Nanomechanical mass measurements through feature-based time series clustering

Area of Science:

• Sensor Technology
• Data Analysis
• Spectrometry

Background:

• Miniaturized sensors generate time-series data with abrupt signal jumps.
• Nanoelectromechanical systems (NEMS) used in mass spectrometry exhibit finite-time frequency jumps due to analyte adsorption.
• This finite-time response causes ambiguity in detecting adsorption events, especially at high rates.

Purpose of the Study:

• To develop a computational algorithm for robustly eliminating ambiguity in NEMS sensor data.
• To automate the identification of single-event jumps in time-series sensor data.

Main Methods:

• A moving-window statistical test was employed.
• A feature-based clustering algorithm was utilized for jump identification.
• The method was validated using numerical simulations and experimental NEMS data.

Main Results:

• The developed algorithm successfully automates the identification of single-event jumps.
• Ambiguity in detecting adsorption events, even at high rates, was robustly eliminated.
• The method demonstrated practical applicability with experimental NEMS sensor data.

Conclusions:

• The computational algorithm effectively resolves ambiguity in NEMS sensor data analysis.
• This advancement enables new applications, including high-throughput, single-molecule proteomics.
• The automated identification of adsorption events enhances sensor data reliability and utility.