Figure 1
Recruitment Flowchart. Participant-organizers and thirty-five prospective participants met at the Quantified Self 2017 Global Conference to propose and discuss a participant-led lipid tracking project. Responders to a follow up survey confirmed their interest in participation and their goal for the project with an organizer via phone call. These individuals then received equipment from participant-organizers and attended online discussions to brainstorm risks and benefits of participation, and then to plan experiments. In total, twenty-one participants completed a project.
Table 1
Single Subject Hypotheses. This table lists the hypotheses, framed by individual participants, tested in single-subject, natural experiments during the study.
| Participant ID(s) | Hypothesis |
|---|---|
| 1–18 | Our lipids may vary significantly within-a-day. |
| 1–12 | Our lipids may vary significantly across mornings in the fasted state. |
| 1, 17 | My blood cholesterol and triglycerides may show ultradian and daily rhythms. |
| 2 | My lipids may cross a risk category within a day. |
| 3 | My post-prandial triglyceride rise may vary predictably based on the kind of food I eat. |
| 4 | My cholesterol and triglycerides may show ultradian rhythms that correlate with those in my electrogastrogram power or body temperature. |
| 5 | I can use my post-prandial triglyceride responses to create a “personal lipidemic index” comparable to a glycemic index of different foods. |
| 6, 7 | My subjectively and/or HRV-estimated stress may correlate with my cholesterol or triglyceride levels within a day. |
| 7 | Taking repeated multi-time-point “baselines” across different days may reveal stereotyped daily variability in my lipids. |
| 8 | Switching to a plant-based vegan diet may change my lipid levels within two weeks. |
| 9 | Natural variability in my lipids by time of morning may cause me to cross a risk category. |
| 10 | My daily fasting lipids, and 2-hour lipid profile may change in range or shape during very low, medium low, and moderate carb diets. |
| 11, 14 | Running may have a short term (before versus directly after a 30, 60 or 90-minute run) effect on my lipids. |
| 11 | A vegan diet may lower my total cholesterol and triglycerides over three months. |
| 12 | Tracking my lipids may be an effective encouragement for me to lose weight. |
| 13, 16 | Psychological and physical stressors (as measured subjectively and by HRV) may have distinct, measurable effects on my lipids. |
| 15 | My post-prandial triglyceride and cholesterol elevation may differ between days in which I eat three meals, and days in which I eat only one meal. |
| 17 | Changing the macro-nutrient composition of my diet for two-week increments may affect my post-prandial and daily fasted lipid levels. |
| 19 | I am interested in if my lipids and PT/INR (a measure of blood coagulation) co-vary, and if this influences the effectiveness of at home blood testing for me. Perhaps if I clot too quickly the test is ineffective. |
| 19 | I am interested in if my lipids change from before to after a) a long walk or b) a tai chi class. |
| 20 | My fasting lipids may vary predictably across my menstrual cycle. |
| 21 | I hypothesize that marathon training over two months will impact my cholesterol, and that my cholesterol may also differently from pre-to post run depending on run intensity. |
Figure 2
Within and Across Day Lipid Variability Traverses CVD Risk Categories. The daily and across morning ranges of TC, HDL-c, and triglycerides carry many individuals across CVD risk categories. Gradient indicates risk category, with redder as higher risk. Lines indicate risk category boundaries. Data shown are scatters by individual of TC, HDL-c and triglyceride values taken within a single day (A–C), and fasted within a single morning on different days, between 06:00 and 08:00 (D–F). Individuals are sorted from low to high range in TC, and this sorting is maintained in all plots. In total, 100% of individuals cross a risk category in as least one output at one or more time points within a day (A–C). 90% crossed at least one risk category in one output across days (D–F). Forty-seven percent of individuals cross at least one risk category in TC (A) and HDL-c (B) within a day. (C) Seventy-four percent of individuals cross at least one risk category for triglycerides. (D) Fasted between 6–8 am, 15% of individuals cross a TC risk category. (E–F) Thirty-six percent crossed an HDL-c or triglyceride risk category.
Figure 3
Median Total Cholesterol and Triglycerides Show Significant Daily rhythms. Median of all individuals’ percent of maximum TC (A) and triglycerides (B) by time of day show significant sine fit at approximately circadian periodicity, and significant change from fit peak to trough. Two-hour moving means of individual profiles are shown in color for ease of visualization. TC and triglycerides reached a maximum around 16:00, with minima in the early to mid-morning. The median percent change from maximum to minimum was 40% for TC and 35% for triglycerides. * indicate significant difference between peak and trough distributions (5:00–7:00, and 15:00–17:00 in TC and 9:00–11:00 and 16:00–18:00 in triglycerides) (Kruskal-Wallis p = 3.41*10–41 and p = 2.11*10–30). Sine fit statistics and plots of individual data and fits are available at our Open Science Foundation page.
Figure 4
Single-Subject Experiments Illustrate Triglyceride and Total Cholesterol Rhythms Across Timescales. Single subject recordings of (A) Ultradian fluctuation of blood triglycerides inversely corresponding to subjective hunger intensity, and (B) TC fluctuation across the menstrual cycle captured by morning fasted recordings.