Building Healthcare's "Ground Truth" with AI

September 17, 2018

For pharma leaders, AI can bring about a new reality of seeing a multitude of possible paths, forecasting the interplay of multi-dimensional trends, and charting patient journeys that are based on reality, writes Aswin Chandrakantan.

Just like pilots in vertigo are guided to safety by calibrating their position to the “ground truth”, pharma leaders need to adapt to new means of seeing what they don’t see, make informed choices of where to go, and recalibrate proactively to reach their destination.

For decades, pharmaceutical leaders successfully inferred a narrow set of product journeys through cultivated instinct and highly sampled primary research, augmented by narrow visibility into the trillions of interactions that shape healthcare. But our markets are changing. They demand more specificity, more contextualization on the patient’s life and a deeper understanding of how disease is experienced by the patient. The days of simple heuristics and narrow research are over.

Forced into the crosshairs of policy makers and the broader public, pharma leaders are busy figuring out roads less traveled. A new reality, it requires new ways of seeing a multitude of possible paths, forecasting the interplay of multi-dimensional trends, and charting journeys that are based on de-facto reality, on “ground truth”. It is pharma’s new problem and AI’s sweet spot.

The promise of AI

Tackling healthcare’s ever-growing complexity, rate of change, and burden of disease requires the heuristics efficiency and speed that AI offers.

Consider this: Precision medicine and market pressures drive deeper and broader collaboration between healthcare and biopharma. Meanwhile, patients influence as consumers is growing, and our conceptual thinking about population analytics is shifting to cope with omni-channel, always-on, always-connected, AI-enabled mix of conversational and IoT (Internet of Things) sensory-device data streams.

The accretive effect of AI across clinical, science, and commercial domains opens up opportunities not previously practical nor economically viable outside computer science research labs.

With the cumulative effect of AI come new ways of seeing, understanding, and taking action: from “true-north” unmet needs and disease burden to figuring out and driving productive ecosystem collaboration, optimizing future supply with demand, overcoming conceptual and procedural barriers entrenched in traditional approach to drug development, rescuing phase III projects by hearing the ‘sound of silence’ of missing data, new and deeper understanding of diseases and diagnosis, and reducing healthcare access barriers, among many others. 

Failure to launch?

While computational methods and access to them have improved tremendously, there remain significant barriers to utilizing AI more deeply to address unmet medical need:

  • The rapid proliferation of healthcare data. As healthcare data continues to proliferate in fragmented and unexpected ways, leaders are challenged to work with either broad, but highly incomplete, patient journeys (so-called “open data”) or presumably complete, but massively narrow and biased journeys (“closed data”). Data aggregators and healthcare conglomerates have perverse incentives to prevent the linkage of data across systems. The resulting trade-offs have made it impossible for the industry to harness the full power of AI to identify and predict disease.

  • Managing re-identification risk. Nearly 21 years after US Department of Health and Human Services (HHS) released its final report “Analysis of Unique Patient Identifier Options”, the Centers for Medicare and Medicaid Services advises that “there is no adopted standard to identify patients.” As paradoxically or consequently as it may be, HHS has constructed formidable legal, procedural, and scientific barriers to assembling a coherent real-world de-facto account of the multiple interactions an individual patient had with the diverse stakeholders over time. But patient-centric high-fidelity data is the ultimate building block of any conceivable means intended to facilitate stakeholder collaboration on “healthcare-trip-planning”.

  • A talent war like no other. As companies have embarked on the journey to modernize analytic capabilities, the ability to recruit and retain top AI leadership continues to be an Achilles Heel for the healthcare sector. Technology and financial services companies bear significant advantages to recruiting top talent, largely driven by the availability and accessibility of data. As a result, any advantages borne by having access to data are often lost by not having the best talent to create advantage from it.

The barriers are high, but an imperative to adapt should come at no surprise as patient engagement is now the prevailing wisdom, patient out-of-pocket expenses rose 66% over ten years outpacing wage growth by 2X, accelerated market uptake of high-deductible health plans, and the pervasive affordability crisis, among others. The solution?

Build healthcare’s first “Ground Truth”

A new medium of patient-journey, AI constructs and tools to explore them would help leaders figure out ways to get better at what they do, venture into growth opportunities not otherwise plausible, and be successful at an ecosystem play larger than their own by seeing what we need to see but could not see before. The unbiased opportunity to predict unmet needs, formularies context, and outcomes. The capability to broaden understanding of actual care patterns relative to prescribed standards of care that forms the foundation of tackling disease burden.

The task: Re-construct discrete yet anonymized, not re-identifiable patient journeys threaded through trillions of “interaction bread crumbs” they had throughout the healthcare space over time.

Reconstructing and exploring patient-journeys from the enormously fragmented data healthcare produces “in-flight” is a non-trivial undertaking. It requires complex orchestration of massively-contextual analytics; crisscrossing highly diverse data not previously tapped for this purpose; enriching data not previously available in traditional data aggregation; and bringing together cross-functional clinical, bio-pharma, and data science experts to drive it with advanced AI techniques and tools.

Unlocking the keys to AI enhancement

The significance of patient-journey AI structures to healthcare stakeholders equates the transformative value of equipping surgeons with 3D MRIs reconstructed from flat 2D imaging slices for planning and simulating an upcoming operation.

What comes to mind is a sophisticated new medium for healthcare, a “ground truth” reconstructed with AI. Healthcare’s “ground truth” is the singular path to addressing disease burden.

 

Dr. Aswin Chandrakantan, MD, is Chief Medical Officer and Head of Product at Komodo Health, where he leads the Product and Clinical Innovations teams.