LOESS smoothing: sample‑size requirements & span selection for annual‑resolution (18–88 yrs)

[potatolover222]

I’m working on a multi‑omics aging study and hoping to apply LOESS smoothing at yearly resolution (ages 18–88). I’m opening this issue to get advice on sample‑size needs and span tuning.

1. Minimum sample size for annual resolution
   In your experience, what is the minimum total n required so that a LOESS fit can reliably produce a smoothed curve at every 1‑year point over a 70‑year span?

2. Feasibility with ~100 samples
   We have ≈100 donors uniformly distributed across 18–88 yrs. Would that be sufficient to support LOESS smoothing with e.g. span ≈ 0.2–0.3 (i.e. 14–21‑sample neighborhoods)?

3. Smaller omics panels (~60 samples)
   For some layers (e.g. targeted proteomics or metabolite panels), only ~60 samples are available.
   In that scenario, would you recommend increasing span (larger neighborhood), or switching to an alternative approach to avoid edge artifacts and over‑fitting?

4. Span selection via leave‑one‑out CV
   Is using LOO cross‑validation to choose the optimal span generally sufficient to ensure the fit is both statistically robust?

Any guidance, rules of thumb, or example parameter settings would be greatly appreciated. Thank you for your time and for sharing your insights!

My sample ages
16, 17, 17, 18, 19, 19, 20, 20, 22, 23, 23, 23, 24, 25, 26, 26, 30, 31, 31, 32,
32, 33, 33, 34, 34, 35, 35, 35, 35, 35, 35, 36, 37, 38, 38, 39, 39, 39, 39, 39,
40, 41, 42, 42, 42, 42, 43, 44, 44, 44, 45, 49, 49, 49, 50, 50, 51, 51, 51, 53,
53, 53, 54, 54, 56, 56, 56, 57, 58, 58, 59, 59, 61, 61, 61, 62, 64, 64, 64, 64,
65, 66, 67, 68, 69, 69, 71, 71, 71, 72, 72, 73, 73, 75, 75, 76, 78, 81, 81, 82,
83, 84, 88