Since 2003, when the first human genome sequence was finished, doctors have been expecting a change in health care that is driven by data. New molecular and phenotypic questions would help doctors make better diagnoses, come up with better ways to treat people, and find ways to keep people from getting sick.
Some of what was promised has already come true. Researchers use healthcare data all the time to find new things. They are finding the genomic causes of cancer and many other common and rare diseases, making new therapies that target molecules, and using new machine learning methods to make the most of the huge amount of computing power available. We’re starting to see the results of all this work.
According to new research, there are several ways to speed up the promise of precision medicine in a fair way, and we’re going to show you what the future of precision medicine looks like.
future of precision medicine
- Huge, longitudinal cohorts
- Diversity and inclusion
- Big data and artificial intelligence
- Routine clinical genomics
- Electronic health records (EHRs)
- Phenomics and environment
- Privacy, trust, return of value
Huge, longitudinal cohorts
Over the past 20 years, researchers have studied large groups of people using genomic, laboratory, and lifestyle tests, as well as long-term follow-up on their health. The amount and variety of data is mind-boggling, as are the chances to find new things in every area of medicine. Now, we need to make it easier for researchers to combine data from different groups. We need to create and coordinate common data models and file formats to make it easier for people to work together and share information. It would be hard to say how important this work could be for research around the world.
Diversity and inclusion
The lack of different kinds of people in research studies is one of the biggest problems (and opportunities) facing the biomedical industry right now. Less than 3% of the people who took part in published genome-wide association studies were of African, Hispanic, or Latin American descent, and 86% of the people who took part in clinical trials were white.
The lack of diversity in research could make health differences worse and also hurt biological discoveries that could help all people. Routine collection of social determinants of health in both research and clinical care, along with more precise measurements of environmental factors, habits, and genetic ancestry, can lead to more rational, etiology-based adjustments and better risk stratifications and treatments.
Big data and artificial intelligence
The growth of clinical data (including image, narrative, and real-time monitoring data), the development of molecular technologies (genomics being the most prominent of these technologies), and the availability of devices and wearables that are capable of providing high-resolution data streams will significantly expand the availability of specific phenotype and environmental data that was not previously available at this scale. It is possible that applications of machine learning will lead to new taxonomies of disease by using genetic, phenomic, and environmental predictors.
Routine clinical genomics
Clinical genomic analysis is usually only done today when certain cancers or rare genetic diseases are being looked at, and many commonly ordered tests only look at a few genetic loci. As time goes on, whole-genome approaches will become a normal, early step in understanding, preventing, detecting, and treating both common and rare diseases.
The last 10 years have also shown that many common diseases, like diabetes or high blood pressure, can be linked to genetic risks at thousands of loci. These links are often found by putting together data from hundreds of thousands of people who took part in huge genetic studies.
Also, using polygenic risk scores may help providers categorize the risk of people who would have been missed by traditional screening methods. This could help them find new groups of people who need treatment or screening.
Also, pharmacogenomics can make drugs work better, reduce side effects, and save money. Genomic-guided therapies are becoming the standard of care for some types of cancer, but germline pharmacovariants have only been used by a small number of medical centers to help decide what drugs to give. There are still many things that make precision pharmacotherapy hard to use, and scientists still have a long way to go.
Electronic health records (EHRs)
Detailed assessments of phenotype, exposure, and health outcomes are the key to any longitudinal cohort, and EHRs and other health data are the key to getting up to 20 years of disease and treatment information that can be used for research.
EHRs provide phenotypes and data as well as new ways to design studies that aren’t always possible with research collections because they don’t collect health information in a systematic way.
EHR data need to be cleaned up and made consistent, and they can be affected by clinical and insurance biases. Unstructured EHR data, like narrative reports or image data, often needs advanced methods like natural language processing or machine learning to be useful on a population scale. But all of these tools are becoming more and more available and useful, giving people access to data on a scale, in a depth, and with a level of detail that would not be possible with data only from research.
Also, as clinical sequencing grows, the number of genotypes that can be found through clinical care will quickly grow to be much larger than those that can be found through research. Large-scale genetic studies may not need to collect separate research biospecimens for as many genomic studies as they do now. Then, measuring other biomarkers, cell-free DNA, exposures, and epigenomics could replace collecting biospecimens for research.
Phenomics and environment
Over the next 10 years, research and clinical will uses different ways to measure clinical phenotypes, exposures, and lifestyle. Wearable devices can be used to track activity, physical measurements, and exposures.
Activity monitors that measure things like single-lead electrocardiograms and oxygen saturation are getting cheaper and easier to share with providers. Since most of a patient’s life happens outside of the healthcare system, adding wearable devices and other information from the patient would improve the EHR and make telehealth more useful. Also, putting these tools together could increase health-related data from places other than hospitals.
Privacy, trust, return of value
Precision medicine can only be useful if a lot of people use it, and for a lot of people to use it, they have to trust it, have their privacy protected, and get something out of it. We know that science hasn’t always been dependable or treated everyone the same.
Transparency, real engagement with communities, and letting participants help decide how research is run can build trust, turn participants into advocates, and make sure that research is headed in a way that is more thoughtful and sensitive to different cultures.