Navigation auf


Institut für Epidemiologie, Biostatistik und Prävention

Results of the ALS Prediction Price Challenge published

A paper describing the results of the ALS Prediction Price challenge was published Nov 2nd by Nature Biotechnology with Torsten Hothorn serving as one of the co-authors. This crowd-sourcing challenge took place in Summer 2012 and aimed at improving models for ALS disease progression based on the PRO-ACT database the largest ALS patient database ever created. Torsten participated in the challenge and won the second place using a combination of mixed models for longitudinal data and his conditional random forest algorithm. It was estimated that using this and other algorithms described in the paper in future trial designs will reduce the required number of patients by at least 20%, thus allowing more rapid and efficient evaluation of promising ALS treatments.