Abstract
Frequency-Domain Weak-Signal Analysis for Bias-Controlled Medical Response Assessment
Context: Weak-signal instrumentation proposed for medical response assessment re-quires clear separation between descriptive signal changes, subjective user response, and therapeutic efficacy claims.
Objective: To evaluate whether a document-informed custom signal acquisition device produces reproducible frequency-domain differences across ambient, blank control, and labeled test-sample recordings, and to frame a bias-controlled path for pilot user feedback.
Methods: Sixteen-bit binary recordings were processed using bias removal, non-overlapping 2048-sample framing, Hann-windowed Fourier analysis, perframe median normalization, difference spectra, and per-bin effect-size estimation. A ten-participant anonymous, non-clinical feasibility feedback summary was reported descriptively.
Results: Ambient-air comparisons showed setup-dependent narrowband changes near 30–33 kHz. Blank-control versus test-sample comparisons showed repeated peaks at normalized frequencies near 0.095, 0.119, 0.167, 0.238, 0.333, and 0.489 cycles/sample, with typical shifts of 6–8 dB and within-recording Cohen’s d around 1.0–1.35. The test sample-A versus test-sample-B comparison was smaller, with a maximum shift of about 2.18 dB and |d| ≤ 0.50. Seven of ten participants favored the signal-conditioned eyedrop sample.
Conclusions: The findings support a reproducible exploratory signal-analysis work-flow and preliminary anonymous feasibility feedback. They should not be interpreted as validation of therapeutic efficacy.
Author(s): Jui-Hung Lu1* and Hung-Chiang Cheng2,3
Abstract | Full-Text | PDF
Share this
