Poster: A Machine Learning ALS Survival Model Lacking Vital Capacity for use in Clinical Trials during the COVID-19 Pandemic

Dec 9, 2020

Presented at the 31st International Symposium on ALS/MND, December 9, 2020.

Background: The ongoing COVID-19 pandemic has made it difficult in some clinical settings to measure vital capacity for the purpose of eligibility determination. We developed a machine learning survival model without the use of baseline vital capacity measures and asked whether it could stratify clinical trial patients and a wider ALS patient population derived from a tertiary care ALS center (Emory University).

Conclusions:

  1. A survival model that doesn’t use baseline vital capacity as a predictor can be built using machine learning
    • Model degrades little compared to model that includes VC
    • Model remains useful for stratification
  2. The “VC-free” survival model returns useful predictions for a clinical trial population as well as a broad ALS clinic population
  3. The model can be used to stratify ALS patients for ALS clinical trials in the absence of baseline vital capacity measurements
    • Useful for inclusion/exclusion during COVID19 pandemic

Authors: Danielle Beaulieu, Jonavelle Cuerdo, David L. Ennist

Download this Poster by Entering Your Email Address Below:

Share This