Izzy Chaiken

—

Unclaimed · AI-structured profileOpenAlex

Clinical NLP researcher applying LLM-extracted features from respiratory therapy notes to extubation-failure prediction

papers here

reviewed by Evolve

first-author

corresponding

—

works · OpenAlex

—

h-index · OpenAlex

Bibliometric figures come from OpenAlex (2026-09-21) and cover the author's entire public record, not just the papers listed here. Reproducibility signal comes from Evolve's own reviews.

Research profile

Output—

Unscored: no publication-volume data is available beyond the single reviewed 2026 arXiv paper. This axis would measure the count, recency, and mix of outputs — journal articles, preprints, datasets — and remains empty until more of the record is indexed.

Impact—

Unscored: no citation counts, h-index, or reception data were provided. This axis would measure how the 2026 extubation-failure study and future work are cited and taken up; with one fresh preprint, no meaningful impact metric exists yet.

Focus100/100

This axis measures how tightly a researcher's output clusters within fields. The score is the maximum, 100, because the field distribution contains exactly one entry — health informatics (clinical NLP and machine learning) — covering the sole reviewed paper, the 2026 extubation-failure study. Perfect concentration is accurate here but rests on a single first-author arXiv record, so it describes a deliberately narrow agenda rather than a proven long-term specialization.

Collaboration32/100

This axis reflects the breadth of the co-author network. The score is 32, computed from just 4 distinct coauthors in the bibliometric data; the recurring set on the 2026 extubation-failure paper comprises Aditya Khowal, Neha A. Sathe, Mark M. Wurfel, and Lucy Lu Wang. A compact team built around one single-center clinical study explains the modest score, which is typical of an early-stage record rather than a signal of closed working habits.

Activity—

Unscored: the data covers a single publication year (2026), with no multi-year cadence, submission rhythm, or volume trend. This axis would measure ongoing research output over time; one data point cannot indicate whether activity is rising, steady, or paused.

Reproducibility55/100

The score is 55, the mean reproducibility reading across exactly 1 reviewed paper: the 2026 extubation-failure study, whose Evolve assessment tags reproducibility readiness as medium. The same review notes the ~+2 AUROC improvement is reported without significance tests, which tempers confidence in the headline gain. With one paper behind the number, treat 55 as an early signal about method reporting rather than a stable, demonstrated reproducibility habit.

Research summary

Izzy Chaiken is a health-informatics researcher whose reviewed work sits in clinical NLP and machine learning applied to intensive care. Evolve's catalog contains one entry: a 2026 first-author arXiv preprint, 'Enhancing Extubation Failure Prediction with LLM-Derived Features from Respiratory Therapy Clinical Notes.' The study applies large language models to extract structured features from respiratory therapy notes within electronic health records and uses them to predict extubation failure in mechanically ventilated patients. Evolve's review characterizes it as a single-center study in which LLM-derived features add roughly +2 AUROC, while flagging that the gains lack significance testing and fade over time — a genuine but preliminary contribution to ventilator-liberation analytics. The bibliometric footprint is narrow: the field distribution maps entirely to one domain, and the network includes only 4 distinct coauthors, with Aditya Khowal, Neha A. Sathe, Mark M. Wurfel, and Lucy Lu Wang recurring. Radar scores capture this shape: focus 100, collaboration 32, and reproducibility 55 from a single medium-readiness paper, while output, impact, and activity remain unscored for lack of data. Read as a whole, this is an early-stage profile whose trajectory will depend on whether the extubation line of work matures into validated, multi-center research.

Collaboration pattern

The bibliometric record shows a compact network: 4 distinct coauthors, giving a collaboration score of 32. The recurring group on the reviewed paper comprises Aditya Khowal, Neha A. Sathe, Mark M. Wurfel, and Lucy Lu Wang. Single-center clinical NLP work of this kind typically depends on close clinician-methodologist pairing, so the small team is consistent with an early-stage record; expect the network to widen if multi-center validation follows.

Research keywords

Extubation failureMechanical ventilationLarge language modelsClinical notesInformation extractionElectronic health recordsRespiratory therapy documentationLLM-derived featuresClinical prediction modelsTemporal generalizationAUROC evaluation

Highlights

LLM-extracted respiratory-therapy note features add ~+2 AUROC to extubation-failure prediction
Evolve's review of this single-center study reports that features large language models extract from respiratory therapy clinical notes add roughly +2 AUROC to extubation-failure prediction in mechanically ventilated patients, suggesting free-text EHR documentation carries weaning-relevant signal.
Reported gains lack significance tests and fade temporally
The same review flags two core limitations: the ~+2 AUROC improvement is reported without significance testing, and the benefit degrades over time — a caution for anyone deploying these features at the bedside or building directly on the result.
First-author arXiv study pairs LLM information extraction with ICU prediction
The 2026 preprint is a first-author, non-corresponding entry in Evolve's catalog, classified under health informatics (clinical NLP and machine learning) with medium reproducibility readiness — the sole paper anchoring this profile.

Timeline

2026First-authored the arXiv preprint 'Enhancing Extubation Failure Prediction with LLM-Derived Features from Respiratory Therapy Clinical Notes' (non-corresponding).
2026The preprint entered Evolve's reviewed catalog, assessed as a single-center study where LLM-derived features add roughly +2 AUROC.
2026Worked with a compact collaborator set — Aditya Khowal, Neha A. Sathe, Mark M. Wurfel, and Lucy Lu Wang — yielding 4 distinct coauthors in the bibliometric record.

Papers in the Evolve catalog

2026-09-17first author
Enhancing Extubation Failure Prediction with LLM-Derived Features from Respiratory Therapy Clinical Notes

Frequent co-authors

Aditya Khowal×1 Neha A. Sathe×1 Mark M. Wurfel×1 Lucy Lu Wang×1

About this profile

This profile is AI-structured by Evolve.Pub from public bibliographic data (OpenAlex) and Evolve's own paper reviews. It is not written or endorsed by the researcher unless the profile is claimed. Radar dimensions are computed from real bibliometrics with fixed formulas — raw figures are shown alongside; missing dimensions mean insufficient data, never zero. Errors? Claim the profile to correct or hide it.

Bibliometric scores are computed from public bibliometric data only; no institution, country, or ORCID was provided.
The name 'Izzy Chaiken' is resolved by name matching without an ORCID, so identity has not been independently confirmed.
All substantive evidence comes from a single reviewed paper (2026 arXiv preprint); treat this as an early-stage profile.
Output, impact, and activity dimensions are unscored because no publication-volume or citation metrics were available.