Can digital health become a true medical therapy?

I quit my job in the pharmaceutical industry in 2011 to found 100Plus because I think technology will have a greater impact on improving people’s health than any new drug I could have worked on.  It is still very early in the digital health revolution, and it remains to be seen which areas of digital health will find consumer adoption and successful payment models.  However, there are now a few digital health companies emerging that are addressing areas of unmet medical need, focusing on key health outcomes measures and generating preliminary data on efficacy and cost effectiveness. In this sense, they are beginning to look a lot like early biotech companies.

I tend to think of digital health companies on a simplistic 2-axis framework of target population (from healthy to managing chronic disease) and payment type (from traditional (payers and employers) to direct consumer payment).

The companies focusing on specific chronic conditions and traditional payer models are the ones who are starting to look like early biotech companies and will follow a similar trajectory if successful.

Pharma industry primer

When most people think of how the pharmaceutical industry works they think of chemistry and biology.  Pharma companies study the underlying biology of different diseases, then develop therapeutics (small molecules or larger proteins) to change some step in the disease process and improve the underlying diseased physiology.  In my experience this is only one of the core competencies of most pharma companies and is actually a diminishing focus.  The rapid rise in biotech companies over the last two decades, who ultimately partner or sell to big pharma is proof of Pharma’s diminishing reliance on this competency.  Big Pharma companies have become mature distribution channels for medical products. Biotech companies serve as outsourced R&D and pharma companies essentially option them around Phase 2 (with proof-of-concept data established).

Thinking broadly, the art of drug development is more about data than chemistry.  A marketed drug is actually a compound (or protein) surrounded by a robust data package.  While drug development and commercialization starts with chemistry and biology, the real competency of pharma is generating a data package (pre-clinical, pharmacokinetics, pharmacodynamics, clinical efficacy, clinical safety, outcomes, etc) that facilitates approval, marketing and reimbursement for the drug.  A modern day big pharma company is more adept at designing and executing clinical studies, gaining FDA approvals, gaining reimbursement from payers, marketing to physicians and distributing to pharmacies, than chemistry or biology.

Digital health as the next biotech

Most successful biotechnology companies have a core expertise in some underlying science and use that expertise to generate therapeutic candidates; most, but not all, generally focus on one therapeutic indication (like type 2 diabetes).  They begin to generate data on their candidates to show safety and efficacy.  If the data look promising, they partner with a Big Pharma company (around Phase 2) to help design and execute late-stage clinical programs, plan for FDA approval, plan reimbursement strategy and marketing strategy.  They get an upfront payment and a share of revenue, but generally turn the reigns over to Big Pharma.

Now lets substitute an app for the molecule in the example above.  Create an app that is focused on improving HbA1c in type 2 diabetics, begin to generate clinical data that the app is safe and effective, distribute through physicians and seek reimbursement for the product via traditional payors.  This is WellDoc, a digital health company based in Baltimore.

There is a group of digital health companies who are beginning to follow a traditional biotech model: therapeutic focus, generating clinical data via randomized clinical trials, distributing via physicians and seeking reimbursement based on pharmacoeconomic data.  WellDoc is focused on type 2 diabetes, has done a clinical trial demonstrating a 1.2% reduction in HbA1c (over usual care) and has recently launched a prescription-only app with limited reimbursement.  Propeller Health, focused on respiratory diseases, has a 500 patient trial ongoing to prove efficacy and cost-effectiveness, has a dozen paying commercial programs (with a mix of payers, integrated health systems and at-risk medical groups) and will distribute through physicians.  Ginger.io is working with physicians to manage a number of therapeutic areas including depression and diabetes, and has pilots ongoing to generate data on outcomes.  Omada Health has run a study comparing their program outcomes to those typically seen in face-to-face versions of lifestyle programs for pre-diabetics, and is actively being paid by traditional payers.  Finally, Qualia Health, a new startup out of the University of Chicago (my business school alma mater) is taking this same approach in Chronic Heart Failure.

