Magic Markers

A statistical debunk of a suicide-biomarker study (Niculescu et al 2015) hyped as 'a blood test predicts suicide with 92% accuracy.' Scott reads the actual results table and shows the biomarkers fail 7 of 8 tests (p 0.16-0.94), the one 'success' is suspicious (n=29, effects jump across diagnoses and don't survive a slightly different outcome measure), and the 92% comes entirely from the two clinical prediction instruments (asking people if they're depressed), not the genes. Section II: even if it worked, base-rate math means a positive result implies <50% actual suicide risk — not actionable — and suicide isn't purely biological anyway.

Quality 68 · Strong. 68 — low Strong. A careful, useful two-pronged debunk (the biomarkers-hidden-behind-clinical-instruments analysis + the base-rate/actionability point), a good methods lesson, but a single-study takedown that's somewhat in-the-weeds. Firm low-Strong.

Claude’s paradigm shift 48 · Moderate. 48 — Moderate. A single-study debunk with a useful base-rate frame; not novel.

Real-world impact 2 · Minor. 2 — minor/within-blog. Niche psychiatry/stats debunk; no material reach.