Developing the POTOMAC model: A Novel Prediction Model to Study the Impact of LymphOpenia kineTics on Survival Outcomes in Head and Neck Cancer via an Ensemble Tree-based MAChine Learning Approach

Case ID:
C17597

Survival Outcome Prediction Model for Lymphopenia in Head and Neck Cancer

JHU Ref #: [C17597]


Unmet Need

Lymphopenia, a disorder in which your blood doesn't have enough white blood cells, is a potential marker of survival outcomes for many solid tumors of the head and neck. Currently, researchers disagree on whether lymphopenia reflects the patient’s frailty at baseline, or if it is an actionable target for treatment modifications to optimize cancer outcomes. Although multiple studies have attempted to delineate the prognostic values of baseline versus treatment-related lymphopenia (TRL), these studies have primarily relied on the use of a traditional Cox regression model which is not designed to handle complex, high-dimensional survival data. The Cox model is susceptible to overfitting when handling complex high dimensional data. To avoid overfitting using the Cox model, it is necessary for researchers to handpick a subset of variables to include in the final analysis. However, the subset of handpicked variables often reflects a partial experience and thus, current lymphopenia research relies on the Cox model that often necessitates selection of only 1-2 lymphopenia metrics (LMs). Therefore, there is a strong need for a technology that can handle complex, high-dimensional data and consider a comprehensive set of LMs.

 

Value Proposition

·        First model to outline survival risks with lymphopenia in head and neck squamous cell carcinoma (HNSCC)

·        Model has lower risks of overfitting when handling high dimensional data

·        Model allows for exploration of non-linear effects and possible interactions amongst covariates


Technology Description

Researchers at Johns Hopkins have developed a novel prediction model to study the impact of lymphopenia kinetics on survival outcomes in head and neck cancer through a machine learning approach called POTOMAC. In contrast to the Cox model, the POTOMAC model is based on the random survival forest approach and has a lower risk of overfitting when handling high dimensional data. Additionally, it allows for exploration of non-linear effects and possible interactions amongst covariates.

The POTOMAC model includes the consideration of a comprehensive set of 9 LMs with both traditionally reported metrics and novel lymphopenia. As such, the POTOMAC model provides an objective, disciplined approach that seeks to visualize the data in various perspectives, which is essential in the quest to understand complex data patterns in a survival prediction model and move towards consensus and unification of previously described findings.


Stage of Development

·        The product is in the prototype stage of development.

·        Further research is underway for external validation datasets.


Publication

N/A

Patent Information:
Title App Type Country Serial No. Patent No. File Date Issued Date Expire Date Patent Status
MODELING THE IMPACT OF LYMPHOCYTE KINETICS ON SURVIVAL OUTCOMES IN HEAD AND NECK CANCER VIA AN ENSEMBLE TREE-BASED MACHINE LEARNING APPROACH PCT: Patent Cooperation Treaty PCT PCT/US2024/021722   3/27/2024     Pending
Inventors:
Category(s):
Get custom alerts for techs in these categories/from these inventors:
For Information, Contact:
Lisa Schwier
lschwie2@jhu.edu
410-614-0300
Save This Technology:
2017 - 2022 © Johns Hopkins Technology Ventures. All Rights Reserved. Powered by Inteum