This is the 4th post in the Health Information Economy series. We are outlining a number of elements outlining a vision for a robust health information economy as we put together a submission for the Economist-Innocentive Health Information Economy Idea Challenge
In our previous post, we outlined the rationale for having a Personal Vitality Score. Here, we will outline the components that could be used to create one.
Objective of the Personal Vitality Score
The goal for the Personal Vitality Score is to create an objective, up-to-date score that summarizes the health of an individual. Much as a FICO score measures credit-worthiness, an objective individual-level Vitality Score can deliver a snapshot of where a person is and what services/ products will most help them to improve their health. We believe that a Personal Vitality Score -type currency is required to create the metrics for a health system that would focus on improving the health of an individual (and populations).
Components of the Personal Vitality Score
A Personal Vitality Score would summarize the health status of an individual. It would need to combine multiple elements:
- Current state of an individual’s health, based on their individual health information
- Programs and treatments the individual is enrolled/ engaged in
- Projected health of the individual given (1) and (2)
1. Individual Health Information
The Individual Health Information dataset provides a current picture of the individual’s health. For it to do so, it must combine what is known about the person’s current status with a mix of how they live and their medical history. Far beyond the hospital-centric EMR, this dataset is a synthesis of all the factors known about this individual that impact their health. It combines information across:
a) Daily Activities and Lifestyle
b) Personal Characteristics and Abilities
c) Medical History
For this snapshot to be relevant, this information must be updated in a regular fashion. A data collection infrastructure to capture and synthesize relevant information must be put in place for this comprehensive snapshot to come together. We’ll give some examples of what this may look like below:
1a. Daily Activities and Lifestyle
This category captures the day to day “healthstream” of choices made by an individual and activities they participate in. We know that things like diet and exercise have a tremendous impact on health status, and this category enables us to capture that information. Likely sources for this information include:
- Quantified Self and Activity tracking websites like DailyBurn, Nike+, BodyBugg, diet logs, etc
- Health profile updates from profiles on medical sites like PatientsLikeMe
- Wireless health data including sensors, scales, and monitors
- Grocery purchases logged on credit cards or store loyalty cards
- Health club visits and digital workout logs
1b. Personal Characteristics and Abilities
We must know an individual’s baseline to appropriately understand their risks and their potential. This category captures the underlying characteristics of the individual and their current capabilities, including:
- Genome and genetic markers
- Physical and mental capabilities: IQ test, reading levels, abilities to perform activities of daily living, fitness, visual acuity, etc
- Molecular performance: activity levels or protein expression of various enzymes and protein markers
- Life skills
- Weight/ body composition
1c. Medical History
The presence of disease markers or diagnosed conditions can significantly impact health risks and future capabilities. This category captures diagnosed conditions, disease markers, family history, and other relevant components of the medical history. Sources could include things like:
- Personal Health Record
- Electronic Medical Record
- Aggregators like Microsoft HealthVault or Google Health
- Profiles from disease communities like ACOR or PatientsLikeMe
2. Program / Treatment Enrollment and Engagement
The programs and treatments being utilized or considered are elements that can change the trajectory of someone’s future health. They may impact the individual by improving their capabilities or reducing their risks. Today, programs and treatments are often reviewed in isolation. In tomorrow’s world, the total impact of the program to the individual, which may be measured by the Personal Vitality Score, is what will matter. Impact will depend on the:
- potential improvement and side effects of the program or treatment chosen
- context of the individual
- degree of engagement
- interaction with other programs or treatments
As we better understand the relative impact of various programs and treatments on individuals, we will finally begin to tap the potential of comparative effectiveness research and understand the opportunity cost of choosing one program over another.
The set of data captured in this database includes:
a) Participation in programs and treatments: CPT codes for procedures, prescription data, use of herbal medicines, treatments by allied health providers, etc
b) Engagement and adherence: Prescription fill data, office visits, program specific metrics, self-reported participation, etc.
c) Options considered: List of alternatives under consideration
d) Treatment interaction: Likely interactions between treatments and programs in which individual is enrolled, cumulative burden of all treatments and programs enrolled, etc.
3. Projected Individual Health Outcomes
Creating a personalized projection for future health capabilities and risks must mine what we know from longitudinal studies and treatment databases and combine it with what we know about the individual.
The approach required to create a projected individual health outcome is the opposite of the one used to create a clinical trial. In the clinical trial, we look to isolate a variable across a number of people. In this projection, we look to drill down into the segment specific data for all the conditions and treatments pertaining to the individual and synthesize them.
This category will require development of detailed, individualized algorithms that can mine deep datasets including:
- Segmented population databases enabling projection for the individual based on a number of relevant factors (e.g., Framingham)
- Deep longitudinal datasets from disease-oriented communities (e.g., PatientsLikeMe, ACOR), clinical trials, payers, and providers
- Segmented or individualized clinical trial data (e.g., enabling drill-down by various factors)
- Provider-specific and treatment-specific outcome data
Vitality Algorithm
The Vitality Algorithm pulls these three disparate streams of information together into one comprehensive score with a number of sub-elements. The algorithm is likely to be a work in progress — updated as information needs and relative importance change over time — and so a structure that balances an accepted standard that may be replaced with disruptors is likely to be required. It is unlikely that a structure based on consensus or a scoring system based on a monopoly held by a nonprofit or government will be dynamic enough to keep up with the advances. Therefore, a dedicated scoring model that gains share but faces multiple disruptors (e.g., FICO) is likely to become the best answer over time.
Synthesis
The Personal Vitality Score creates a synthesis of a person’s health by combining in one score: 1) a personalized view across all known health data, 2) what we know about the individual, and 3) what actions the individual is taking.
By packaging all these elements together, we get a robust picture of an individual’s health, where it is going, and what can and is being done to improve it. Abstracting a level, we can create a Vitality Index across a population, which would help to allocate resources in the most effective way.
The creation of this Personal Vitality Score enables us to measure value creation in health and could create an ecosystem with the appropriate incentives to improve health in a given individual or population. What that new ecosystem looks like is the subject of our next post.
