January 5, 2022
Precision health aims to optimize a person’s health by measuring factors that affect health and using that data to target treatments or other actions that will benefit that person. In the months and years to come, advances in and greater adoption of precision health practices will change the way healthcare operates.
More research participants
Precision health researchers want to define exactly what keeps us healthy, as well as what makes us sick. But in order to pinpoint the unique combinations of genetic, behavioral, and environmental factors that contribute to or hinder good health, researchers require access to diverse and deep datasets for analysis. The All of Us Research Program aims to fill this gap and speed up discoveries by gathering health data on 1 million volunteer participants from the United States and monitoring those participants over a decade.
Participant recruitment for such efforts is challenging. Advances in telehealth access and home health tracking devices can address geographic challenges by reducing the need for participants to travel. Proper investment in data platform security and research ethics training and clear communication with participants about those investments can allay fears about healthcare privacy. Harder to overcome is the mistrust of communities who have been exploited by healthcare researchers in the past. Overcoming this hurdle will require a precision population health approach that incorporates participant education, engagement, and outreach.
Every year about 4 million U.S. newborns have their heels pricked and their blood spots sent to labs to be screened for 29 or more serious metabolic, endocrine, and hemoglobin disorders. As a result of these and other newborn screenings, about 13,000 babies each year are diagnosed and referred for additional treatment, preventing lifelong disability and premature death.
As lab testing technology has improved, the number of disorders newborns are screened for has increased. The Department of Health and Human Services (HHS) has a Recommended Uniform Screening Panel (RUSP) of disorders for newborn screening, but each state determines the tests for the babies born there. The increased availability of next-generation gene sequencing has the potential to greatly increase the scope of newborn screening, but we need to increase the usability of the genetic data generated by such testing in order to fulfill this potential, as highlighted in the HHS ONC’s Health IT Buzz article, “A Strong Start: Enhancing Newborn Screening for Precision Public Health.”
Early diagnosis is just the first step in achieving better outcomes. Care gaps for children with disorders identified during newborn screening exist and long-term data collection about these patients is insufficient to assess or improve the quality of care, as explained in the Pediatrics journal article “Ensuring the Life-Span Benefits of Newborn Screening.” The authors identify several potential strategies to address these deficits, such as creating and maintaining nationwide disease-specific registries and linking data from public health records and a patient’s various healthcare providers in health information exchanges.
Fewer medicine-related ER visits and hospitalizations
Tens of thousands of people experience adverse drug reactions as a result of taking prescribed medications every year. Greater adoption of precision prescribing strategies could change that. The September 2021 Diseases journal article titled, “Precision Medicine and Adverse Drug Reactions Related to Cardiovascular Drugs,” focuses on this class of drugs because it is responsible for a large percentage of medication-related hospitalizations.
The article points out that prescribing best practices in the United Kingdom lag in including pharmacogenomic guidance, even when numerous studies have shown links between specific gene variants and bleeding and other adverse effects from commonly prescribed drugs such as warfarin and clopidogrel. Beyond discovering these individual gene-drug interactions, research has revealed how multiple genes contribute to the body’s processing of medication. Technology now makes it possible to combine this research on multiple genes into a genetic risk score to help guide prescribing. The article ends by emphasizing the need to integrate pharmacogenomics into primary care and acknowledging the technological barriers to doing so, “Ultimately, the success of pharmacogenomic prescribing will rest on the interface in which healthcare providers interact with the system. A well-designed clinical decision support system (CDSS) that compliments the physician’s normal practice is needed.”
More people exercising regularly
Would people exercise more if their physical activity tracking devices were gamified? Doctors X. Shirley Chen and Mitesh S. Patel explored this question in their article for Harvard Business Review, “Digital Health Tools Offer New Opportunities for Personalized Care.” The study described in the article compared a control group using a tracker without a game to groups using trackers with games designed to focus on either competition, support, or collaboration. The group using the competitive game crushed it, being the only group who continued their increased physical activity in the months after the game was over.
However, the authors were quick to point out that although the competitive game was the clear winner at the group level, a follow-up study showed a more complex answer at the personal level, “We found that taking an individual’s behavioral and physiological characteristics into account led to dramatically different results.” The authors advocate using “behavioral phenotypes” that are based on the way a patient uses technology to customize digital health tools to encourage changes in patient behavior.
Better chronic disease management
In the September 2021 Frontiers in Pharmacology review article “Machine Learning Techniques for Personalised Medicine Approaches in Immune-Mediated Chronic Inflammatory Disease: Applications and Challenges,” authors Jungjie Peng et al. summarize studies that use machine learning (ML) models in researching chronic inflammatory diseases such as multiple sclerosis, autoimmune chronic kidney disease, and inflammatory bowel disease. The massive growth in volume of electronic health records and other health data types and the improvements in health data standardization have made such studies possible. The goal of many of these studies is to identify the biomarkers that indicate the presence or progression of these diseases.
Machine learning models have several applications:
- Precision diagnosis. A complex condition such as chronic kidney disease has many potential causes, and machine learning can help identify and classify patient data according to disease subtypes. This assistance has the potential to prevent missed diagnoses and reduce invasive diagnostic procedures, such as biopsies.