In the future, it should not matter to pharmaceutical companies, payers, physicians or patients whether an intervention for a specific disease is chemical or technological.  The only thing that matters is whether the data package generated on that intervention proves it is efficacious, safe and cost-effective over the long term (meaning patients have to use it long-term).  In the near term I believe we will see pharma companies begin to embrace these new technologies and work with these companies to develop robust data packages for approval and reimbursement, then distribute them via their existing massive sales organizations.  This may create a viable distribution option for many disease-focused digital health products. This also creates an opportunity for a few early companies to try to build the next generation of pharma company – fully integrated, data driven, reimbursement focused with both development, marketing and sales.  But these companies will develop technology instead of chemistry.

The Promise of User-Generated Health

Recently I have been thinking a lot about the concept of user-generated health.  Some also call this Interactive Health and some just call it Health 2.0.  I can’t keep up with all the labels, but what I am thinking of is when we, as users of the healthcare system, start adding as much value to the system as we receive.  We will do this by tracking personal health data, sharing our data with phsyicans and peers, connecting with others like us to share advice and experiences and by generally being more engaged with our health and bodies.  I think of this like a bank account.  Even if we are not withdrawing value from the system today (if we are generally healthy), we will eventually be net withdrawers, so we should all start adding value to our accounts today.

Below are two talks I gave this year on this topic.  I think there are a few themes that are consistent through both:

1. The biggest problem in healthcare today is not medical; it is behavioral.  Most of us have become incredibly detached from our bodies and health and being detached leads to poor health.
2. Increasing engagement should be the key goal of any modern day health company.
3. Patients have a tremendous amount of value and information to add to the healthcare equation that has been missing up to this point.
4. All of us will start tracking more and more data about our bodies and health, whether we know it or not (think iPhone background monitoring).
5. These data we collect will transform the relationship we have with our bodies.  Data will also change how we interact with the healthcare system and fundamentally change the relationship we have with our physicians.
6. Data lead to personalization and relevance.  Sites based on a foundation of personal health data will be able to provide superior services and experiences to users.

TEDx Silicon Valley talk entitled “Data as the next blockbuster drug” from May 14, 2011 (length: 7:21)

Slides from a talk to Stanford CS247 Human Computer Interaction: Interaction Design Studio entitled “User Generated Health” on 2/11/11

The Future of the Health Graph (Health Social Layer) – Part 2



In Part 1 of this post I presented the argument that there is currently no true social layer in health. I think that a true health network layer (the Health Graph) will evolve that organizes users based on their health profile from multiple data points and will allow groupings based on a high degree of similarity between users (and thus relevance).

Personal health data exists today in many digital silos, and increasingly, APIs are being opened to allow app developers to provide services to users based on their personal health data.  Microsoft Healthvault and Google Health were designed as platforms from the outset, and recently, EMR vendors (such as PracticeFusion) have announced intentions to open APIs allowing development of personalized applications for physicians and patients.

While current applications built on these platforms have some degree of personalization based on an individual’s data, they are still mostly one-to-one connections between one app and one individual.  With the addition of the Health Graph, ‘social health applications’ will become possible across these platforms, greatly increasing value to users.

I know the common Health Graph API and Login are a long-shot, but it is the way things should be, so I hope it will happen someday.

 

Applications of the Health Graph

I think the most immediate application of the Health Graph is the ability to target content or questions from one user to another based on a higher degree of similarity and relevance than is possible in today’s one-dimensional “health networks”.  However, I think the real value of this Health Social Layer will come from the aggregation and sharing of personal health data and health content among personalized peer groups that are based on an individual’s health profile.  The Health Graph will allow applications to automatically segment users into meaningful peer groups, based on what the app is trying to accomplish.  Below are a few thoughts on features and applications that could be built on the Health Graph:

Q&A

Rather than posting a question to the appropriate forum or site, users will have health questions targeted to users with very similar health histories and characteristics increasing the relevance of the response.  A response will carry more weight to the user if they know what they have in common with the respondent.  For instance, a response about a potential therapy for depression is valuable.  But a response about this therapy from another user who has the same symptoms, history of depression, age, gender and sequence of therapies is much more relevant.  An analogy from standard Q&A sites is the value of Quora or Aardvark versus Yahoo! Answers.  I am an expert about my own body and health history, so if we share a very similar health history, my ‘expertise’ is relevant to you, and vice versa.

