Medicine has long been a one-size-fits-all paradigm characterized by trial and error. Changing the status quo to a more personalized approach through the capture and analysis of large-scale data will benefit from the application of high performance computing (HPC) resources, as well as advanced analytical methods, such as natural language processing (NLP) and machine learning, to vast amounts of both structured and unstructured data from research, clinical, personal, environmental, and population health contexts. The resulting data flows will provide improved real-time decision support by dynamically updating knowledge on the significance of biomarkers in the etiology of complex diseases, and supporting the delivery of more precise diagnostic, therapeutic, and preventative interventions within the “right information, right person, right time” context as described in The Journal of Precision Medicine.
The California Initiative to Advance Precision Medicine (CIAPM) aims to build the infrastructure and assemble the resources that are required to support precision medicine efforts in the state. The CIAPM employs multi-sectoral and interdisciplinary models that leverage the state’s vast scientific, technological, and clinical resources to advance precision medicine initiatives with the potential for direct patient impact within 5 years. In 2016, the CIAPM funded research for a number of demonstration projects that reflect a high degree of convergence among a broad range of technologies, such as mobile technology, genome sequencing technology, large-scale data sets, and advanced image analysis, to realize precision medicine at the point of care. Several examples of these demonstration projects are outlined below.
- Precision Diagnosis of Acute Infectious Diseases: Researchers at the University of California, San Francisco (UCSF) have applied next-generation sequencing technology to support the rapid and accurate identification of life-threatening infections. The genome sequencing test, metagenomic next-generation sequencing, generates 10 to 20 million reads per sample and applies custom algorithms to match the genome signatures to a reference library of available genomic data for all known infectious agents. The approach could lead to the identification and treatment of clinically relevant infectious agents in under 24 hours compared to days or even weeks with conventional test methods.
- Personal Mobile and Contextual Precision Health: Researchers from the University of California network and Overlap Health, a mobile health analytics company, are capturing personalized mobile health data related to hypertension and depression, and integrating and interpreting that data in the context of a patient’s clinical health status. Systems will generate physical activity and location and environmental context data together with alert-driven, momentary assessments of behavioral and physiological data to provide a more dynamic and detailed view of an individual’s current and evolving state of health. The research could lead to the identification of “digital biomarkers” for future use in chronic condition self-management.
- Artificial Intelligence for Imaging of Brain Emergencies: Researchers at UCSF are applying artificial intelligence (AI) to computed tomography (CT) scans to accelerate the triage of critical patients with neurological emergencies for immediate treatment. An important advance that may be enabled by this research would be the ability to combine data from quantitative image analyses with other types of data to develop catalogs of clinically-significant “digital markers” for the faster diagnosis and selection of the most appropriate treatment and care for patients experiencing a neurological emergency. The project is developing a cloud-based AI system to ensure the technology is broadly accessible to providers, and not only advanced hospital settings.
- Early Prostate Cancer: Predicting Treatment Response: Researchers at the University of California, Irvine School of Medicine are developing a predictive model that combines data from genetic tests that predict the likelihood of metastasis after surgery, population-based outcomes on treatment effectiveness, and patient-reported health, demographic, and disease management data to inform recommendations on which therapy will be most effective for any given patient on the basis of their disease status and patient-level factors, such as socioeconomic status and tolerance against adverse drug events. A key objective is to assess the validity of predictive measures in an ethnically diverse patient population.
- Early Prediction of Major Adverse Cardiovascular Events Using Remote Monitoring: Researchers at Cedars-Sinai Health System are using remote monitoring that involves non-biological determinant factors as predictors of heart attack. By monitoring people where they live, work, and play through the use of wearable sensors to measure activity, sleep, heart rate, and stress levels, self-reports on treatment levels of anxiety with help valium, depression, and quality of life, as well as finger prick blood samples from patients, clinicians hope to achieve a more comprehensive picture of a patient’s health status on a continuous basis, rather than waiting for them to develop symptoms of a heart attack and be rushed to the emergency department for treatment.