- Precision treatment. Advances in laboratory technology means that the different ways the same disease affects different people can be precisely measured. When machine learning is applied to these precise measurements and corresponding treatment outcomes, it can give clinicians a more informed view of which treatments are likely to work on a given patient’s disease.
- Drug research. By quickly identifying small similarities in how different diseases affect the body, machine learning can help researchers identify new applications for existing drugs.
The depth and breadth of data required for effective machine learning approaches presents many challenges. For example, small sample sizes can mean that models that performed well in a study scenario do poorly in a real-world clinical setting. Datasets that lack diversity can train machine learning algorithms to discriminate. The article’s authors conclude, “Despite several challenges which might impede some of the ML applications in clinical research and practice, the contribution of AI [artificial intelligence] and ML techniques to personalized medicine for improved patient care is no doubt revolutionary.”
New ways to pay for drugs
The targeted medications that result from genomic research don’t come cheap. As more of these medications become available, healthcare payers will have to rethink how they approach reimbursement for prescriptions that are too new or too narrow in application to be addressed with existing data models for payment. In a recent article for Future Healthcare Journal, Sanjay Budhdeo et al. advocate for what they term “precision reimbursement” to address rising drug costs.
Precision reimbursement uses machine learning on a common dataset that is accessible to payers, providers, and pharmaceutical companies to identify patient risk factors for disease progression or treatment failure. Providers use this data to inform clinical recommendations, and payers and pharmaceutical companies use this data to establish the value of a medication for an individual patient. This more granular form of value-based healthcare requires significant investment in data management and infrastructure across all healthcare organizations, but promises an equally significant return. “Patients will be the greatest beneficiary of precision reimbursement, as the response to treatment models can be used to tailor treatment choice for each individual,” says Budhdeo et al.
Cures for specific cancers
In addition to working to prevent Alzheimer’s disease and type 2 diabetes, the Indiana University Grand Challenge Precision Health Initiative has targeted 3 cancers for research toward a cure: multiple myeloma, triple negative breast cancer, and pediatric sarcoma. Each research team is guided by the precision health principle of understanding disease through examining differences in patients’ genetics, environment, and lifestyle. For example, the myeloma team is doing a study on how a patient’s weight loss affects the presence of the MGUS (monoclonal gammopathy of undetermined significance) protein, which can develop into multiple myeloma. Another effort combines state cancer registry data with community data to discover how environmental factors and other social determinants of health affect myeloma patient treatments and outcomes.
Faster detection of hospital-associated infection outbreaks
Traditional hospital infection prevention depends on clinicians alerting an infection prevention team to investigate when multiple patients develop a similar infection. This labor-intensive process relies on a clinician first noticing this commonality and flagging it for further investigation, which means there could be delays in investigation if the infected patients are not being treated by the same people.
Scientists from the University of Pittsburgh and Carnegie Mellon University aimed to speed up the process of detecting outbreaks of infectious disease at hospitals by combining machine learning analysis of patient records with genome sequencing of infection specimens. When the system, named EDS-HAT, found two infection specimens that were genetically alike, it would examine the related patients’ hospital records to try to find connections, such as the same location, procedure, or provider. The scientists ran the EDS-HAT system at a 6-month lag against data used with traditional infection prevention methods at a single hospital. The results, published in November 2021, estimated that running EDS-HAT in real-time would have prevented an additional 63 disease transmissions and saved the hospital $692, 500.
Proactive primary care
Humanwide was Stanford Medicine’s 2018 pilot project to apply precision health principles in a primary care clinic setting. A 2021 BMC Family Practice journal research article studied the outcomes of this project to understand how such principles might be applied on a wider scale.
The Humanwide program expanded typical primary care offerings in the following ways:
- Digital health tools. A blood pressure cuff, digital scale, glucometer, and pedometer app supplied health measurements taken at home to each patient’s electronic health record (EHR) at the clinic.
- Pharmacogenomic and genetic testing. This testing looked for gene-drug interactions and risk for breast and ovarian cancer, Lynch syndrome, and familial hypercholesterolemia.
- Team-based care approach. Genetic counselors, pharmacists, health coaches, and dieticians were all part of the onsite clinical team.
All patient participants had some change in their care as a result of this program, such as medicine adjustments as a result of pharmacogenetic testing or digital health monitoring, and both patients and providers successfully adopted the precision health principles. To expand this program to a larger scale, however, would require addressing the care coordination, provider training, technology integration, and cost gaps that were highlighted even in this limited program. The researchers Cati G. Brown-Johnson et al. concluded, “On the cusp of a new type of medicine that is personalized and focused on prevention and wellbeing, it is critical to focus on implementation science—the how of integrating cutting-edge approaches into the clinical mainstream.”
Greater patient engagement
In October 2021, the Personalized Medicine Coalition launched the More Than a Number website to provide information to patients about genetic testing, health insurance coverage, clinical trials, and other related topics. This type of nuts-and-bolts information and focus on the core patient/provider relationship is especially valuable for patients who are dealing with long-term disease. But broadening the view of personalized medicine to encompass the vision of precision health requires all participants in a healthcare system to recognize their role and the role of the communities they belong to in improving health outcomes for everyone.
To learn more about how hc1 views the future of healthcare, watch Our Vision for Precision Health video.