Surveys and Statistics

In addition to the type of Q&A described above, users will be able to ask simple quantitative questions that can be targeted at the appropriate audience.  Again using a depression example, if a user wants other users to rate the effectiveness of a given treatment, it is useful to have a result from a broad audience, but it is much more relevant if the answers are filtered to only users highly similar to the initial user.  Using this approach of targeted quant questions, an app could generate highly relevant data and statistics of how a given user responds relative to their peer group, and relative to the population as a whole.

Treatment Recommendations

I remember reading a series of Q&A posts about a certain drug from a popular health Q&A site.  User #1 was asking if this drug was appropriate for them, and if others had experience with the drug.  He listed a pretty detailed medical history in his initial post.  After numerous posts on whether the drug worked or did not work for respondents, but without context, another answer jumped out.  This respondent went on to describe a very similar medical history to User #1 (which was quite complex), and then gave a very detailed description of their trial with the drug (which was positive).  The relevance and weight of that post was very impressive.  The Health Graph will enable this level of relevance for all questions and recommendations.

Health Games

It seems to me that it is a lot more fun to compete with others when I know a bit about them.  I think FitBit has done a nice job on this front by allowing users to see group data on activity level, but allowing filtering by important characteristics: age, BMI, etc.  This peer group is relevant to me, but comparing my activity level to an 18 year old marathon runner probably is not.  True peer groups open many opportunities to add game mechanics into health apps.

Social Tracking – Longitudinal Data Collection by Peer Group

I think one of the most powerful outgrowths of the Health Graph and ability to define highly relevant peer groupings will be the ability to track specific pieces of data over time.  I have written previously about the rapid rise in health trackers facilitated by smart phones.  While most of these trackers are for the individual, the power of social tracking far surpasses individual tracking.  With a well defined and relevant peer group, I think the value of tracked data will grow exponentially.  In addition to seeing how you are trending relative to your peer group, the Graph makes it possible to compare two groups who are essentially equivalent, but differ on one key characteristic: drug A vs. drug B, migraine with aura vs. without aura, frequent exercise vs. infrequent, etc.  With enough data on each user, over time these could become well matched and well controlled studies.

Please let me know of your thoughts on these concepts, or other potential uses for the Health Graph …

The Future of the Health Graph (Health Social Layer) – Part 1

Lack of a True Health Social Layer
There is currently no true social layer in health akin to the Social Graph for social connections. Most of the successful health networks or health communities that exist today make connections based on one significant characteristic of users, typically one specific medical condition (e.g. Type 1 Diabetes or ALS). What needs to evolve is a true health network layer (the Health Graph) that organizes users based on multiple data points and connects people based on a high degree of similarity between users (and thus relevance).

People typically organize themselves (offline and online) based on the number of things they have in common, or some specifically strong tie (i.e. family):

Since we cannot see health characteristics and experiences, we cannot see whom we should ask for advice, insight or understanding of health questions.  Of course one of the great promises of the Health 2.0 movement is that technology will facilitate this ability in health.

Turning Health Data into Information – The Health Graph

I previously wrote a post about the details one can learn about a person from a few health data points, in that case medication history.  Taking this concept a step further, I think that with a few simple data points from each individual in a population (i.e. age, gender, BMI, medical conditions, medications/treatments, medical events, or any subset of these data), it is possible to map any individual in context (similarity and dissimilarity) to all other members in that population.  Mapping users based on heath data is facilitated because many of these data points are inherently connected:

In the example above Zocor (simvastatin) (drug) implies high cholesterol (medical condition), which implies cardiovascular disease (disease system).  Lipitor (atorvastatin) also implies high cholesterol and cardiovascular disease.  A heart attack (event) implies coronary artery disease (medical condition), which implies high cholesterol, which implies cardiovascular disease. So, someone on simvastatin is inherently related to someone on atorvastatin, and both are indirectly related, (whether they like it or not) to someone who has had a heart attack (severe manifestation of same disease).

Due to the inherent connectedness of these data points, a meaningful health profile can be built for a user with only a few points and some educated assumptions.  Of course then you can layer in relationships between other medical conditions and data points, and begin to include attitudinal characteristics.  The more points you have in common with another user, the more similar you are to that user, and the closer you are to that user in the Health Graph.  This Health Graph will become stronger and more robust with each additional user and/or data point added.

Once you know these basic health characteristics for each member of a population and map them in relation to other members, you can create a true network, a true health social layer, and a strong platform for personalized services, features and “social health applications” (discussed in greater detail in Part 2 of this post).

One of the key benefits of the Health Graph is the ability to create meaningful peer groups based on multiple dimensions. As an example, knowing only a user’s chronic medical conditions allows segmentation into meaningful ‘verticals’.  In the diagram below, even though all users share some attribute with another user, users 1 and 3 are more similar to each other than to 2,4,5.  Note that non-shared characteristics (diabetes) also differentiate patients.

Limitations of Current Health Networks

I think it is interesting that the most successful health networks to date (PatientsLikeMe, type 1 diabetes networks, oncology networks such as ACOR, etc.) are really targeting the “tail”, meaning they are indications with small numbers of patients (compared to high cholesterol or type 2 diabetes), but where that one indication has a significant impact on the user’s life.  For these patients, these networks are imensely valuable.  But for patients without this ‘defining health characteristic’, no true network exists.

In truth, however, the majority of people are complex, made of many equally relevant factors that make up the whole. The successful companies targeting this complex audience have historically focused on content (WebMD) or on Q&A forums (MedHelp), where the user must know what they are searching for or to which ‘forum’ to post a question.

Future of the Health Graph

In the future, the health info needed to add a new user to the Health Graph will be pulled down effortlessly from some other source (like HealthVault, Google Health, Blue Button Initiative, PracticeFusion, etc.), but for now, this is a non-trivial task, and early adopters must enter this data.  But I am pretty sure that whoever can attract the early adopters, begin to build this Health Graph and build engaging features on top of the Graph will be pretty successful when the second wave of users can join with the simple push of a Blue Button.

Flatlining: Can Health 2.0 be Resuscitated by Access to Prescription Drug Histories?

Where is the firehose of data we were promised?

Access to personal health data is the foundation requirement for any movement toward patient engagement. Don’t get me wrong, access to the data itself will not solve any problem.  In fact, in the early stages, access to vast amounts of personal health data will probably create a lot more problems than it solves.  However, with widespread access to personal health data, enterprising entrepreneurs will have the keys to provide users with highly customized and personalized  health-related experiences, or such is the much hyped promise of Health 2.0.  And when I say access, I am talking about simple and seamless access to my data.  I recently tried to link my Google Health account to Walgreens Pharmacy.  It took 8 minutes to complete a profile at Walgreens.com, and I still did not get my data.  It took another few days for the data to drop into GHealth.  I am pretty sure I clicked a release allowing Google to send a fax to Walgreens for permission.  That would be pretty funny if it wasn’t so sad.  Call me lazy, but I think I deserve an OAuth for my data, where, as a user, I simply need to provide permission and credentials.  Then let the app do the rest.

Given the current state of access to health data, most ‘successful’ Health 2.0 companies are limited, and will continue to be limited, to very early adopters.  Health 2.0 will not succeed on any meaningful scale as long as apps require users to input their own health data to receive a service in return.  Most users are not that motivated (some might even say lazy like me), most do not have access to their information and many do not feel comfortable enough with the data to use it.  The watershed for Health 2.0 will come when big data starts to flow effortlessly into consumers hands in huge chunks.

