Skip to content

Citing publications sweep 2026-11: 17 new catch22 papers - #8

Merged
benfulcher merged 1 commit into
mainfrom
citations/2026-11
Nov 17, 2026
Merged

benfulcher merged 1 commit into
mainfrom
citations/2026-11

Conversation

@benfulcher

Copy link
Copy Markdown
Contributor

Monthly citation sweep for catch22 (run 2026-11-17). Discovery found 118 citing works not yet in the ledger, the first sweep since 1 October plus a backlog of repository and preprint copies that OpenAlex has only now linked. 110 were read in full (cached full text, arXiv, Europe PMC, or in the browser); 8 could not be reached.

Added to the page (verdict A): 17 bullets

Paper Section Evidence (quoted)
Köhler et al., Frontiers in Neurology (2026), doi Physiological and other sensors "Due to the limited sample size and subsequent risk to overfitting, we chose the catch22 framework"
Irani et al., Journal of Movement Disorders (2026), doi Physiological and other sensors "In addition, we incorporated time-series features from the CATCH22 feature set,19 a collection of 22 canonical time-series descriptors" (best model: "Dynamic (basic+CATCH22)", 90.1%)
Farrukh et al., arXiv (2026), doi Physiological and other sensors "we extracted Catch-22 (Canonical Time-series Characteristics) features from both EEG and pupillometry signals."
Pradeep et al., arXiv (2026), doi Physiological and other sensors "HRV features (meanNN, SDNN, RMSSD, SD1, SD2) and Catch22 features were extracted from the ECG"
Wijendran, University of Toronto thesis (2025), link Physiological and other sensors "Using Catch22 features extracted from EVG recordings as input to tree-based machine learning models, BP was predicted with high accuracy (<5 mmHg error)."
Landelle et al., Nature Communications (2026), doi Neuroimaging "the dynamic properties of the z-scored BOLD time-series were summarized using a set of features extracted with the Canonical Time-series Characteristics python toolbox, catch22"
Unterberg et al., Research Square (2026), doi Industry, energy and chemistry "the Python library catch22 [20] was used to extract features from each shearing phase due to its fast feature calculation and minimal between-feature correlation [21]."
Nikhal et al., arXiv (2025), doi Industry, energy and chemistry "In addition, we extract a set of canonical features (e.g., distributional characteristics, autocorrelation structure, and outlier dynamics), summarizing key statistical properties of each time series [21]."
Konopka et al., Science Advances (2026), doi Biology, physiology, neuroscience, pathology, and ecology "First, we used the catch22 software ( 36 ) to extract 24 features (F1 to F24)"
Kuppel et al., arXiv (2026), doi Biology, … "a feature vector computed from the catch22 statistical descriptors [20]"
Cebedio et al., Elektron (2026), doi Biology, … "se extraen características temporales mediante el extractor catch22 [17], obteniendo vectores de 22 atributos para cada fragmento."
Zhang et al., medRxiv (2026), doi Biology, … "The second set was drawn from the catch22 library [17], accessed via the Rcatch22 package"
Rojas et al., Scientific Reports (2026), doi General "The objective of this work is to investigate the conditions under which Catch22 is informative then only short and noisy time-series are available."
Thäns and Latoschik, arXiv (2026), doi General "We instantiate the timing-signature representation with 22 canonical time-series characteristics from catch22"
Owens et al., arXiv (2026), link General "We chose the catch24 feature set because it is concise, interpretable, computationally efficient"
Usova, University of Pisa thesis (2026) General "The main tool used here is catch22, a canonical set of 22 time series features selected for strong classification performance and low redundancy."
Chizaki et al., Japanese Journal of JSCE (2026), doi General "本研究では,catch22を汀線時系列に適用しUMAP-HDBSCANで変動パターンを分類する手法を提案する."

Worth a second look before merging:

  • Nikhal et al. is the least certain A: catch22 is cited ([21]) but never named in the text, and the features are clustered together with weekly usage profiles.
  • Chizaki et al. was judged from the J-STAGE abstract and title only (full text needs a subscription), and is marked 📕 closed access.
  • Kuppel et al.: the catch22 vector is one of three input modalities (it follows Hoßbach et al., already on the page).
  • Pradeep et al., Irani et al. and Zhang et al. combine catch22 with domain features rather than using it alone.
  • The two theses (Wijendran, Usova) had their venue labels hand-edited to "Thesis, University of …"; arXiv labels were shortened from "arXiv (Cornell University)". Section choices for Usova, Thäns and Chizaki ("General") are a guess.

Preprint → published

None: every preprint on the page is still a preprint.

Not added, and why

  • B (46): catch22 as one classifier or feature set in a comparison. E.g. Middlehurst et al. (multivariate TSC archive), Feofanov et al. (MantisV2, "Catch22+" baseline), Henderson & Fulcher (six feature sets via theft), Pacheco & Pedroso (IEEE Access), Brito et al. (tsMIST). Two borderline cases you may prefer on the page: Kassem (Aalto thesis; catch22 was the best of four feature libraries for activity recognition) and Thelin (Dalarna thesis; catch22 is one of the descriptor families in a telemetry-monitoring representation).
  • C (17): partial or auxiliary use. E.g. Kasab et al. (a few catch22 features in a large EEG feature set), Ailion et al. (catch22 waveform features beside network features for sEEG), Moutonnet et al. (catch22 to compare real and synthetic EEG), Weng et al. (a reduced subset as LLM context).
  • D (30): cited only.

Needs a human look (U)

🤖 Generated with Claude Code

Add 17 bullets across five sections and record 118 verdicts in the ledger.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
@benfulcher
benfulcher merged commit f889d4e into main Nov 17, 2026
4 checks passed
@benfulcher
benfulcher deleted the citations/2026-11 branch November 17, 2026 01:09
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant