Publications using catch22
A list of all scientific publications that have used catch22.
On this page, we keep track of scientific research publications that use catch22. Articles are labeled as follows:
π = Journal is open access.
π = Journal is not open access.
π = Freely accessible pre-print.
π» = Code/Data available.
If you have used catch22 in your published work, please contact us by email and we'll add it this growing list!
π₯οΈ Algorithms
catch22 features have formed the basis of new algorithms:

Combine the Matrix Profile with catch22 features, as C22MP, yielding a state-of-the-art anomaly detector. π Knowl Inf Sys (2024).

For developing semi-synthetic time series to understand algorithm performance.
π arXiv (2023).
"COCALITE: A Hybrid Model COmbining CAtch22 and LITE for Time Series Classification". πBadi et al. IEEE International Conference on Big Data (2024).
A toolbox for predictive analytics (feature extraction and selection) demonstrated on wind turbine bearing condition classification, hydraulic systems condition monitoring, CNC tool wear estimation, and respiratory health monitoring. π Zhou et al. Annual Symposium on Reliability and Maintainability (2025).
Applications
catch22 features have been used in applications to:
β Physiological and other sensors

Distinguish pain responses to hot and cold stimuli from electrodermal Activity and electromyography. π Ozek et al., Research Square (2024).

Classify hyperkinetic, tonic, and tonic-clonic seizures using unsupervised clustering of video signals.

Estimate pain intensity from physiological sensors.π arXiv (2023).

Classify human exercises using wearable sensors and video data.
π Joint European Conference on Machine Learning and Knowledge Discovery in Databases (2023).

Classify advertising engagement using affect and physiological signals of heart rate, electrodermal activity, pupil dilation, and skin temperature.

Automate general movements assessment for Cerebral Palsy from smartphone videos.
πPLOS Digital Health (2024).
π» Code.

Detect stress levels in real time from multiple physiological signals (heart rate, blood pressure, electrodermal activity, and respiration).

Predict behavioural change from physiological signals. π Sensors (2022).
𧬠Biology, physiology, neuroscience, pathology, and ecology

Predict pathological complete response in breast cancer from dynamic contrast-enhanced magnetic resonance images. π Breast Cancer Research (2024).

Classify calf behaviour from accelerometer signals.
π arXiv (2024).

Detect early signals of disease outbreaks from incidence data.
πPLOS Computational Biology (2025).
πarXiv (2024).

Distinguish chemical stimuli from C. elegans chemosensory system recordings.
π bioRxiv (2024).

Predict when an individual patient can switch from IV to oral antibiotic treatment from routinely collected clinical parameters from over 10,000 ICU stays.

Predict of cardiomyocytes differentiation outcome from oxygen consumption rate time series from human-induced pluripotent stem cells.

Evaluate similarity of synthetically generated peripheral nerve signals.
π 10th International IEEE/EMBS Conference on Neural Engineering (NER) (2021).
To quantify neural coding from MEG time series. π Maleki & Karimi-Rouzbahani. bioRxiv (2025).
To classify peptides from blockage current time series measured from a nanopore-based device. πHoΓbach et al., The Journal of Chemical Physics (2025). π arXiv (2024).
As a baseline for classifying marmoset monkey calls from audio. π Interspeech 2024 satellite event (2024).π Sarkar et al. arXiv (2024).
To identify methamphetamine users from EEG recordings. π Meynaghizadeh-Zargar et al., Biogeosciences (2024).
π Industry, energy and chemistry

Compare the influence of imputation strategies for classifying household devices from electricity usage.
π2023 Tenth International Conference on Social Networks Analysis, Management and Security (SNAMS).

Detect chemical analytes from chemiresistive hardware sensor arrays.
π arXiv (2023).

Appliance detection from very low-frequency smart meter time series.
πACM International Conference on Future Energy Systems (2023).

Detect anomalies in cloud services, where using catch22 resulted in the highest performance.
π IoT (2022).

Predict anode effects in aluminium production at least 1 min of advance from TRIMET Aluminium SE Essen (TAE) time-series data.

Classify vehicle trajectories from unmanned aerial vehicle-derived trajectories. πTeo et al. ISPRS Int. J. Geo-Inf. (2024).
Others:
Audio classification. π Marzano et al. (2025) arXiv.
Hard disk drive failure prediction. π Li et al. (2025) Engineering Applications of Artificial Intelligence.
Predicting different types of induction motor failure. π Li et al., IEEE Annual Symposium on Reliability and Maintainability (2025).
Analyzing appliance consumption by automatically detecting and classifying appliance activations in industrial kitchens. π Martins et al. Computers and Electrical Engineering (2025).
Sound Analysis of Drop Characteristics by Evaluation of Impact on Water Pool. Arogeti et al. πIEEE 2024 Signal Processing: Algorithms, Architectures, Arrangements, and Applications (2024).
π§ Neuroimaging

EEG classification using AutoML.
π Anais do Symposium on Knowledge Discovery, Mining and Learning (KDMiLe) (2023).
Infer rules underlying experience of subjective liking of artwork stimuli from EEG.
π PLOS ONE (2023).
Classify Multiple Sclerosis from full-field visual evoked potentials. π Banijamali et al., Doc Ophthalmol (2024).
Investigate the neural correlates of dreaming. π Wong et al. PsyArXiv (2025).
π Meteorology and astronomy

Detect earthquakes.
π arXiv (2022).
Predict solar flares from solar active region magnetic field data. π arXiv (2024).
ποΈ Finance and databases

As part of a meta-learning strategy for predicting market price movement.
π Anais do Symposium on Knowledge Discovery, Mining and Learning (KDMiLe) (2023).

Representations for time series in database management systems, such as Apache IoTDB, InfluxDB, OpenTSDB.

Analyze US market price data.
π arXiv (2023).

Capture meaningful properties of financial time series (performance is lower than using domain features).

Distinguish texts generated by AI (LLMs) from text authored by humans. πProceedings of the 19th Workshop on Innovative Use of NLP for Building Educational Applications (BEA 2024).
π General

To improve the efficiency and accuracy of time-series forecasting on target datasets using transfer learning.
π arXiv (2024).

Compare and cluster long time series.
π IEEE International Conference on Knowledge Graph (ICKG) (2022).

To understand dataset differences in evaluating foundation models for probabilistic time-series forecasting.

Evaluate time-series imputation methods. πProc. VLDB Endowment (2024).
Tutorials
Last updated

