The Data Watershed: E-Prescribing and Medication History

Fortunately, I think this data watershed may be coming.  Today, there are only a few sources of big data in healthcare.  Data is captured in medical charts (or EMRs for the select few), it is captured in medical claims databases and it is captured as prescription drug histories.

Due to the rise in e-prescribing, the first of these firehoses of data to reach patients will probably be prescription drug histories.  E-prescribing is pumping digital data into the system in large and growing quantities.  And in March 2010, Google Health announced a partnership with Surescripts, the largest e-prescribing network in the US, to make it possible for millions of users to access to their medication histories.

The Importance of Medication History

So, what’s the a big deal? Medication history is only one small part of the complete data package, right?  Wrong. Medication history is the backbone of a patient’s medical history, and, for better or for worse, pharmaceutical use is the foundation of our current medical system.  In the vast majority of cases, when a patient gets sick, they go to the doctor, get diagnosed, and get a prescription.  So a patient’s medication history gives a detailed, longitudinal medical history for the patient.  One just needs to know how to translate it.

Below are a few examples showing simplified medication histories for three hypothetical patients.  Let’s see what we can learn about these patients from these simple data points.

At first glance, patient #1 has a pretty random assortment of medications, including some pain medications, a prescription stool softener (Doc-Q-Lace), a 4 times a day antibiotic and hemorrhoid medication.  But any woman who has given birth probably recognizes this assortment of medications as a pretty typical post-childbirth discharge list.  All of these meds prescribed on 3/20/08 tells us that this is a woman of childbearing age who gave birth on 3/18/08 +/- 2 or 3 days.  We also know from the Diclo antibiotic that she breastfed her child for at least 2 months, and had mastitis, a common breast feeding infection.  I bet that Diapers.com would like this information.  Or possibly Mattel, knowing that this woman now has a two year old child?  How about the entrepreneur who is developing the next BabyCenter.com app for new moms?

Patient #2 has a pretty straight-forward and easily identifiable medical history.  The point of this example is to show the depth of personal information that can be derived from five simple data points.  We know that this patient is HIV positive and began treatment at the end of 2006.  We also know that they are probably not doing well, as they have recently failed on their first-line antiretroviral therapy and have progressed to second-line treatment with a protease inhibitor (Reyataz).  We can also make some assumptions, based on demographics, and assume there is about a 75% chance this patient is male and make assumptions on age.  I would have to assume that this patient would want to strictly control who has access to these five data points.

Patient #3 is a little more complicated, but is a pretty typical patient with cardiovascular disease.  We know this patient has high cholesterol (Lipitor), high blood pressure (Norvasc) and type 2 diabetes (Metformin).  We know from the pretty rapid intensification in oral anti-diabetic therapy (Metformin 500 to Metformin 1000 to Janumet, which is a combination of Metformin and Januvia) that this patient’s diabetes is less well controlled than the average type 2 diabetic.  We also know their high cholesterol is unconrolled on a standard dose of Lipitor, due to the increase to Lipitor 80.  The addition of Coreg (a beta-blocker), Plavix (an anti-platelet drug) and Altace (an ACE inhibitor), all on August 15, 2009, leads us to believe that this patient had a major cardiac event, most likely a heart attack requiring percutaneous intervention (PCI) and a stent to open a clogged coronary artery, as this list is a pretty common myocardial infarction discharge list.  What else can we infer from this data?  We can be pretty sure this patient has good health insurance, and if they are on Medicare, they are probably in a Medicare Advantage plan, due to the number of branded medications and lack of generic substitution (i.e. generic simvastatin vs. branded Lipitor).

