On June 27th, ImpriMed's Data Science team released a new generation of AI models for canine lymphoma with improved predictive performance. The models, which predict the likelihood of a positive clinical response to 10 different chemotherapy drugs, were optimized to give higher sensitivity and specificity, as well as high positive predictive value (PPV) of our top-ranked predictions. The top-ranked drug prediction now has a 20% higher probability of eliciting a positive clinical response than historical drug selections made during the course of lymphoma care.
Here is a brief history of ImpriMed's single drug response model releases:
2019
- Version 1 of our single drug models (SDM1) consisted of predictive mathematical models for 6 different drugs based on drug sensitivity data and a limited number of known clinical outcomes. These predictions were provided for free as ImpriMed collected additional outcome data.
2020
- The SDM2 release added predictive mathematical models for 4 new drugs, bringing the total number of drug predictions to 10.
- When our database of known outcomes grew to 500 patients, we made our first commercial model release, SDM3. This release replaced 10 mathematical models with more accurate AI models and introduced new math models for three additional drugs, bringing the total number of personalized predictions to 13.
2021
- The SDM3.1 release updated the 13 models from the SDM3 release with a much larger dataset of 840 patients.
2022
- The SDM4 release, introduced this month, replaces our 10 AI models with new algorithms optimized to give higher sensitivity and specificity, as well as high PPV of our top-ranked predictions.
Thanks to your contributions, we are continually updating our machine learning models to improve the accuracy of our predictions and support for additional drugs.