Using hybrid PKPD/ML models to accurately predict risk of neutropenia

  • Published October 12, 2023

In an article recently published in the online journal, CPT: Pharmacometrics & Systems Pharmacology, a team of researchers from InsightRX and Memorial Sloan Kettering Cancer Center (MSK) in New York shows how a hybrid pharmacokinetic/pharmacodynamic (PKPD)/machine learning (ML) approach can improve the ability of clinicians to predict the risk of chemotherapy-induced neutropenia in individual patients relative to PKPD or ML alone.

Patients with neutropenia have a lower than normal number of neutrophils, one of several types of white blood cells, which can increase a patient’s risk of developing an infection. Chemotherapy-related neutropenia hospitalizations in the U.S. cost $2.3 billion for adults and $439 million for children. In a 2012 study, total spend topped 8.3% of all cancer-related hospitalization costs. 

The collaborative InsightRX-MSK study suggests that hybrid models hold promise for predicting patients at risk of neutropenia at the point of care – helping to increase patient safety while maintaining therapeutic efficacy.

Published in July, the research paper highlights important outcomes from ongoing collaboration between InsightRX and the MSK Innovation Hub, reporting not only on improved models for prediction of neutropenia, but also proposing new modeling methods applicable to other disease areas. 

InsightRX was one of only four companies chosen in 2021 for the MSK Innovation Hub’s inaugural cohort. The program brings together health technology innovators and a community of researchers, clinicians, and digital health professionals to develop cutting-edge healthcare products designed to transform cancer care. This research paper further validates InsightRX’s focus on AI-enabled computational oncology, resulting in improvements to pharmacology models that enable clinicians to effectively predict neutropenia. 

As co-authors of this paper, we want to thank our study colleagues from InsightRX – Dominic Tong, Vanessa Burns, and Ron Keizer – as well as Bobby Daly, Pedram Razavi, and Jaap Boelens of MSK, for their contributions and hard work. 

You can read the full paper here: Clinical decision support for chemotherapy-induced neutropenia using a hybrid pharmacodynamic/machine learning model.