Predicting Survival of Neuroblastoma Patients Using Gene Expression

Neuroblastoma (NB) is the most common solid extracranial tumor found in children, but there is little support for families dealing with a NB diagnosis. The goal of this project is to create a machine learning model (MLM) that can accurately predict survival time for patients diagnosed with NB based on gene expression results from lab tests. Using the R2 platform and a dataset curated by the Amsterdam Institute (consisting of eighty-four complete samples), information was compiled in three areas: the INSS (International Neuroblastoma Staging System), MYCN amplification, and ALK mutations. A MLM was developed and tested using Google Colab, borrowing and modifying code from a project done by DataProfessor. The coefficient of determination (R2) was calculated through both Random Forest and linear regression models, and graphs of the linear regression models were produced through the Scikit-learn (sklearn) library. While the model still has room to improve, it is able to predict survival time with fairly high accuracy (up to R2=0.93). The next step would be to create a user interface that makes the MLM more user-friendly, allowing medical professionals to interact with the data and better understand the likely prognosis of their NB. In the future, more datasets and more genes will be added to the model to increase overall accuracy, allowing healthcare professionals to gain a greater understanding of their patient’s diagnosis.