Chowell2022 - Random Forest model to predict efficacy of immune checkpoint blockade across multiple cancer patient cohorts

  public model
Model Identifier
BIOMD0000001066
Short description
This is a Random Forest algorithm-based machine learning model called RF16, which incorporates a total of 16 genomic, molecular, demographic, and clinical features to predict the immunotherapy response for a patient. The model assigns a value of 0 for NonResponder and 1 for Responder. Please be aware that the column names in the GitHub code and the downloaded dataset from the publication may vary. Users are advised to make minor adjustments to either the code or the dataset to ensure compatibility. The curated version of the model has modified the column names in the training code to align with the dataset. GitHub repository: https://github.com/CCF-ChanLab/MSK-IMPACT-IO
Format
Open Neural Network Exchange
Related Publication
  • Improved prediction of immune checkpoint blockade efficacy across multiple cancer types.
  • Chowell D, Yoo SK, Valero C, Pastore A, Krishna C, Lee M, Hoen D, Shi H, Kelly DW, Patel N, Makarov V, Ma X, Vuong L, Sabio EY, Weiss K, Kuo F, Lenz TL, Samstein RM, Riaz N, Adusumilli PS, Balachandran VP, Plitas G, Ari Hakimi A, Abdel-Wahab O, Shoushtari AN, Postow MA, Motzer RJ, Ladanyi M, Zehir A, Berger MF, Gönen M, Morris LGT, Weinhold N, Chan TA
  • Nature biotechnology , 4/ 2022 , Volume 40 , Issue 4 , pages: 499-506 , PubMed ID: 34725502
  • Human Oncology and Pathogenesis Program, Memorial Sloan Kettering Cancer Center, New York, NY, USA.
  • Only a fraction of patients with cancer respond to immune checkpoint blockade (ICB) treatment, but current decision-making procedures have limited accuracy. In this study, we developed a machine learning model to predict ICB response by integrating genomic, molecular, demographic and clinical data from a comprehensively curated cohort (MSK-IMPACT) with 1,479 patients treated with ICB across 16 different cancer types. In a retrospective analysis, the model achieved high sensitivity and specificity in predicting clinical response to immunotherapy and predicted both overall survival and progression-free survival in the test data across different cancer types. Our model significantly outperformed predictions based on tumor mutational burden, which was recently approved by the U.S. Food and Drug Administration for this purpose. Additionally, the model provides quantitative assessments of the model features that are most salient for the predictions. We anticipate that this approach will substantially improve clinical decision-making in immunotherapy and inform future interventions.
Contributors
Submitter of the first revision: Divyang Deep Tiwari
Submitter of this revision: Divyang Deep Tiwari
Curators: Divyang Deep Tiwari

Metadata information

is (2 statements)
BioModels Database BIOMD0000001066
BioModels Database MODEL2304250001

isDescribedBy (3 statements)
PubMed 34725502
GitHub repository unknown
PMID unknown

hasTaxon (1 statement)
Taxonomy Homo sapiens

hasProperty (34 statements)
NCI Thesaurus OBO Edition Hemoglobin
NCI Thesaurus OBO Edition Heterozygosity
Experimental Factor Ontology (EFO) tumor stage
STATO: the statistical methods ontology random forest procedure
NCI Thesaurus OBO Edition Neutrophil to Lymphocyte Ratio Measurement (NLR)
GitHub repository Scikit-Learn
OBI: Ontology for Biomedical Investigations supervised machine learning
NCI Thesaurus OBO Edition Overall Survival
NCI Thesaurus OBO Edition Receiver Operator Characteristics
NCI Thesaurus OBO Edition Immune Checkpoint Inhibitor
OGG: Ontology of Genes and Genomes CTLA4 (cytotoxic T-lymphocyte-associated protein 4)
NCI Thesaurus OBO Edition Immunotherapy
NCI Thesaurus OBO Edition Tumor Mutation Burden (TMB)
NCI Thesaurus OBO Edition Pembrolizumab
OBCS: Ontology of Biological and Clinical Statistics specificity
OBCS: Ontology of Biological and Clinical Statistics sensitivity
NCI Thesaurus OBO Edition Nivolumab
Software Ontology (SWO) Python
NCI Thesaurus OBO Edition Genome
NCI Thesaurus OBO Edition Atezolizumab
Experimental Factor Ontology (EFO) programmed death-ligand 1 measurement (PDL1)
NCI Thesaurus OBO Edition Microsatellite Instability (MSI)
NCI Thesaurus OBO Edition PD-1 Ligand Inhibitor
Experimental Factor Ontology (EFO) progression free survival
STATO: the statistical methods ontology precision
OBI: Ontology for Biomedical Investigations machine learning
NCI Thesaurus OBO Edition Checkpoint Blockade Immunotherapy
DICOM Controlled Terminology Response Evaluation Criteria In Solid Tumors (RECIST)
NCI Thesaurus OBO Edition Chemotherapy
NCI Thesaurus OBO Edition Albumin
NCI Thesaurus OBO Edition Platelet
CLO: Cell Line Ontology PDL-1 Cell
NCI Thesaurus OBO Edition Cancer Immunotherapy
Experimental Factor Ontology (EFO) body mass index (BMI)

