Automated real-world data integration improves cancer outcome prediction

The digitization of health records and growing availability of tumour DNA sequencing provide an opportunity to study the determinants of cancer outcomes with unprecedented richness. Patient data are often stored in unstructured text and siloed datasets. Here we combine natural language processing annotations<sup>1,2</sup> with structured medication, patient-reported demographic, tumour registry and tumour genomic data from 24,950 patients at Memorial Sloan Kettering Cancer Center to generate a clinicogenomic, harmonized oncologic real-world dataset (MSK-CHORD). MSK-CHORD includes data for non-small-cell lung (n = 7,809), breast (n = 5,368), colorectal (n = 5,543), prostate (n = 3,211) and pancreatic (n = 3,109) cancers and enables discovery of clinicogenomic relationships not apparent in smaller datasets. Leveraging MSK-CHORD to train machine learning models to predict overall survival, we find that models including features derived from natural language processing, such as sites of disease, outperform those based on genomic data or stage alone as tested by cross-validation and an external, multi-institution dataset. By annotating 705,241 radiology reports, MSK-CHORD also uncovers predictors of metastasis to specific organ sites, including a relationship between SETD2 mutation and lower metastatic potential in immunotherapy-treated lung adenocarcinoma corroborated in independent datasets. We demonstrate the feasibility of automated annotation from unstructured notes and its utility in predicting patient outcomes. The resulting data are provided as a public resource for real-world oncologic research.

Authors

Institutions

Publication Details

Journal
Nature
Published
2024-11-06
DOI
https://doi.org/10.1038/s41586-024-08167-5
Citations
123
Primary Topic
Radiomics and Machine Learning in Medical Imaging
Type
article
Field-Weighted Citation Impact
49.72
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
article

Automated real-world data integration improves cancer outcome prediction

Debyani Chakravarty, Ritika Kundra, Kelly R. Pekala, Thinh Ngoc Tran et al.
123 citations
Nature
Radiomics and Machine Learning in Medical Imaging
49.72
article

Automated real-world data integration improves cancer outcome prediction

Debyani Chakravarty, Ritika Kundra, Kelly R. Pekala, Thinh Ngoc Tran, Marzia Capelletti, Julio García‐Aguilar, Kevin Boehm, Chenlian Fu, Jordan E. Eichholz, Henry S. Walch, Robert E. Sheridan, Helena A. Yu, Jianjiong Gao, Aaron James Stonestrom, Anisha Luthra, Clare Wilhelm, Ross L. Levine, Calla Chennault, Francisco Sánchez-Vega, Benjamin Groß, Mirella L. Altoé, Pedram Razavi, Xiang Li, Anita Kohli, Anton Safonov, Jian Carrot‐Zhang, Nikolaus Schultz, Wungki Park, Sohrab P. Shah, Lior Z. Braunstein, Robert Pimienta, Mitchell I. Parker, Michael F. Berger, Mark T.A. Donoghue, Christopher J. Fong, Rona Yaeger, Bob T. Li, Paolo Manca, Manda Wilson, Peter D. Stetson, Mark E. Robson, Marc Ladanyi, Nicholas D. Socci, Baby A. Satravada, Charles M. Rudin, Surya Narayana Rangavajhala, Tyler Aprati, Aaron Lisman, Steven B. Maron, Ronglai Shen, Daniel Richard Gomez, Jason C. Chang, Arfath Pasha, Justin Jee, Gregory J. Riely, Avery Wang, Wassim Abida, Gaofei Zhao, Kenneth L. Kehl, David R. Jones, Gayathri Subramanian, Maria Perry, Brooke Mastrogiacomo, Anyi Li, Howard I. Scher, Axel Martin, Karl Pichotta, Eileen Mary O'Reilly, Jorge S. Reis‐Filho, Ino de Bruijn, Xiao-Jun Bi, Subhiksha Nandakumar, Michele Waters, Tom Pollard, Pablo Sánchez Vela, Chaitanya Bandlamudi, Deborah Schrag, A. Rose Brannon, Mono Pirun, Tejiri Agbamu, Ayush Kris, Lisa Loudon, Yuan Chen, Siyang Liu, David Liu, Hongxin Zhang, Xuechun Bai, Susie Kim, David B. Solit, Raymond Lim, Darin Moore, J.A. Valverde García, Mehnaj Ahmed, Naveen Rachuri, U Justin
article en
123 citations

Abstract

The digitization of health records and growing availability of tumour DNA sequencing provide an opportunity to study the determinants of cancer outcomes with unprecedented richness. Patient data are often stored in unstructured text and siloed datasets. Here we combine natural language processing annotations<sup>1,2</sup> with structured medication, patient-reported demographic, tumour registry and tumour genomic data from 24,950 patients at Memorial Sloan Kettering Cancer Center to generate a clinicogenomic, harmonized oncologic real-world dataset (MSK-CHORD). MSK-CHORD includes data for non-small-cell lung (n = 7,809), breast (n = 5,368), colorectal (n = 5,543), prostate (n = 3,211) and pancreatic (n = 3,109) cancers and enables discovery of clinicogenomic relationships not apparent in smaller datasets. Leveraging MSK-CHORD to train machine learning models to predict overall survival, we find that models including features derived from natural language processing, such as sites of disease, outperform those based on genomic data or stage alone as tested by cross-validation and an external, multi-institution dataset. By annotating 705,241 radiology reports, MSK-CHORD also uncovers predictors of metastasis to specific organ sites, including a relationship between SETD2 mutation and lower metastatic potential in immunotherapy-treated lung adenocarcinoma corroborated in independent datasets. We demonstrate the feasibility of automated annotation from unstructured notes and its utility in predicting patient outcomes. The resulting data are provided as a public resource for real-world oncologic research.

NatureVol. 636(8043)
Memorial Sloan Kettering Cancer Center (US), Cornell University (US), Dana-Farber Cancer Institute (US), Caris Life Sciences (United States) (US)
Good health and well-being
Openalex Percentile: Top 0%
Radiomics and Machine Learning in Medical Imaging
49.72
DOIOpen Access PDFOpenAlex
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.