Multidisciplinary teams of Cedars-Sinai researchers advance emerging technologies, such as AI and proteomics, to establish new norms in oncology practice.
In late 2021, Cedars-Sinai launched the new Department of Computational Biomedicine, attracting some of the world’s top researchers in the field of big data. This resulted in rapid growth in infrastructure, faculty and funding to support a new wave of research using machine learning and AI.
The department now has 16 faculty with expertise in AI, bioinformatics, biostatistics and data science. Since its inception, it has grown to house the new Center for Artificial Intelligence Research and Education (CAIRE), which focuses on AI automation, ethics and education. The scientific, data and architectural synergies between the department, CAIRE and vertically integrated clinical care centers aim to catalyze practice-changing discoveries.
Initial examples of oncology research projects and programs supported by Department of Computational Biomedicine investigators include:
Molecular Twin Precision Oncology Platform (MT-POP). By creating virtual replicas of patients and their real-life tumor samples, this AI-based data-analysis tool has three fundamental and interrelated aims:
To generate better predictors of disease behavior in patients
To enhance the tools that researchers have for discovery, and thus accelerate their progress
To democratize personalized oncology by providing predictors that can be more cost-effectively deployed in resource- and technology-limited settings globally
AI processes clinical and “multiomic” molecular data from patients’ normal and tumor tissues and multiple body fluids. Additional data integrated from wearables, microbiome profiling, tumor sequencing and other sources will make MT-POP one of the world’s most robust oncology data sources. Clinical outcomes entered into the system then allow AI-based discoveries of therapeutic targets, biomarkers and prognostic indicators.
More than 1,000 patients have consented to be part of the platform, with thousands more anticipated in the coming years.1
SynTwin. This extension of AI-based molecular data analysis creates digital, synthetic “twins” by studying the distance between data points of actual patient characteristics. The algorithm then generates data for similar synthetic patients and pairs up the closest matches of hypothetical and real patients for comparison. The approach combines nearest-neighbor statistical methods, network science and AI. SynTwin tactics may reduce privacy concerns while increasing the available data from which AI investigations can draw conclusions. In a test application, SynTwin significantly improved mortality prediction among breast cancer patients in the National Cancer Institute’s Surveillance, Epidemiology and End Results (SEER) registry compared to real-life SEER data.2
Cedars-Sinai Biomedical Imaging Research Institute collaborations. As members of one of the most high-tech and advanced cancer-imaging departments in the country, Biomedical Imaging Research Institute (BIRI) faculty hold numerous federally funded cancer-imaging grants, including efforts to study novel high-resolution MRI techniques to better detect small prostate lesions,3 predict pancreatic ductal adenocarcinoma through AI analysis of pre-diagnostic CT scans4 and better target therapy for lethal prostate cancer. They also have developed dynamic contrast-enhanced MRI techniques to accurately differentiate and diagnose pancreatic cancer versus chronic pancreatitis.5 BIRI has partnered with the Department of Computational Biomedicine to contribute computational imaging of patients enrolled in MT-POP. In the first such collaboration, the teams integrated pancreatic-cancer imaging and AI to improve prediction of cancer outcomes using MT-POP data.
The research cores at Cedars-Sinai are a vital collection of centralized labs and services offering state-of-the-art technology, instruments and resources to staff and investigators. From centers supporting biobanking, imaging and data science to novel programs offering proteomics, applied genomics, flow cytometry, metabolomics and much more, Cedars-Sinai Cancer researchers have a wealth of tools and resources at their fingertips.
Board of Governors Innovation Center (BOGIC). The BOGIC is a state-of-the-art single-cell deep phenotyping center that performs single-cell genomics, transcriptomics, epigenomics, proteomics and metabolomics.
BOGIC is one of only two sites in the nation to currently have a Bruker timsTOF SCP, which is capable of assaying up to 4,000 proteins from a single cell. While BOGIC investigators explore many different topics, 60% focus on cancer, including the biology underlying the initiation, maintenance and development of therapy resistance.
BOGIC and the Department of Computational Biomedicine collaborate to analyze large orthogonal datasets and integrate these data with clinical phenotyping, radiology and pathology studies using AI and machine learning.
Precision Biomarker Laboratories (PBL). This independent contract research organization owned by Cedars-Sinai focuses on providing complete proteomics solutions across biomarker discovery, validation and CAP/CLIA-certified laboratory testing capabilities. PBL collaborates with Cedars-Sinai investigators (and outside companies) who have candidate biomarkers for large-scale validation, such as the rapid, cost-effective proteomics panel developed by the MT-POP work in pancreatic cancer.1 The assay will be used in a large-scale clinical evaluation of the panel’s predictive performance compared with CA 19-9, the current standard-of-care marker.
Poised to Make a Difference
Pilot studies and clinical trials using these resources and technologies are underway across numerous diagnoses. Other endeavors, such as ChatGPT-driven education tools for patients with liver cancer, are also in development.
New therapeutic targets and more precise prognostics hold practice-changing potential for everything from lung and breast cancer to brain tumors and hidden prostate lesions—democratizing the promise of AI and precision medicine.
“Profiling of pancreatic adenocarcinoma using artificial intelligence-based integration of multiomic and computational pathology features.” Nat Cancer. In press.
“SynTwin: A graph-based approach for predicting clinical outcomes using digital twins derived from synthetic patients.” Pacific Symposium on Biocomputing. 29:96-107(2024)
“3D high-resolution diffusion-weighted MRI at 3T: Preliminary application in prostate cancer patients undergoing active surveillance protocol for low-risk prostate cancer.” Magn Reson Med. PMID: 25761871.
“Predicting pancreatic ductal adenocarcinoma using artificial intelligence analysis of pre-diagnostic computed tomography images.” Cancer Biomarkers. PMID: 35213359.
“Multitasking dynamic contrast enhanced magnetic resonance imaging can accurately differentiate chronic pancreatitis from pancreatic ductal adenocarcinoma.” Front Oncol. PMID: 36686811.