In brief

QuestionCan an electrochemical biosensor recognize its environment before estimating concentration?
ResultThe pH classifier exceeded 85% accuracy across four environments from pH 3.6 to 8.0; pH-specific regressors estimated CEA over five orders of magnitude.
Why it mattersInstead of treating environmental variation as noise, the system identifies the environment and adapts its interpretation.
Adaptive Electrochemical Aptamer Sensing across Biofluid-Relevant pH Ranges via a Dual-Stage Machine Learning Framework
Wenjun He et al. · ACS Sensors · Published 9 August 2026
DOI: 10.1021/acssensors.6c01536

The electrochemical problem

Electrochemical biosensor signals are not controlled by analyte concentration alone. Changes in pH, ionic environment, adsorption, and matrix composition can shift peak positions and change peak shapes, causing a calibration curve built in one condition to fail in another.

This study asks whether the sensor can first infer an environmental variable from the electrochemical waveform and then use the appropriate concentration model.

What the researchers did

The model system uses ferrocene-labeled DNA–gold nanoparticle probes and differential pulse voltammetry. Shape descriptors including peak potential, prominence, width, and signal-to-noise ratio are used by a first-stage classifier to identify the pH regime.

A second stage then applies pH-specific regression models to estimate carcinoembryonic antigen concentration.

What they found

The first-stage classifier achieved greater than 85% accuracy across four pH-variable environments spanning pH 3.6 to 8.0. The second-stage models predicted CEA concentrations across five orders of magnitude.

The combined strategy adapts to distorted electrochemical profiles without adding a separate internal standard for every measurement.

Why this matters

The architecture is interesting because it explicitly separates context from concentration. In many sensors, environmental variation is treated as an unwanted disturbance to be averaged away; here it becomes a variable that the model identifies and uses.

That approach could be more robust when the major source of calibration drift is structured and classifiable rather than purely random.

One important caution

The pH environments in this study are controlled calibration media. Real biofluids vary simultaneously in ionic strength, protein content, fouling, temperature, and many other variables.

The paper presents the framework as a step toward complex-matrix and portable validation; those later validations are still necessary before treating the method as a general solution to biosensor drift.

Primary source

Wenjun He et al. Adaptive Electrochemical Aptamer Sensing across Biofluid-Relevant pH Ranges via a Dual-Stage Machine Learning Framework. ACS Sensors (2026). DOI: 10.1021/acssensors.6c01536.

Research-note facts and quantitative values were checked against the publisher record linked above. The “Why this matters” and “One important caution” sections are EchemAI editorial interpretation.