This paper introduces a conceptual framework to guide research and education into the practice of personal science, which we define as using empirical methods to pursue personal health questions. Personal science consists of five activities: questioning, designing, observing, reasoning, and discovering. These activities are conceptual abstractions derived from review of self-tracking practices in the Quantified Self community. These practices have been enabled by digital tools to collect personal real-world data. Similarities and differences between personal science, citizen science and single subject (N-of-1) research in medicine are described. Finally, barriers that constrain widespread adoption of personal science and limit the potential benefits to individual well-being and clinical and public health discovery are briefly discussed, with perspectives for overcoming these barriers.

ObjectivesParticipant-led research (PLR) is a rapidly developing form of citizen science in which individuals can create personal and generalisable knowledge. Although PLR lacks a formal framework for ethical review, participants should not be excused from considering the ethical implications of their work. Therefore, a PLR cohort consisting of 24 self-trackers aimed to: (1) substitute research ethics board procedures with engagement in ethical reflection before and throughout the study and (2) draft principles to encourage further development of the governance and ethical review of PLR.MethodsA qualitative case study method was used to analyse the ethical reflection process. Participants discussed study risks, risk management strategies and benefits pre-project, during a series of weekly webinars, via individual meetings with the participant-organisers, and during semi-structured interviews at project completion. Themes arising from discussions and interviews were used to draft prospective principles to guide PLR.ResultsData control, aggregation and identifiability were the most common risks identified. These were addressed by a commitment to transparency among all participants and by establishing participant control via self-collection and self-management of data. Group discussions and resources (eg, assistance with experimental design and data analysis) were the most commonly referenced benefits of participation. Additional benefits included greater understanding of one’s physiology and greater ability to structure an experiment. Nine principles were constructed to encourage further development of ethical PLR practices. All participants expressed interest in participating in future PLR.ConclusionsProjects involving a small number of participants can sustain engagement in ethical reflection among participants and participant-organisers. PLR that prioritises transparency, participant control of data and ongoing risk-to-benefit evaluation is compatible with the principles that underlie traditional ethical review of health research, while being appropriate for a context in which citizen scientists play the central role.

No abstract

Cardiovascular disease risk assessment relies on single time-point measurement of risk factors. Although significant daily rhythmicity of some risk factors (e.g., blood pressure and blood glucose) suggests that carefully timed samples or biomarker timeseries could improve risk assessment, such rhythmicity in lipid risk factors is not well understood in free-living humans. As recent advances in at-home blood testing permit lipid data to be frequently and reliably self-collected during daily life, we hypothesized that total cholesterol, HDL-cholesterol or triglycerides would show significant time-of-day variability under everyday conditions. To address this hypothesis, we worked with data collected by 20 self-trackers during personal projects. The dataset consisted of 1,319 samples of total cholesterol, HDL-cholesterol and triglycerides, and comprised timeseries illustrating intra and inter-day variability. All individuals crossed at least one risk category in at least one output within a single day. 90% of fasted individuals (n = 12) crossed at least one risk category in one output during the morning hours alone (06:00–08:00) across days. Both individuals and the aggregated group show significant, rhythmic change by time of day in total cholesterol and triglycerides, but not HDL-cholesterol. Two individuals collected additional data sufficient to illustrate ultradian (hourly) fluctuation in triglycerides, and total cholesterol fluctuation across the menstrual cycle. Short-term variability of sufficient amplitude to affect diagnosis appears common. We conclude that cardiovascular risk assessment may be augmented via further research into the temporal dynamics of lipids. Some variability can be accounted for by a daily rhythm, but ultradian and menstrual rhythms likely contribute additional variance.

No abstract