Access to robust, multimodal, and longitudinal clinical and genetic patient real-world data (RWD) is paramount to advancing the understanding of rare and ultrarare diseases. Driving insights from this data is the goal of Sema4’s pioneering health intelligence platform, Centrellis®. The AI-led process involves curating unique RWD to understand the foundational natural history of diseases, using that data to make predictions about patient subgroups and to develop metrics to guide drug development, and then testing those predictions in a retrospective or prospective setting with real patients and health systems.
Deep Curation of Patient Data Using Natural Language Processing (NLP) and Expert Review
This approach uses machine learning and NLP to collect and curate unique patient data to understand rare and complex disease progression. Using an innovative genomic infrastructure, Centrellis® can generate, analyze, and interpret a combination of longitudinal clinical and genetic data (clinicogenomic data) from electronic health records (EHRs), including free text clinical notes, to create diverse patient profiles to inform a deeper understanding of rare disease trajectories. RWD can also be retrieved from lab results, insurance claims, clinical trials, medical reports, omics and safety data, scientific literature, and more.
In addition to leveraging the power of AI, Sema4 also incorporates key scientific opinion leader knowledge from renowned health systems, like Mount Sinai. For example, an expert consultant is an important component in creating an accurate data dictionary.
This combination of expert guidance with AI-led understanding is key to creating research-ready data. This data provides decision makers with an invaluable tool to gain real-world
insights to make data-driven decisions.
Leveraging AI-Led Insights to Make Data-Driven Decisions
The significance of curating patient records in aggregate from disparate sources allows for a deeper understanding of the rare disease patient journey. For example, Sema4’s clinicogenomic dataset can be used to develop foundational natural history studies to generate deep longitudinal and multimodal insights on rare diseases like Lysosomal Storage Disorders (LSDs).
Informed findings from this dataset can also be used to predict and test patient metrics, such as rate of disease progression and therapeutic response to help guide drug development. Additionally, this integrated, RWD can be used in combination with advanced molecular profiling to go beyond general comprehensive genomic profiling to more targeted analyses, including single-cell RNA analysis and other advanced omics. For example, biobanked samples or those from clinical practice under IRB can be profiled with these technologies to confirm initial findings.
Connecting the Dots from R&D through Commercialization
Current linear R&D processes are inefficient, resulting in time and cost sinks. AI engines and other innovative technologies that continually learn are game changers, creating a paradigm shift in R&D by enabling a more cyclical, iterative process.
This can help drive real-world insights to power informed decision making across the drug discovery pipeline, accelerating go-to-market progress to ultimately deliver safe, cost-effective therapies for patients in dire need.
Learn more sema4.com/rwd