subpopulations. Skin cancer machine learning algorithms should be developed using inclusive and representative populations to reduce algorithmic biases.…”

“…Furthermore, the recently developed concept of health data povertysystematic data disparities leading to inequalities in health care-emphasises the need to ensure diversity, transparency, and usability of datasets. 14,36,37 Demographic and clinical metadata can also greatly influence machine learning algorithm development and validation, with classification accuracy increasing for skin lesions when subject and lesion metadata are integrated into models. 38,39 This more closely reflects clinical practice, as dermatologists conduct in-person history taking and evaluation of skin lesions via naked eye, dermoscopic, and physical examination (eg, lesion Review palpation), and not through image analysis alone.…”

Learning our lesson: using past policies to improve digital and ethnic inequalities beyond the pandemic

“…There is a well described gap in ethnicity data within healthcare and routine databases [ 7 ], despite statutory requirements for collection, analysis and reporting [ 6 ]. There have been increasing concerns that a lack of representativeness in clinical trial recruitment, the data science workforce, and in health data more generally, contributes to “health data poverty”, in which particular individuals, groups, and populations do not benefit equally, or may even be harmed, by a lack of representative data [ 8 ]. This was borne out early during the pandemic, with few publications reporting ethnicity disaggregated data, reflecting a lack of data collection [ 9 ].…”