The Good and Bad of Medication History

I believe widespread access to medication histories will infuse new life into Health 2.0 and the entrpreneurs working in this field.  Any of these entrepreneurs not actively thinking about how to use medication histories in their apps are missing the boat. With easy access to medication history, users can give permission for App X to access their medication history (hopefully with one click), and receive highly personalized services in return.  Keas is beginning to do this with taylored Care Plans, but I am waiting for the apps that look at my medication history and tell me other drugs I can take (to save money), whether my physician is following guidelines, news I should read, and FDA alerts I should know about.  I am waiting for the app that connects me to similar patients based on our medication histories, the app that looks at my history and asks simple, taylored questions to help me further populate a more robust PHR, and the app that pays me to allow pharma companies to see my de-identified medication history and ask me questions about my data (pharma loves medication history information, just ask IMS Health, which recently sold for $5.2 billion).

But, of course, with this much personal data coming into the mix so quickly, there are significant privacy risks coming too.  The same algorithms used to translate a medication history into a health profile for customized applications can also be used by others to learn deeply personal facts about individuals.  So, with all the good that will come from access to medication histories, the first major privacy breaches and big scare stories are probably coming as well.  Get ready, the Health 2.0 ride is getting more interesting.

Patient-Reported Data: FDA Cares, So Should You

The FDA has long recognized patient-reported data in clinical trials.  Many widely used drugs have been approved primarily based on patient-reported improvements.  Migraine drugs are typically approved using patient diaries measuring levels of pain relief, phonophobia, photophobia and nausea; the primary endpoint of depression trials is the change from baseline in a standard depression rating scale; Pfizer’s Lyrica (pregabalin) showed efficacy in neuropathic pain based on a difference in average self-reported pain scores between patients treated with Lyrica and placebo; and the list goes on.  Whether FDA’s acceptance of these patient-reported endpoints is due to FDA’s belief in the value of patient-reported outcomes (PROs), or due to the lack of a ‘harder’ endpoint for these trials is debatable.  However, below is a selected list of drugs primarily approved based on PROs, and their peak WW sales.  One can see that relying on patient-reported clinical data has not hindered the success of these products.

FDA Guidance on Patient-Reported Outcome Data

In December 2009, FDA issued a guidance document instructing industry on how to use PROs in drug development and clinical trials.  Although FDA has historically accepted patient-reported outcomes for approval when a more objective endpoint is lacking, they have been very reluctant to include secondary PRO endpoints (such as improvements in quality of life) in approved product labeling.  Some in industry have viewed this new guidance as sign of a shift within FDA toward the importance of patient experience and quality of life measures.  If true, this could have a broad impact on the way drugs are designed, developed and marketed.

To date, PRO data has been underutilized in clinical practice.  However, if FDA begins to recognize PRO measures such as quality of life, pharma companies will begin to generate data on their drugs’ impact on these measures. Physicians will have additional data available when deciding which drug might be best for their patients and will begin to rely more heavily on PRO data.

Importance of PRO Measures

There are many reasons to value PRO measurements.  Some treatment effects are only known to the patient, and the patient provides a unique perspective on treatment effectiveness that goes beyond traditional clinical measurements.  European pricing authorities have long valued the impact of quality of life on drug evaluation, using the quality-adjusted life year (QALY) measurement of disease burden, which includes both the quality and the quantity of life lived.  But even for those who do not believe a focus on the patient is reason enough, there is evidence of the clinical importance of PRO measures.  A study of 293 heart failure patients showed that patient-reported functional status was a prognostic predictor of hospitalizations, quality of life and death; and a meta-analysis of oncology research from 1982-2008 showed a correlation between patient-reported quality of life and survival in cancer patients.

Opportunity for mHealth

All of this, of course, creates a tremendous opportunity for companies working to collect patient-reported data, and I have previously written about the value of longitudinal personal health data streams, or healthstreams, that include both physician/system-generated data (EMR), combined with more frequently added patient-generated data.  Traditionally, patient-reported data has been captured disproportionaley via written diaries.  However, since the most common measures for a PRO instrument are based on the Likert scale and Visual Analog Scale (VAS), it seems clear that any PRO data commonly captured via written patient diaries can be more easily and efficiently captured by electronic means, and work is being done to validate electronic capture methods.