occursIn (20 statements)
BTO (BRENDA Tissue Ontology) breast cancer cell
BTO (BRENDA Tissue Ontology) hepatobiliary carcinoma cell
BTO (BRENDA Tissue Ontology) uterine endometrial cancer cell
BTO (BRENDA Tissue Ontology) Renal Cancer Cell Line
BTO (BRENDA Tissue Ontology) pancreatic cancer cell
BTO (BRENDA Tissue Ontology) Kidney Cancer Cell
BTO (BRENDA Tissue Ontology) esophageal cancer cell
BTO (BRENDA Tissue Ontology) colorectal cancer cell
BTO (BRENDA Tissue Ontology) Skin Cancer Cell
BTO (BRENDA Tissue Ontology) Bladder
BTO (BRENDA Tissue Ontology) ovary cancer cell
BTO (BRENDA Tissue Ontology) gastric cancer cell
BTO (BRENDA Tissue Ontology) breast cancer cell line
BTO (BRENDA Tissue Ontology) NSCLC Cell
BTO (BRENDA Tissue Ontology) mesothelioma cell
BTO (BRENDA Tissue Ontology) small cell lung cancer cell
BTO (BRENDA Tissue Ontology) head and neck cancer cell
BTO (BRENDA Tissue Ontology) ovary cancer cell line
BTO (BRENDA Tissue Ontology) Sarcoma Cell
BTO (BRENDA Tissue Ontology) Melanoma Cell

hasInput (4 statements)
NCI Thesaurus OBO Edition Demographic Data
ERO: eagle-i resource ontology molecular data
Experimental Factor Ontology (EFO) clinical data
Experimental Factor Ontology (EFO) genomic data

hasOutput (1 statement)
hasDataset (1 statement)
unknown unknown

Curation status
Curated
Modelling approach(es)
computational model
Tags
FAIR-AIML Machine Learning Model

Connected external resources

Name Description Size Actions

Model files
RF16.onnx ONNX file of the Random Forest (RF16) trained model.
 1.80 MB Preview | Download

Additional files
ICB_annotation.csv Model annotation in csv format. 6.88 KB Preview | Download
Reproduced Figures.pdf Comparison between original figures (Fig. 1c, 1d and 2a from manuscript) and reproduced figures.
 727.33 KB Preview | Download
Reproduced Metrics.pdf Comparison between original confusion matrix (Fig. 2d-g from manuscript) and reproduced confusion matrix.
 64.12 KB Preview | Download
dockerICB_predict.zip Docker file to load the 'RF16.onnx' model and make predictions. 529.05 KB Preview | Download
dockerICB_train.zip Docker file and dependencies to train the RF16 model using train.py script.
 304.04 KB Preview | Download
train.py Python script which loads the dataset to train the RF16 and RF11 model and create text files with prediction probabilities.
 3.36 KB Preview | Download

  • Model originally submitted by : Divyang Deep Tiwari
  • Submitted: Jun 29, 2023 2:26:38 PM
  • Last Modified: Jun 29, 2023 2:26:38 PM
Revisions
  • Version: 1 public model Download this version
    • Submitted on: Jun 29, 2023 2:26:38 PM
    • Submitted by: Divyang Deep Tiwari
    • With comment: First submission of the Immune Checkpoint Blockade (ICB) machine learning model.