Abstract WP167: Machine Learning Modelling To Predict 90 Day Home Time In Patients Undergoing Endovascular Thrombectomy
Bibliographic record
Abstract
Background: 90-day home-time, the number of days a patient is back at their premorbid living situation without an increase in level of care in the first 90 days post stroke, is routinely collected in administrative data. We evaluated the prediction of 90-day home-time using machine learning modelling in patients undergoing endovascular treatment (EVT). Methods: We used the QuICR provincial stroke registry and administrative data from Southern Alberta from Jan 2015-Dec 2019 to identify patients who underwent EVT. Imaging data were scored by 2-physician consensus. The primary outcome was 90 day home-time, which has a highly non-normal distribution with excess zero’s. We modelled using generalized boosting machine model with Gaussian distribution. Contribution of different covariates to hometime was determined using partial dependence plots. Results: We identified 659 EVT patients from Jan 2015 to Dec 2019 treated in Calgary, Alberta. Overall,median predicted 90d home-time was 41days (IQR 5.5 to 77.8) with good model accuracy (Root mean square error 6.96). Holding other covariates constant, factors predicting lower 90d-hometime were diabetes mellitus(-14.1d), hypertension (-7d), low baseline ASPECTS (≤5) (-5.6d) and symptomatic intracerebral hemorrhage (sICH) on follow up scan (-13.8d). There was a consistent improvement in the predicted home-time over the last 5 years. There was no meaningful difference in predicted 90d-home-time by age, baseline NIHSS, sex, atrial fibrillation, occlusion site, tandem occlusion, thrombolysis, or successful reperfusion (Figure 1) Conclusions: Predicting 90-d hometime using boosting machine learning modelling is useful to assess complex relationships between predictors and home-time. Hypertension, diabetes, low ASPECTS and sICH were predictors of lower 90-d home-time in this registry.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.002 | 0.008 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.001 | 0.000 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.003 | 0.001 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".