Abhay Chakra Sadineni

Research paper · November 2026

Spectral Passport

Abhay Chakra Sadineni

Machine learningConformal predictionRaman spectroscopyBiophotonics

Abstract

A Raman calibration model that comes with a conformal prediction interval carries a promise that, on new samples, the true concentration falls inside the interval at a stated rate. The promise rests on the new samples being exchangeable with the calibration samples. This paper tests what happens to it when the spectrometer changes. It uses a public collection of 2,261 Raman spectra of glucose, acetate and magnesium sulfate mixtures measured on eight commercial instruments, 45 of the mixtures on all eight. For each instrument it trains a partial least squares model, calibrates a split conformal interval at 90 percent nominal coverage using one spectrum per mixture, and applies the result on each of the other seven. Across the 168 cross-instrument routes (56 instrument pairs, three analytes), median empirical coverage falls to 0.47, half the routes cover fewer than half of the test spectra, and fewer than one in five reach 0.85. A simple classifier separates every pair of instruments perfectly from a single preprocessed spectrum, so the change is never subtle. Because the instruments' spectra do not overlap at all, weighted conformal prediction, which needs overlap, returns intervals of infinite width for every test sample. Recalibrating with ten labelled spectra from ten distinct mixtures on the target instrument restores coverage on most routes (median 0.93, minimum 0.68) at a cost, since the median interval becomes 1.9 times wider than the full concentration range, and only one route in seven yields an interval narrower than that range. The paper proposes a decision rule, accept, recalibrate or abstain, and a portable record, the Spectral Passport, that binds a prediction to the instrument, the calibration evidence and the decision taken.

Cite

Abhay Chakra Sadineni (2026). Spectral Passport. Research paper. https://doi.org/10.5281/zenodo.24210927 https://www.abhaychakra.com/papers/spectral-passport/