Unlearn to use APST data to develop digital twin models for ALS trials

AI developer Unlearn will use data from APST Research to make digital twin models for amyotrophic lateral sclerosis (ALS)-focused clinical trials under a new agreement.

The deal – details announced last week – will see Unlearn build the twins using data from 8,000 people taking part in a longitudinal ALS study.

Unlearn founder Aaron Smith told Clinical Insider, “We will use the data to enhance the accuracy and add capabilities to our ALS digital twin generator model. The twins created will be used in Phase II and III trials to increase power and reduce sample sizes.

“In addition,” Smith continued, “the models will be used to inform future study designs and discover subgroups with enhanced treatment responses. “

The data will include patient self-assessments, biomarker analyses, as well as common ALS clinical trial metrics such as ALSFRS-R, SVC, FVC, and neurofilament light chain (NfL) measurements.

Smith said, “A particular feature of the APST data is the richness of the NFL measurements, which are taken alongside key endpoints, disease history, genetic information, and demographics.

“This will allow Unlearn’s ALS model to better serve contemporary studies which incorporate NfL measurements,” he added.

Smith cited the ability to model complex, non-linear interactions as the key advantage AI systems have over traditional analytical approaches.

“This allows us to create comprehensive digital twins that contain predictions for different variables of interest. Development teams can use these twins to plan faster and more powerful studies and answer a myriad of questions during development.”

Unlearn and APST Research will also collaborate on research publications focused on digital twin technology and its applications in research. The AI firm has also agreed to provide digital twins of study participants to APST.

DepositPhotos/WrightStudio