{
  "abstract": "Background Before mechanical thrombectomy for acute large-vessel occlusion, families must authorize treatment with limited prognostic context. Clinicians possess key information—age, stroke severity, comorbidities, functional status—yet no validated framework converts these data into a structured outcome estimate. Existing tools require post-procedural imaging, computational resources, or achieve only moderate discrimination (AUC 0.70-0.83). We developed a machine learning-derived bedside risk score using exclusively pre-operative variables available during family consent.Methods We conducted a single-center retrospective study of 421 consecutive thrombectomy patients (January 2022-June 2024). All intra- and post-operative variables were excluded. Gradient boosting machine models with recursive feature elimination (GBM-RFE) identified optimal predictors from 15 pre-operative candidates. An ordinal logistic regression model predicted the full modified Rankin Scale (mRS) spectrum (0-6), and coefficients were transformed into integer points (scaling factor of 20) to yield a simplified arithmetic score. Data were split 80/20 into discovery (n=337) and validation (n=84) cohorts. Performance was assessed by AUC, sensitivity, and specificity for functional independence (mRS ≤2) at 90 days.Results Of 421 patients (median age 70; median NIHSS 16), 36.6% achieved functional independence. GBM-RFE identified 10 optimal predictors; NIHSS, time since last known normal, BMI, and age dominated. The ordinal model achieved AUC 0.845 (95% CI 0.764-0.927) with 92.3% sensitivity and 67.2% specificity. Four variables with coefficients rounding to zero were eliminated, yielding a 6-variable equation: Risk Score = (2 × NIHSS) + (22 × Cancer History) + (7 × Diabetes) + (7 × Ambulatory Status) + (6 × Pre-stroke mRS) + (2 × Smoking History). At threshold ≥42, the score achieved AUC 0.854 (95% CI 0.775-0.934) with 92.3% sensitivity and 67.2% specificity ( figure 1). Scores below 42 corresponded to predicted independence probabilities of 72-95%.Conclusions This explainable pre-operative risk score enables data-driven family counseling before mechanical thrombectomy using only consent-conversation variables. It exceeds existing pre-treatment scores (AUC 0.70-0.83) without imaging-derived inputs or computational tools, addressing an unmet need for structured prognostic communication at the treatment authorization decision point. External validation is needed.Disclosures J. Keller: None. I. Bediako: None. C. Sollenberger: None. A.Q. Wu: None. E. Ozcan: None. Z. Hoglund: None. J. Catapano: None. J. Burkhardt: None. S. Kandregula: None. R. Rahmani: None. V. Srinivasan: None.Abstract E-064 Figure 1Abstract E-064 Table 1Pre-op predictive performance comparison cross-modelVariableGBM-RFEOrdinal Logistic RegressionRisk ScoreAUC0.8430.8450.85495% CI0.759 - 0.8540.759 - 0.9270.775 - 0.934Sensitivity96.2%92.3%92.3%Specificity60.3%67.2%67.2%",
  "authors": [
    {
      "affiliations": [
        "Neurosurgery, Perelman School of Medicine at the University of Pennsylvania, Philadelphia, PA"
      ],
      "name": "J Keller"
    },
    {
      "affiliations": [
        "Neurosurgery, Perelman School of Medicine at the University of Pennsylvania, Philadelphia, PA"
      ],
      "name": "I Bediako"
    },
    {
      "affiliations": [
        "Neurosurgery, Perelman School of Medicine at the University of Pennsylvania, Philadelphia, PA"
      ],
      "name": "C Sollenberger"
    },
    {
      "affiliations": [
        "Neurosurgery, Perelman School of Medicine at the University of Pennsylvania, Philadelphia, PA"
      ],
      "name": "AQ Wu"
    },
    {
      "affiliations": [
        "Neurosurgery, Perelman School of Medicine at the University of Pennsylvania, Philadelphia, PA"
      ],
      "name": "E Ozcan"
    },
    {
      "affiliations": [
        "Neurosurgery, Perelman School of Medicine at the University of Pennsylvania, Philadelphia, PA"
      ],
      "name": "Z Hoglund"
    },
    {
      "affiliations": [
        "Neurosurgery, Perelman School of Medicine at the University of Pennsylvania, Philadelphia, PA"
      ],
      "name": "J Catapano"
    },
    {
      "affiliations": [
        "Neurosurgery, Perelman School of Medicine at the University of Pennsylvania, Philadelphia, PA"
      ],
      "name": "J Burkhardt"
    },
    {
      "affiliations": [
        "Neurosurgery, Perelman School of Medicine at the University of Pennsylvania, Philadelphia, PA"
      ],
      "name": "S Kandregula"
    },
    {
      "affiliations": [
        "Neurosurgery, Perelman School of Medicine at the University of Pennsylvania, Philadelphia, PA"
      ],
      "name": "R Rahmani"
    },
    {
      "affiliations": [
        "Neurosurgery, Perelman School of Medicine at the University of Pennsylvania, Philadelphia, PA"
      ],
      "name": "V Srinivasan"
    }
  ],
  "title": "E-064 Machine-learning derived equation to predict functional outcomes before mechanical thrombectomy",
  "uid": "0bc16e2d-2148-5e96-8d8d-ade349fb5ea8"
}