In the future, we will see this type of PRO data regularly captured by patients via mobile devices and mobile apps, and this data will become a common tool for physicians in clinical practice.  Some companies, such as Ringful Health and ReliefInsite (recently acquired by PatientsLikeMe), are already moving in this direction with mobile versions of asthma diaries and pain journals.

Patient-reported outcomes data is valuable and gaining importance in the practice of medicine.  When you see your first pharmaceutical ad featuring claims of improvements to quality of life, you will know the era of patient-reported outcomes has arrived.

Completing the Picture: The Value of User-Generated Health Data Streams

The healthcare industry is undergoing a dramatic and necessary transformation. Empowered patients are taking action and participating in their own healthcare, marking a significant change from the long-standing paternalistic relationship between physician and patient.  Underlying this transformation is a new found access to health information via the internet and personal health data via consumer health devices and applications.

Much has been written about this long coming transformation, which has spawned the Health 2.0 movement, a successful conference series of the same name started by Mathew Holt and Indu Subaiya, the peer reviewed Journal of Participatory Medicine and a series of grassroots unconferences championed by Mark Scrimshire.

To date, many have focused on the challenge of digitizing the vast amounts of health data in the current health system and on the one-way transfer of existing data from the system to the patient via EMRs and PHRs.  Google and Microsoft have launched data warehouses for personal health information, and a few large institutions have embraced this one-way flow of information. The Mayo Clinic recently partnered with Microsoft HealthVault to provide patients access to their medical records, and Kaiser Permanente has been very successful with their “My Health Manager” offering.

Providing access to health information and health data is a logical first step.  The foundation of the modern health industry is health data.   Each time a prescription is dispensed (3.8 billion times in 2007), a medical claim is submitted, or a lab is ordered, a piece of personal health data is generated.  Evidence-based treatment guidelines often focus on specific health data values, such as BP, LDL or HbA1c, for treatment recommendations.  And a modern medical record is simply an aggregation of health data generated at the point of care, including patient reported symptoms, diagnoses, treatments, test results, physician commentary and observations.  However, medical records are punctuated data streams with large gaps in data from time periods in between physician visits.  It is the doctor’s difficult job to fill in these gaps with imperfect information recounted by the patient to complete the picture and treat the patient effectively.

Focusing only on the EMR, or the one-way transmission of personal health data from the system to the patient, misses half of the health data equation.  With increasing frequency, patients are beginning to generate and report their own personal health data.  Sites such as CureTogether and PatientsLikeMe are collecting patient reported data, analyzing the data and generating valuable services for their members.  And recently, we have seen a rapid rise in ‘health trackers’ to record health and wellness information (MedHelp, TheCarrot, DailyMile, hLog, Polka, Ringful).  Data is also being collected and uploaded in the home by personal health devices such as BP monitors with SD cards, WiFi weight scales, USB glucose meters and activity monitors, some now sold at Best Buy.  To date, these patient generated data streams have existed in silos, both disconnected from traditional data streams generated within the current health system and from data generated on other sites.

In the future, each individual’s traditional health data stream (EMR) and user-generated data streams will merge into one personal health data stream, or healthstream, and this healthstream will be owned solely by the patient.  Diagnoses, treatment decisions and lab results will be overlaid on top of more frequently added data from the patient including symptoms, adherence to medication, daily weight, diet, mood, activity levels, sleep patterns, etc.  This complete longitudinal data stream will be valuable to both the physician and patient and become a part of everyday medical practice.  Physicians will prescribe trackers, apps and home health devices, along with their standard Rx, and will receive a report at the patient’s next visit.  Rather than relying on recollections, physicians will have more complete pictures of their patients, and patients will begin to feel like partners in their own care.

This longitudinal healthstream will become a valuable asset for the patient, and will increase in value with each added data point.  Patients will have the right to grant permission to use all or part of their healthstream for medical care, for research and discovery, or even for profit.  I will explore the topics of permission and use of the healthstream in future posts.

For another view of the future of the healthstream, see Dr. Vijay Goel’s post “HealthStreaming: What data would you need in your stream to make your health decisions?”