Скачать или смотреть Using machine learning predictions to design more efficient MS clinical trials

Multiple sclerosis (MS) has a very heterogenous presentation, with some patients worsening steadily over time, whilst others experience sporadic autoimmune attacks, making measuring treatment efficacy very challenging. David Li-Bland, PhD, Unlearn.ai, Oakland, CA, discusses the need to increase the efficiency of MS clinical trials, to reduce the cost and increase the number of studies. Dr Li-Bland explains how 2400 patients with MS, including relapse-remitting, secondary-progressive, and primary-progressive MS, were used to train a machine learning model to predict MS progression. The model generates clinical predictions, such as Expanded Disability Status Score, MS Functional Composite (MSFC), and relapse events, based on a patient’s baseline characteristics. The model is used to create a digital twin for each patient, predicting the long-term effects of disease and relapse occurrence when the patient is given placebo. Digital twins can be used to simulate the outcomes of a control group, understand the effects of particular inclusion or exclusion criteria, or to select cohorts with different types of progression, allowing a reduction in the control group size and increased clinical trial efficiency. This interview took place at the ACTRIMS Forum 2022 in West Palm Beach, Florida.