Personalized Medicine—Humanity’s Ultimate Big Data Challenge Figure 2: The Virtuous Personal Healthcare Cycle


Oracle EHA Data Warehouse


GATAAATCTGGTCTTATTTCC 120


130 -20000 Omics Genotype


Biomarker Identification


B Id


Phenotype pe


Our Product Candidate Set vs. 9 Known Clinical Risk Markers


Analytics at the Point of Care


Enterprise Healthcare Analycs Enter rise Healthcare Analrprilh ise Healthcare Analycs


1.0 0.8 0.6 0.4 0.2 0.0


0.0 -40000 -30000 -20000 -10000 -40000 -3000


Cohort


10000 Identification 0 -10000


Our Marker


Conventional Marker


BGM Marker


Sensitivity 87 % 0.89


Specificity 87 % AUC


Numbers of 20 analytes


0.2 0.4 0.6 1 -Specificity 0.8 1.0


Translaonal Research Center New Evidence-based Care Guideline deline Hospitalize, assess fluid status PCWP ≤ 18 mm HG No PCWP < 15 mm HG


PCWP 15-18 mm HG CI < 2.2 L/min/m2 SBP > 90 mm HG?


Yes No


Administer fluids until PCWP 15-18 mm HG CI < 2.2 L/min/m2


Yes Adjust oral drugs Maintain CI


Monitor for arrhytmia Begin titration of oral drugs


Dopamine or milrinone


CI < 2.2 L/min/m2 Yes No


Consider adjunctive metolazone renal-dose dopamine, nitroprusside, nesiritide, or nitroglicerin


Monitor for arrhytmia SBP > 90 mm HG PCWP 15-18 mm Hg


No No


Titrate dopamine to SBP > 90 mm HG CI < 2.2 L/min/m2


? Yes


PCWP >18 mm HG SBP > 90 mm HG?


Ye Yes


Intermittent infusion of loop diuretic ± metolazone for PCWP 15-18 mm Hg plus dobutamine or milrinone


CI < 2.2 L/min/m2 Yes ? No


Consider combination inotropic agents Add nitroprusside, nesiritide, or nitroglicerin, renal-dose dopamine or metalozone (if not alresdy done)


ttent infusion of uretic ±


1.0 0.8 0.6 0.4 0.2 0.0


1.0 0.8 0.6 0.4


0


Hypothesis Testing


H Te


Segregation by group


1000 Group 1


The virtuous personal healthcare cycle (see Figure 2) is in motion already, but just barely. At the tip of the healthcare ‘spear’, more often than not, we do not really know what works. Many of the interventions we render today have not been studied rigorously. While there is a growing body of evidence-based clinical guidelines (www.guidelines.gov), most are relatively blunt instruments, with little tailoring based on omics or other personalizing factors. Moreover, their evidence basis is still relatively crude. Even for a common condition such as chronic obstructive pulmonary disorder, the support for the recommended treatment is mostly characterized as “moderate quality evidence.”35


A meta-analysis


of published clinical guidelines found that 75 % of those surveyed needed (often major) updating and half of the guidelines were outdated after 5.8 years.36


Medicine in 2009, 13 % were subsequently reversed.37


Of 124 articles published in the New England Journal of As the virtuous


personalized healthcare cycle gains momentum, these guidelines will be continually refined, hopefully at an ever-accelerating pace.


Medicine, like Zen, is in a constant state of becoming, even when we think we are ‘there.’ Capturing this ever-evolving evidence basis is yet another dimension to the big data problem.


iHEALTH CONNECTIONS Gathering the Good Big Data


The big data challenge begins with integrating a patient’s high-volume omics data with phenotypic and environmental data from his or her electronic medical record (EMR) and other data sources. Some data in healthcare are highly structured, i.e., we know exactly what they contain and mean (e.g., date of birth). However, most are loosely structured or unstructured data, the content of which has to be interpreted. For example, dictated notes, such as the history of present illness, are typically unstructured, as are most social media data. The big data challenge in personalized medicine requires these variably encoded data to come together in a physical or virtual data ‘warehouse.’ This big data warehouse will not only serve as a repository but also as the arbiter of ‘good’ data.38,39


Healthcare is rife with ‘bad data’, i.e., errors that can be subtle or not so subtle, omissions, mis-categorizations, etc. These data defects often are not a problem for clinicians who can easily extrapolate, interpolate, or look past them. They are, however, a problem for automated analysis. For example, one EMR system regularly submits patient body weights in excess of 2,000 pounds (nurses enter their premature infants’


93


2000 Time in days Group 2


3000


4000


Proportions (...) oportions (...)


Sensitivity


Page 1  |  Page 2  |  Page 3  |  Page 4  |  Page 5  |  Page 6  |  Page 7  |  Page 8  |  Page 9  |  Page 10  |  Page 11  |  Page 12  |  Page 13  |  Page 14  |  Page 15  |  Page 16  |  Page 17  |  Page 18  |  Page 19  |  Page 20  |  Page 21  |  Page 22  |  Page 23  |  Page 24  |  Page 25  |  Page 26  |  Page 27  |  Page 28  |  Page 29  |  Page 30  |  Page 31  |  Page 32  |  Page 33  |  Page 34  |  Page 35  |  Page 36  |  Page 37  |  Page 38  |  Page 39  |  Page 40  |  Page 41  |  Page 42  |  Page 43  |  Page 44  |  Page 45  |  Page 46  |  Page 47  |  Page 48  |  Page 49  |  Page 50  |  Page 51  |  Page 52  |  Page 53  |  Page 54  |  Page 55  |  Page 56  |  Page 57  |  Page 58  |  Page 59  |  Page 60  |  Page 61  |  Page 62  |  Page 63  |  Page 64  |  Page 65  |  Page 66  |  Page 67  |  Page 68  |  Page 69  |  Page 70  |  Page 71  |  Page 72  |  Page 73  |  Page 74  |  Page 75  |  Page 76  |  Page 77  |  Page 78  |  Page 79  |  Page 80  |  Page 81  |  Page 82  |  Page 83  |  Page 84